entry_point
stringlengths 1
65
| original_triton_code
stringlengths 4.5k
619k
| python_code
stringlengths 208
60.9k
| triton_code
stringlengths 1.15k
275k
| repo_name
stringlengths 7
115
| module_name
stringlengths 1
65
| synthetic
bool 1
class | uuid
int64 0
18.5k
| licenses
sequencelengths 1
6
| stars
int64 0
19.8k
| sha
stringlengths 40
40
| repo_link
stringlengths 72
180
| pytorch_code
stringlengths 200
4.05k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
ANN | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/xk/cxkoqb3kxz2yd7scflge247h2s4tm7sn45x5eovbo5agxcenjofr.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.sigmoid]
# Source node to ATen node mapping:
# x_1 => sigmoid
# Graph fragment:
# %sigmoid : [num_users=2] = call_function[target=torch.ops.aten.sigmoid.default](args = (%view_1,), kwargs = {})
triton_poi_fused_sigmoid_0 = async_compile.triton('triton_poi_fused_sigmoid_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sigmoid_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sigmoid_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1280
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 20
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.sigmoid(tmp2)
tl.store(in_out_ptr0 + (x2), tmp3, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/7z/c7zsuucunqdovb2xa6tywxjxwmolzjzdk72ratro7fi3qvgyqb7c.py
# Topologically Sorted Source Nodes: [x_7], Original ATen: [aten.sigmoid]
# Source node to ATen node mapping:
# x_7 => sigmoid_3
# Graph fragment:
# %sigmoid_3 : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%view_7,), kwargs = {})
triton_poi_fused_sigmoid_1 = async_compile.triton('triton_poi_fused_sigmoid_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sigmoid_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sigmoid_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (20, 4), (4, 1))
assert_size_stride(primals_2, (20, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (20, 20), (20, 1))
assert_size_stride(primals_5, (20, ), (1, ))
assert_size_stride(primals_6, (20, 20), (20, 1))
assert_size_stride(primals_7, (20, ), (1, ))
assert_size_stride(primals_8, (1, 20), (20, 1))
assert_size_stride(primals_9, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 20), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 20), (320, 80, 20, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.sigmoid]
stream0 = get_raw_stream(0)
triton_poi_fused_sigmoid_0.run(buf1, primals_2, 1280, grid=grid(1280), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf1, (64, 20), (20, 1), 0), reinterpret_tensor(primals_4, (20, 20), (1, 20), 0), out=buf2)
buf3 = reinterpret_tensor(buf2, (4, 4, 4, 20), (320, 80, 20, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.sigmoid]
triton_poi_fused_sigmoid_0.run(buf3, primals_5, 1280, grid=grid(1280), stream=stream0)
del primals_5
buf4 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf3, (64, 20), (20, 1), 0), reinterpret_tensor(primals_6, (20, 20), (1, 20), 0), out=buf4)
buf5 = reinterpret_tensor(buf4, (4, 4, 4, 20), (320, 80, 20, 1), 0); del buf4 # reuse
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.sigmoid]
triton_poi_fused_sigmoid_0.run(buf5, primals_7, 1280, grid=grid(1280), stream=stream0)
del primals_7
buf6 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf5, (64, 20), (20, 1), 0), reinterpret_tensor(primals_8, (20, 1), (1, 20), 0), out=buf6)
buf7 = reinterpret_tensor(buf6, (4, 4, 4, 1), (16, 4, 1, 1), 0); del buf6 # reuse
# Topologically Sorted Source Nodes: [x_7], Original ATen: [aten.sigmoid]
triton_poi_fused_sigmoid_1.run(buf7, primals_9, 64, grid=grid(64), stream=stream0)
del primals_9
return (buf7, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, buf3, buf5, buf7, primals_8, primals_6, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((20, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((20, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((20, 20), (20, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((20, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((20, 20), (20, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((20, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((1, 20), (20, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from _paritybench_helpers import _mock_config
import torch
from torch import nn
class ANN(nn.Module):
def __init__(self, args, name):
super(ANN, self).__init__()
self.name = name
self.len = 0
self.loss = 0
self.fc1 = nn.Linear(args.input_dim, 20)
self.relu = nn.ReLU()
self.sigmoid = nn.Sigmoid()
self.dropout = nn.Dropout()
self.fc2 = nn.Linear(20, 20)
self.fc3 = nn.Linear(20, 20)
self.fc4 = nn.Linear(20, 1)
def forward(self, data):
x = self.fc1(data)
x = self.sigmoid(x)
x = self.fc2(x)
x = self.sigmoid(x)
x = self.fc3(x)
x = self.sigmoid(x)
x = self.fc4(x)
x = self.sigmoid(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'args': _mock_config(input_dim=4), 'name': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_sigmoid_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 1280
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 20
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.sigmoid(tmp2)
tl.store(in_out_ptr0 + x2, tmp3, xmask)
@triton.jit
def triton_poi_fused_sigmoid_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + x0, tmp4, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (20, 4), (4, 1))
assert_size_stride(primals_2, (20,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (20, 20), (20, 1))
assert_size_stride(primals_5, (20,), (1,))
assert_size_stride(primals_6, (20, 20), (20, 1))
assert_size_stride(primals_7, (20,), (1,))
assert_size_stride(primals_8, (1, 20), (20, 1))
assert_size_stride(primals_9, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 20), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 20), (320, 80, 20, 1), 0)
del buf0
get_raw_stream(0)
triton_poi_fused_sigmoid_0[grid(1280)](buf1, primals_2, 1280,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf1, (64, 20), (20, 1), 0),
reinterpret_tensor(primals_4, (20, 20), (1, 20), 0), out=buf2)
buf3 = reinterpret_tensor(buf2, (4, 4, 4, 20), (320, 80, 20, 1), 0)
del buf2
triton_poi_fused_sigmoid_0[grid(1280)](buf3, primals_5, 1280,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_5
buf4 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf3, (64, 20), (20, 1), 0),
reinterpret_tensor(primals_6, (20, 20), (1, 20), 0), out=buf4)
buf5 = reinterpret_tensor(buf4, (4, 4, 4, 20), (320, 80, 20, 1), 0)
del buf4
triton_poi_fused_sigmoid_0[grid(1280)](buf5, primals_7, 1280,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_7
buf6 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf5, (64, 20), (20, 1), 0),
reinterpret_tensor(primals_8, (20, 1), (1, 20), 0), out=buf6)
buf7 = reinterpret_tensor(buf6, (4, 4, 4, 1), (16, 4, 1, 1), 0)
del buf6
triton_poi_fused_sigmoid_1[grid(64)](buf7, primals_9, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_9
return buf7, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf1, buf3, buf5, buf7, primals_8, primals_6, primals_4
class ANNNew(nn.Module):
def __init__(self, args, name):
super(ANNNew, self).__init__()
self.name = name
self.len = 0
self.loss = 0
self.fc1 = nn.Linear(args.input_dim, 20)
self.relu = nn.ReLU()
self.sigmoid = nn.Sigmoid()
self.dropout = nn.Dropout()
self.fc2 = nn.Linear(20, 20)
self.fc3 = nn.Linear(20, 20)
self.fc4 = nn.Linear(20, 1)
def forward(self, input_0):
primals_1 = self.fc1.weight
primals_2 = self.fc1.bias
primals_4 = self.fc2.weight
primals_5 = self.fc2.bias
primals_6 = self.fc3.weight
primals_7 = self.fc3.bias
primals_8 = self.fc4.weight
primals_9 = self.fc4.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
| luoyang97/FedProx-PyTorch | ANN | false | 7,133 | [
"MIT"
] | 1 | b19263e22420251ad8c3a9701951a37b5c0a3569 | https://github.com/luoyang97/FedProx-PyTorch/tree/b19263e22420251ad8c3a9701951a37b5c0a3569 | from _paritybench_helpers import _mock_config
import torch
from torch import nn
class Model(nn.Module):
def __init__(self, args, name):
super().__init__()
self.name = name
self.len = 0
self.loss = 0
self.fc1 = nn.Linear(args.input_dim, 20)
self.relu = nn.ReLU()
self.sigmoid = nn.Sigmoid()
self.dropout = nn.Dropout()
self.fc2 = nn.Linear(20, 20)
self.fc3 = nn.Linear(20, 20)
self.fc4 = nn.Linear(20, 1)
def forward(self, data):
x = self.fc1(data)
x = self.sigmoid(x)
x = self.fc2(x)
x = self.sigmoid(x)
x = self.fc3(x)
x = self.sigmoid(x)
x = self.fc4(x)
x = self.sigmoid(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
Predict_Network1_combine | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/cy/ccy2s3bgoyu6vw3wuf772mifqyy4tokqrm5khaji772ruw2shslv.py
# Topologically Sorted Source Nodes: [mean, std, sub, add, output, output_1, h], Original ATen: [aten.mean, aten.std, aten.sub, aten.add, aten.div, aten.relu]
# Source node to ATen node mapping:
# add => add
# h => relu
# mean => mean
# output => div
# output_1 => add_1
# std => sqrt, var
# sub => sub
# Graph fragment:
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%view_1, [-1], True), kwargs = {})
# %var : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%view_1, [-1]), kwargs = {correction: 1.0, keepdim: True})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%var,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view_1, %mean), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sqrt, 1e-06), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %add), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%div, %primals_4), kwargs = {})
# %relu : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%add_1,), kwargs = {})
triton_poi_fused_add_div_mean_relu_std_sub_0 = async_compile.triton('triton_poi_fused_add_div_mean_relu_std_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mean_relu_std_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mean_relu_std_sub_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp28 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = 4.0
tmp9 = tmp7 / tmp8
tmp10 = tmp0 - tmp9
tmp11 = tmp1 - tmp9
tmp12 = tmp11 * tmp11
tmp13 = tmp2 - tmp9
tmp14 = tmp13 * tmp13
tmp15 = tmp12 + tmp14
tmp16 = tmp4 - tmp9
tmp17 = tmp16 * tmp16
tmp18 = tmp15 + tmp17
tmp19 = tmp6 - tmp9
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = 3.0
tmp23 = tmp21 / tmp22
tmp24 = libdevice.sqrt(tmp23)
tmp25 = 1e-06
tmp26 = tmp24 + tmp25
tmp27 = tmp10 / tmp26
tmp29 = tmp27 + tmp28
tmp30 = tl.full([1], 0, tl.int32)
tmp31 = triton_helpers.maximum(tmp30, tmp29)
tl.store(out_ptr0 + (x0 + (8*x1)), tmp31, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/y7/cy7dwphppgpb4tjogyhecgnwlzbv37vngcsd2jy2ongmfasz4h2u.py
# Topologically Sorted Source Nodes: [h_1], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# h_1 => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%relu, %primals_5], -1), kwargs = {})
triton_poi_fused_cat_1 = async_compile.triton('triton_poi_fused_cat_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tl.store(out_ptr0 + (x0 + (8*x1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/gm/cgmflgdlpeeb52xctoa47uvw47ycyf7ahlj5wdscxdatpbwcboco.py
# Topologically Sorted Source Nodes: [h_2], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# h_2 => relu_1
# Graph fragment:
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_3,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_1, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_2 = async_compile.triton('triton_poi_fused_relu_threshold_backward_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_2(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, ), (1, ))
assert_size_stride(primals_5, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_6, (4, 8), (8, 1))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, 4), (4, 1))
assert_size_stride(primals_9, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf3 = empty_strided_cuda((4, 4, 4, 8), (128, 32, 8, 1), torch.float32)
buf1 = reinterpret_tensor(buf3, (4, 4, 4, 4), (128, 32, 8, 1), 0) # alias
# Topologically Sorted Source Nodes: [mean, std, sub, add, output, output_1, h], Original ATen: [aten.mean, aten.std, aten.sub, aten.add, aten.div, aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_mean_relu_std_sub_0.run(buf0, primals_4, buf1, 256, grid=grid(256), stream=stream0)
buf2 = reinterpret_tensor(buf3, (4, 4, 4, 4), (128, 32, 8, 1), 4) # alias
# Topologically Sorted Source Nodes: [h_1], Original ATen: [aten.cat]
triton_poi_fused_cat_1.run(primals_5, buf2, 256, grid=grid(256), stream=stream0)
del primals_5
buf4 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf3, (64, 8), (8, 1), 0), reinterpret_tensor(primals_6, (8, 4), (1, 8), 0), out=buf4)
buf5 = reinterpret_tensor(buf4, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf4 # reuse
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [h_2], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_2.run(buf5, primals_7, buf7, 256, grid=grid(256), stream=stream0)
del primals_7
buf6 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_9, reinterpret_tensor(buf5, (64, 4), (4, 1), 0), reinterpret_tensor(primals_8, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf6)
del primals_9
return (reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0), primals_4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf0, reinterpret_tensor(buf3, (64, 8), (8, 1), 0), reinterpret_tensor(buf5, (64, 4), (4, 1), 0), primals_8, buf7, primals_6, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 8), (8, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
def weights_init_(m):
if isinstance(m, nn.Linear):
torch.nn.init.xavier_uniform_(m.weight, gain=1)
torch.nn.init.constant_(m.bias, 0)
class LayerNorm(nn.Module):
"""
Simple 1D LayerNorm.
"""
def __init__(self, features, center=True, scale=False, eps=1e-06):
super().__init__()
self.center = center
self.scale = scale
self.eps = eps
if self.scale:
self.scale_param = nn.Parameter(torch.ones(features))
else:
self.scale_param = None
if self.center:
self.center_param = nn.Parameter(torch.zeros(features))
else:
self.center_param = None
def forward(self, x):
mean = x.mean(-1, keepdim=True)
std = x.std(-1, keepdim=True)
output = (x - mean) / (std + self.eps)
if self.scale:
output = output * self.scale_param
if self.center:
output = output + self.center_param
return output
class Predict_Network1_combine(nn.Module):
def __init__(self, num_inputs, hidden_dim, num_outputs, n_agents,
layer_norm=True, lr=0.001):
super(Predict_Network1_combine, self).__init__()
self.linear1 = nn.Linear(num_inputs, hidden_dim)
self.linear2 = nn.Linear(hidden_dim + n_agents, hidden_dim)
self.last_fc = nn.Linear(hidden_dim, num_outputs)
self.layer_norm = layer_norm
if layer_norm:
self.ln1 = LayerNorm(hidden_dim)
self.apply(weights_init_)
self.lr = lr
self.optimizer = optim.Adam(self.parameters(), lr=self.lr)
def forward(self, input, add_id):
if self.layer_norm:
h = F.relu(self.ln1(self.linear1(input)))
else:
h = F.relu(self.linear1(input))
h = torch.cat([h, add_id], dim=-1)
h = F.relu(self.linear2(h))
x = self.last_fc(h)
return x
def get_log_pi(self, own_variable, other_variable, add_id):
predict_variable = self.forward(own_variable, add_id)
log_prob = -1 * F.mse_loss(predict_variable, other_variable,
reduction='none')
log_prob = torch.sum(log_prob, -1, keepdim=True)
return log_prob
def update(self, own_variable, other_variable, add_id, mask):
predict_variable = self.forward(own_variable, add_id)
loss = F.mse_loss(predict_variable, other_variable, reduction='none')
loss = loss.sum(dim=-1, keepdim=True)
loss = (loss * mask).sum() / mask.sum()
self.optimizer.zero_grad()
loss.backward()
torch.nn.utils.clip_grad_norm_(self.parameters(), 1.0)
self.optimizer.step()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'num_inputs': 4, 'hidden_dim': 4, 'num_outputs': 4,
'n_agents': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_div_mean_relu_std_sub_0(in_ptr0, in_ptr1, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp28 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = 4.0
tmp9 = tmp7 / tmp8
tmp10 = tmp0 - tmp9
tmp11 = tmp1 - tmp9
tmp12 = tmp11 * tmp11
tmp13 = tmp2 - tmp9
tmp14 = tmp13 * tmp13
tmp15 = tmp12 + tmp14
tmp16 = tmp4 - tmp9
tmp17 = tmp16 * tmp16
tmp18 = tmp15 + tmp17
tmp19 = tmp6 - tmp9
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = 3.0
tmp23 = tmp21 / tmp22
tmp24 = libdevice.sqrt(tmp23)
tmp25 = 1e-06
tmp26 = tmp24 + tmp25
tmp27 = tmp10 / tmp26
tmp29 = tmp27 + tmp28
tmp30 = tl.full([1], 0, tl.int32)
tmp31 = triton_helpers.maximum(tmp30, tmp29)
tl.store(out_ptr0 + (x0 + 8 * x1), tmp31, xmask)
@triton.jit
def triton_poi_fused_cat_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tl.store(out_ptr0 + (x0 + 8 * x1), tmp0, xmask)
@triton.jit
def triton_poi_fused_relu_threshold_backward_2(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4,), (1,))
assert_size_stride(primals_5, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_6, (4, 8), (8, 1))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4, 4), (4, 1))
assert_size_stride(primals_9, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf3 = empty_strided_cuda((4, 4, 4, 8), (128, 32, 8, 1), torch.float32)
buf1 = reinterpret_tensor(buf3, (4, 4, 4, 4), (128, 32, 8, 1), 0)
get_raw_stream(0)
triton_poi_fused_add_div_mean_relu_std_sub_0[grid(256)](buf0,
primals_4, buf1, 256, XBLOCK=128, num_warps=4, num_stages=1)
buf2 = reinterpret_tensor(buf3, (4, 4, 4, 4), (128, 32, 8, 1), 4)
triton_poi_fused_cat_1[grid(256)](primals_5, buf2, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del primals_5
buf4 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf3, (64, 8), (8, 1), 0),
reinterpret_tensor(primals_6, (8, 4), (1, 8), 0), out=buf4)
buf5 = reinterpret_tensor(buf4, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf4
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused_relu_threshold_backward_2[grid(256)](buf5,
primals_7, buf7, 256, XBLOCK=128, num_warps=4, num_stages=1)
del primals_7
buf6 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_9, reinterpret_tensor(buf5, (64, 4), (
4, 1), 0), reinterpret_tensor(primals_8, (4, 4), (1, 4), 0),
alpha=1, beta=1, out=buf6)
del primals_9
return reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0
), primals_4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf0, reinterpret_tensor(buf3, (64, 8), (8, 1), 0
), reinterpret_tensor(buf5, (64, 4), (4, 1), 0
), primals_8, buf7, primals_6
def weights_init_(m):
if isinstance(m, nn.Linear):
torch.nn.init.xavier_uniform_(m.weight, gain=1)
torch.nn.init.constant_(m.bias, 0)
class LayerNorm(nn.Module):
"""
Simple 1D LayerNorm.
"""
def __init__(self, features, center=True, scale=False, eps=1e-06):
super().__init__()
self.center = center
self.scale = scale
self.eps = eps
if self.scale:
self.scale_param = nn.Parameter(torch.ones(features))
else:
self.scale_param = None
if self.center:
self.center_param = nn.Parameter(torch.zeros(features))
else:
self.center_param = None
def forward(self, x):
mean = x.mean(-1, keepdim=True)
std = x.std(-1, keepdim=True)
output = (x - mean) / (std + self.eps)
if self.scale:
output = output * self.scale_param
if self.center:
output = output + self.center_param
return output
class Predict_Network1_combineNew(nn.Module):
def __init__(self, num_inputs, hidden_dim, num_outputs, n_agents,
layer_norm=True, lr=0.001):
super(Predict_Network1_combineNew, self).__init__()
self.linear1 = nn.Linear(num_inputs, hidden_dim)
self.linear2 = nn.Linear(hidden_dim + n_agents, hidden_dim)
self.last_fc = nn.Linear(hidden_dim, num_outputs)
self.layer_norm = layer_norm
if layer_norm:
self.ln1 = LayerNorm(hidden_dim)
self.apply(weights_init_)
self.lr = lr
self.optimizer = optim.Adam(self.parameters(), lr=self.lr)
def get_log_pi(self, own_variable, other_variable, add_id):
predict_variable = self.forward(own_variable, add_id)
log_prob = -1 * F.mse_loss(predict_variable, other_variable,
reduction='none')
log_prob = torch.sum(log_prob, -1, keepdim=True)
return log_prob
def update(self, own_variable, other_variable, add_id, mask):
predict_variable = self.forward(own_variable, add_id)
loss = F.mse_loss(predict_variable, other_variable, reduction='none')
loss = loss.sum(dim=-1, keepdim=True)
loss = (loss * mask).sum() / mask.sum()
self.optimizer.zero_grad()
loss.backward()
torch.nn.utils.clip_grad_norm_(self.parameters(), 1.0)
self.optimizer.step()
def forward(self, input_0, input_1):
primals_1 = self.linear1.weight
primals_2 = self.linear1.bias
primals_6 = self.linear2.weight
primals_4 = self.linear2.bias
primals_8 = self.last_fc.weight
primals_7 = self.last_fc.bias
primals_9 = self.ln1.center_param
primals_3 = input_0
primals_5 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
| ltzheng/CDS | Predict_Network1_combine | false | 7,134 | [
"Apache-2.0"
] | 1 | 397282147498647a9f26577adfa451e8478de76d | https://github.com/ltzheng/CDS/tree/397282147498647a9f26577adfa451e8478de76d | import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
def weights_init_(m):
if isinstance(m, nn.Linear):
torch.nn.init.xavier_uniform_(m.weight, gain=1)
torch.nn.init.constant_(m.bias, 0)
class LayerNorm(nn.Module):
"""
Simple 1D LayerNorm.
"""
def __init__(self, features, center=True, scale=False, eps=1e-06):
super().__init__()
self.center = center
self.scale = scale
self.eps = eps
if self.scale:
self.scale_param = nn.Parameter(torch.ones(features))
else:
self.scale_param = None
if self.center:
self.center_param = nn.Parameter(torch.zeros(features))
else:
self.center_param = None
def forward(self, x):
mean = x.mean(-1, keepdim=True)
std = x.std(-1, keepdim=True)
output = (x - mean) / (std + self.eps)
if self.scale:
output = output * self.scale_param
if self.center:
output = output + self.center_param
return output
class Model(nn.Module):
def __init__(self, num_inputs, hidden_dim, num_outputs, n_agents,
layer_norm=True, lr=0.001):
super().__init__()
self.linear1 = nn.Linear(num_inputs, hidden_dim)
self.linear2 = nn.Linear(hidden_dim + n_agents, hidden_dim)
self.last_fc = nn.Linear(hidden_dim, num_outputs)
self.layer_norm = layer_norm
if layer_norm:
self.ln1 = LayerNorm(hidden_dim)
self.apply(weights_init_)
self.lr = lr
self.optimizer = optim.Adam(self.parameters(), lr=self.lr)
def forward(self, input, add_id):
if self.layer_norm:
h = F.relu(self.ln1(self.linear1(input)))
else:
h = F.relu(self.linear1(input))
h = torch.cat([h, add_id], dim=-1)
h = F.relu(self.linear2(h))
x = self.last_fc(h)
return x
def get_log_pi(self, own_variable, other_variable, add_id):
predict_variable = self.forward(own_variable, add_id)
log_prob = -1 * F.mse_loss(predict_variable, other_variable,
reduction='none')
log_prob = torch.sum(log_prob, -1, keepdim=True)
return log_prob
def update(self, own_variable, other_variable, add_id, mask):
predict_variable = self.forward(own_variable, add_id)
loss = F.mse_loss(predict_variable, other_variable, reduction='none')
loss = loss.sum(dim=-1, keepdim=True)
loss = (loss * mask).sum() / mask.sum()
self.optimizer.zero_grad()
loss.backward()
torch.nn.utils.clip_grad_norm_(self.parameters(), 1.0)
self.optimizer.step()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'num_inputs': 4, 'hidden_dim': 4, 'num_outputs': 4,
'n_agents': 4}]
|
GeM | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/w7/cw76edgx2byvy63fwff7anzlambkufz45h3okmkxmcgppfib77jc.py
# Topologically Sorted Source Nodes: [clamp, pow_1], Original ATen: [aten.clamp, aten.pow]
# Source node to ATen node mapping:
# clamp => clamp_min
# pow_1 => pow_1
# Graph fragment:
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%primals_1, 1e-06), kwargs = {})
# %pow_1 : [num_users=2] = call_function[target=torch.ops.aten.pow.Tensor_Tensor](args = (%clamp_min, %primals_2), kwargs = {})
triton_poi_fused_clamp_pow_0 = async_compile.triton('triton_poi_fused_clamp_pow_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clamp_pow_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clamp_pow_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp3 = tl.load(in_ptr1 + (0))
tmp4 = tl.broadcast_to(tmp3, [XBLOCK])
tmp1 = 1e-06
tmp2 = triton_helpers.maximum(tmp0, tmp1)
tmp5 = libdevice.pow(tmp2, tmp4)
tl.store(out_ptr0 + (x0), tmp5, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/bw/cbwgbppqaap3oywo4gn6wujtg6clpxlchs4m6kjxievrbuoqr6wh.py
# Topologically Sorted Source Nodes: [avg_pool2d, truediv, pow_2], Original ATen: [aten.avg_pool2d, aten.reciprocal, aten.mul, aten.pow]
# Source node to ATen node mapping:
# avg_pool2d => avg_pool2d
# pow_2 => pow_2
# truediv => mul, reciprocal
# Graph fragment:
# %avg_pool2d : [num_users=2] = call_function[target=torch.ops.aten.avg_pool2d.default](args = (%pow_1, [4, 4]), kwargs = {})
# %reciprocal : [num_users=1] = call_function[target=torch.ops.aten.reciprocal.default](args = (%primals_2,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%reciprocal, 1.0), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Tensor](args = (%avg_pool2d, %mul), kwargs = {})
triton_poi_fused_avg_pool2d_mul_pow_reciprocal_1 = async_compile.triton('triton_poi_fused_avg_pool2d_mul_pow_reciprocal_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_avg_pool2d_mul_pow_reciprocal_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 17, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_avg_pool2d_mul_pow_reciprocal_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (16*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (16*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (16*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (16*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (4 + (16*x0)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (5 + (16*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (6 + (16*x0)), xmask, eviction_policy='evict_last')
tmp13 = tl.load(in_ptr0 + (7 + (16*x0)), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr0 + (8 + (16*x0)), xmask, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr0 + (9 + (16*x0)), xmask, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr0 + (10 + (16*x0)), xmask, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (11 + (16*x0)), xmask, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr0 + (12 + (16*x0)), xmask, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr0 + (13 + (16*x0)), xmask, eviction_policy='evict_last')
tmp27 = tl.load(in_ptr0 + (14 + (16*x0)), xmask, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr0 + (15 + (16*x0)), xmask, eviction_policy='evict_last')
tmp33 = tl.load(in_ptr1 + (0))
tmp34 = tl.broadcast_to(tmp33, [XBLOCK])
tmp2 = tmp1 + tmp0
tmp4 = tmp3 + tmp2
tmp6 = tmp5 + tmp4
tmp8 = tmp7 + tmp6
tmp10 = tmp9 + tmp8
tmp12 = tmp11 + tmp10
tmp14 = tmp13 + tmp12
tmp16 = tmp15 + tmp14
tmp18 = tmp17 + tmp16
tmp20 = tmp19 + tmp18
tmp22 = tmp21 + tmp20
tmp24 = tmp23 + tmp22
tmp26 = tmp25 + tmp24
tmp28 = tmp27 + tmp26
tmp30 = tmp29 + tmp28
tmp31 = 0.0625
tmp32 = tmp30 * tmp31
tmp35 = tl.full([1], 1, tl.int32)
tmp36 = tmp35 / tmp34
tmp37 = 1.0
tmp38 = tmp36 * tmp37
tmp39 = libdevice.pow(tmp32, tmp38)
tl.store(out_ptr0 + (x0), tmp32, xmask)
tl.store(out_ptr1 + (x0), tmp39, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [clamp, pow_1], Original ATen: [aten.clamp, aten.pow]
stream0 = get_raw_stream(0)
triton_poi_fused_clamp_pow_0.run(primals_1, primals_2, buf0, 256, grid=grid(256), stream=stream0)
buf1 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
buf2 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [avg_pool2d, truediv, pow_2], Original ATen: [aten.avg_pool2d, aten.reciprocal, aten.mul, aten.pow]
triton_poi_fused_avg_pool2d_mul_pow_reciprocal_1.run(buf0, primals_2, buf1, buf2, 16, grid=grid(16), stream=stream0)
return (buf2, primals_1, primals_2, buf0, buf1, buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
import torch.nn.functional as F
from torch.nn.parameter import Parameter
class GeM(nn.Module):
def __init__(self, p=3, eps=1e-06):
super(GeM, self).__init__()
self.p = Parameter(torch.ones(1) * p)
self.eps = eps
def forward(self, x):
return F.avg_pool2d(x.clamp(min=self.eps).pow(self.p), (x.size(-2),
x.size(-1))).pow(1.0 / self.p)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
from torch.nn.parameter import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_clamp_pow_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp3 = tl.load(in_ptr1 + 0)
tmp4 = tl.broadcast_to(tmp3, [XBLOCK])
tmp1 = 1e-06
tmp2 = triton_helpers.maximum(tmp0, tmp1)
tmp5 = libdevice.pow(tmp2, tmp4)
tl.store(out_ptr0 + x0, tmp5, xmask)
@triton.jit
def triton_poi_fused_avg_pool2d_mul_pow_reciprocal_1(in_ptr0, in_ptr1,
out_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 16 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp3 = tl.load(in_ptr0 + (2 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp5 = tl.load(in_ptr0 + (3 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp7 = tl.load(in_ptr0 + (4 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp9 = tl.load(in_ptr0 + (5 + 16 * x0), xmask, eviction_policy='evict_last'
)
tmp11 = tl.load(in_ptr0 + (6 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp13 = tl.load(in_ptr0 + (7 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp15 = tl.load(in_ptr0 + (8 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp17 = tl.load(in_ptr0 + (9 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp19 = tl.load(in_ptr0 + (10 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp21 = tl.load(in_ptr0 + (11 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp23 = tl.load(in_ptr0 + (12 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp25 = tl.load(in_ptr0 + (13 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp27 = tl.load(in_ptr0 + (14 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp29 = tl.load(in_ptr0 + (15 + 16 * x0), xmask, eviction_policy=
'evict_last')
tmp33 = tl.load(in_ptr1 + 0)
tmp34 = tl.broadcast_to(tmp33, [XBLOCK])
tmp2 = tmp1 + tmp0
tmp4 = tmp3 + tmp2
tmp6 = tmp5 + tmp4
tmp8 = tmp7 + tmp6
tmp10 = tmp9 + tmp8
tmp12 = tmp11 + tmp10
tmp14 = tmp13 + tmp12
tmp16 = tmp15 + tmp14
tmp18 = tmp17 + tmp16
tmp20 = tmp19 + tmp18
tmp22 = tmp21 + tmp20
tmp24 = tmp23 + tmp22
tmp26 = tmp25 + tmp24
tmp28 = tmp27 + tmp26
tmp30 = tmp29 + tmp28
tmp31 = 0.0625
tmp32 = tmp30 * tmp31
tmp35 = tl.full([1], 1, tl.int32)
tmp36 = tmp35 / tmp34
tmp37 = 1.0
tmp38 = tmp36 * tmp37
tmp39 = libdevice.pow(tmp32, tmp38)
tl.store(out_ptr0 + x0, tmp32, xmask)
tl.store(out_ptr1 + x0, tmp39, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clamp_pow_0[grid(256)](primals_1, primals_2, buf0,
256, XBLOCK=256, num_warps=4, num_stages=1)
buf1 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
buf2 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
triton_poi_fused_avg_pool2d_mul_pow_reciprocal_1[grid(16)](buf0,
primals_2, buf1, buf2, 16, XBLOCK=16, num_warps=1, num_stages=1)
return buf2, primals_1, primals_2, buf0, buf1, buf2
class GeMNew(nn.Module):
def __init__(self, p=3, eps=1e-06):
super(GeMNew, self).__init__()
self.p = Parameter(torch.ones(1) * p)
self.eps = eps
def forward(self, input_0):
primals_2 = self.p
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0]
| lulor/project_vg | GeM | false | 7,135 | [
"MIT"
] | 1 | 27b0c3b3038c5a666dde516a0a265ae8ddf2059f | https://github.com/lulor/project_vg/tree/27b0c3b3038c5a666dde516a0a265ae8ddf2059f | import torch
from torch import nn
import torch.nn.functional as F
from torch.nn.parameter import Parameter
class Model(nn.Module):
def __init__(self, p=3, eps=1e-06):
super().__init__()
self.p = Parameter(torch.ones(1) * p)
self.eps = eps
def forward(self, x):
return F.avg_pool2d(x.clamp(min=self.eps).pow(self.p), (x.size(-2),
x.size(-1))).pow(1.0 / self.p)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
SEModule | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/l3/cl35tzbhrd24dhunkbb6gjs54aklpyr46oikqhoylcgmkcmhujil.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.mean]
# Source node to ATen node mapping:
# x => mean
# Graph fragment:
# %mean : [num_users=2] = call_function[target=torch.ops.aten.mean.dim](args = (%primals_1, [-1, -2], True), kwargs = {})
triton_per_fused_mean_0 = async_compile.triton('triton_per_fused_mean_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[16, 16],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mean_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mean_0(in_out_ptr0, in_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 16
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (16*x0)), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp5 = 16.0
tmp6 = tmp4 / tmp5
tl.debug_barrier()
tl.store(in_out_ptr0 + (x0), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ad/cadccuyhl7stcp3nyqfgohiwbiv5ckfzxsye27ithwsill6dvmh4.py
# Topologically Sorted Source Nodes: [x_1, x_2], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# x_1 => convolution
# x_2 => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%mean, %primals_2, %primals_3, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_1 = async_compile.triton('triton_poi_fused_convolution_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.full([1], 0, tl.int32)
tmp5 = triton_helpers.maximum(tmp4, tmp3)
tl.store(in_out_ptr0 + (x0), tmp5, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/hc/chcuw27vv75itefu6xswbdhopf2pq6oaj4bww4whelczhqtkghqr.py
# Topologically Sorted Source Nodes: [x_3, add, relu6, x_4, mul], Original ATen: [aten.convolution, aten.add, aten.hardtanh, aten.div, aten.mul]
# Source node to ATen node mapping:
# add => add
# mul => mul
# relu6 => clamp_max, clamp_min
# x_3 => convolution_1
# x_4 => div
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %add : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%convolution_1, 3.0), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add, 0), kwargs = {})
# %clamp_max : [num_users=1] = call_function[target=torch.ops.aten.clamp_max.default](args = (%clamp_min, 6), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%clamp_max, 6.0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_1, %div), kwargs = {})
triton_poi_fused_add_convolution_div_hardtanh_mul_2 = async_compile.triton('triton_poi_fused_add_convolution_div_hardtanh_mul_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_convolution_div_hardtanh_mul_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_convolution_div_hardtanh_mul_2(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x4 = (xindex // 16)
x1 = (xindex // 16) % 4
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr1 + (x4), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = 3.0
tmp5 = tmp3 + tmp4
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = 6.0
tmp9 = triton_helpers.minimum(tmp7, tmp8)
tmp10 = 0.16666666666666666
tmp11 = tmp9 * tmp10
tmp12 = tmp0 * tmp11
tl.store(out_ptr0 + (x3), tmp12, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/rh/crhdgkbcu2rbqzaxws3536qfrpjhmp6d6zhyebzqejxvceodywdi.py
# Topologically Sorted Source Nodes: [x_3, add], Original ATen: [aten.convolution, aten.add, aten.hardtanh_backward]
# Source node to ATen node mapping:
# add => add
# x_3 => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %add : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%convolution_1, 3.0), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%add, 0), kwargs = {})
# %ge : [num_users=1] = call_function[target=torch.ops.aten.ge.Scalar](args = (%add, 6), kwargs = {})
# %bitwise_or : [num_users=1] = call_function[target=torch.ops.aten.bitwise_or.Tensor](args = (%le, %ge), kwargs = {})
triton_poi_fused_add_convolution_hardtanh_backward_3 = async_compile.triton('triton_poi_fused_add_convolution_hardtanh_backward_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_convolution_hardtanh_backward_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_convolution_hardtanh_backward_3(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 3.0
tmp4 = tmp2 + tmp3
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tmp7 = 6.0
tmp8 = tmp4 >= tmp7
tmp9 = tmp6 | tmp8
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_3, (1, ), (1, ))
assert_size_stride(primals_4, (4, 1, 1, 1), (1, 1, 1, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 1, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_mean_0.run(buf1, primals_1, 16, 16, grid=grid(16), stream=stream0)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_2, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 1, 1), (1, 1, 1, 1))
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [x_1, x_2], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf3, primals_3, 4, grid=grid(4), stream=stream0)
del primals_3
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(buf3, primals_4, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 4, 1, 1), (4, 1, 1, 1))
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_3, add, relu6, x_4, mul], Original ATen: [aten.convolution, aten.add, aten.hardtanh, aten.div, aten.mul]
triton_poi_fused_add_convolution_div_hardtanh_mul_2.run(primals_1, buf4, primals_5, buf5, 256, grid=grid(256), stream=stream0)
buf6 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_3, add], Original ATen: [aten.convolution, aten.add, aten.hardtanh_backward]
triton_poi_fused_add_convolution_hardtanh_backward_3.run(buf4, primals_5, buf6, 16, grid=grid(16), stream=stream0)
del buf4
del primals_5
return (buf5, primals_1, primals_2, primals_4, buf1, buf3, buf6, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 1, 1, 1), (1, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torchvision.transforms import functional as F
import torch.utils.data
from torch import nn
import torch.nn.functional as F
class Hsigmoid(nn.Module):
def __init__(self, inplace=True):
super(Hsigmoid, self).__init__()
self.inplace = inplace
def forward(self, x):
return F.relu6(x + 3.0, inplace=self.inplace) / 6.0
class SEModule(nn.Module):
def __init__(self, channel, reduction=4):
super(SEModule, self).__init__()
self.avg_pool = nn.AdaptiveAvgPool2d(1)
self.fc1 = nn.Conv2d(channel, channel // reduction, kernel_size=1,
padding=0)
self.relu = nn.ReLU(inplace=True)
self.fc2 = nn.Conv2d(channel // reduction, channel, kernel_size=1,
padding=0)
self.hsigmoid = Hsigmoid()
def forward(self, x):
input = x
x = self.avg_pool(x)
x = self.fc1(x)
x = self.relu(x)
x = self.fc2(x)
x = self.hsigmoid(x)
return input * x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'channel': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torchvision.transforms import functional as F
import torch.utils.data
from torch import nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_per_fused_mean_0(in_out_ptr0, in_ptr0, xnumel, rnumel, XBLOCK:
tl.constexpr):
xnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 16 * x0), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp5 = 16.0
tmp6 = tmp4 / tmp5
tl.debug_barrier()
tl.store(in_out_ptr0 + x0, tmp6, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.full([1], 0, tl.int32)
tmp5 = triton_helpers.maximum(tmp4, tmp3)
tl.store(in_out_ptr0 + x0, tmp5, xmask)
@triton.jit
def triton_poi_fused_add_convolution_div_hardtanh_mul_2(in_ptr0, in_ptr1,
in_ptr2, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x4 = xindex // 16
x1 = xindex // 16 % 4
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr1 + x4, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = 3.0
tmp5 = tmp3 + tmp4
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = 6.0
tmp9 = triton_helpers.minimum(tmp7, tmp8)
tmp10 = 0.16666666666666666
tmp11 = tmp9 * tmp10
tmp12 = tmp0 * tmp11
tl.store(out_ptr0 + x3, tmp12, xmask)
@triton.jit
def triton_poi_fused_add_convolution_hardtanh_backward_3(in_ptr0, in_ptr1,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 3.0
tmp4 = tmp2 + tmp3
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tmp7 = 6.0
tmp8 = tmp4 >= tmp7
tmp9 = tmp6 | tmp8
tl.store(out_ptr0 + x2, tmp9, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_3, (1,), (1,))
assert_size_stride(primals_4, (4, 1, 1, 1), (1, 1, 1, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 1, 1), 0)
del buf0
get_raw_stream(0)
triton_per_fused_mean_0[grid(16)](buf1, primals_1, 16, 16, XBLOCK=8,
num_warps=2, num_stages=1)
buf2 = extern_kernels.convolution(buf1, primals_2, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 1, 1), (1, 1, 1, 1))
buf3 = buf2
del buf2
triton_poi_fused_convolution_relu_1[grid(4)](buf3, primals_3, 4,
XBLOCK=4, num_warps=1, num_stages=1)
del primals_3
buf4 = extern_kernels.convolution(buf3, primals_4, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 4, 1, 1), (4, 1, 1, 1))
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_add_convolution_div_hardtanh_mul_2[grid(256)](
primals_1, buf4, primals_5, buf5, 256, XBLOCK=256, num_warps=4,
num_stages=1)
buf6 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.bool)
triton_poi_fused_add_convolution_hardtanh_backward_3[grid(16)](buf4,
primals_5, buf6, 16, XBLOCK=16, num_warps=1, num_stages=1)
del buf4
del primals_5
return buf5, primals_1, primals_2, primals_4, buf1, buf3, buf6
class Hsigmoid(nn.Module):
def __init__(self, inplace=True):
super(Hsigmoid, self).__init__()
self.inplace = inplace
def forward(self, x):
return F.relu6(x + 3.0, inplace=self.inplace) / 6.0
class SEModuleNew(nn.Module):
def __init__(self, channel, reduction=4):
super(SEModuleNew, self).__init__()
self.avg_pool = nn.AdaptiveAvgPool2d(1)
self.fc1 = nn.Conv2d(channel, channel // reduction, kernel_size=1,
padding=0)
self.relu = nn.ReLU(inplace=True)
self.fc2 = nn.Conv2d(channel // reduction, channel, kernel_size=1,
padding=0)
self.hsigmoid = Hsigmoid()
def forward(self, input_0):
primals_2 = self.fc1.weight
primals_3 = self.fc1.bias
primals_4 = self.fc2.weight
primals_5 = self.fc2.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| ljjyxz123/CenterMask | SEModule | false | 7,136 | [
"BSD-2-Clause"
] | 1 | 443eebde30e209eeb3b953f7ef35d3f7f14aaca5 | https://github.com/ljjyxz123/CenterMask/tree/443eebde30e209eeb3b953f7ef35d3f7f14aaca5 | import torch
from torchvision.transforms import functional as F
import torch.utils.data
from torch import nn
import torch.nn.functional as F
class Hsigmoid(nn.Module):
def __init__(self, inplace=True):
super().__init__()
self.inplace = inplace
def forward(self, x):
return F.relu6(x + 3.0, inplace=self.inplace) / 6.0
class Model(nn.Module):
def __init__(self, channel, reduction=4):
super().__init__()
self.avg_pool = nn.AdaptiveAvgPool2d(1)
self.fc1 = nn.Conv2d(channel, channel // reduction, kernel_size=1,
padding=0)
self.relu = nn.ReLU(inplace=True)
self.fc2 = nn.Conv2d(channel // reduction, channel, kernel_size=1,
padding=0)
self.hsigmoid = Hsigmoid()
def forward(self, x):
input = x
x = self.avg_pool(x)
x = self.fc1(x)
x = self.relu(x)
x = self.fc2(x)
x = self.hsigmoid(x)
return input * x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
Conv2d | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ud/cudtupp4xbsxvl5czwt3p2pj3cknjnhtp6x45zymsucnyg3xzdnf.py
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv2d => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%permute, %primals_2, %primals_3, [1, 1], [4, 4], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = (yindex // 4)
tmp0 = tl.load(in_ptr0 + (x2 + (16*y3)), xmask & ymask)
tl.store(out_ptr0 + (y0 + (4*x2) + (64*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/r6/cr6dadrhpyxhbvayrqsgg5jsar4ckn6ku5xyq7mu3ehavjpdivjo.py
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv2d => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%permute, %primals_2, %primals_3, [1, 1], [4, 4], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1296
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 1, 16, 4), torch.float32)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(primals_2, buf0, 16, 16, grid=grid(16, 16), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(reinterpret_tensor(primals_1, (4, 4, 4, 4), (64, 1, 16, 4), 0), buf0, stride=(1, 1), padding=(4, 4), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 9, 9), (324, 1, 36, 4))
del buf0
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf2, primals_3, 1296, grid=grid(1296), stream=stream0)
del primals_3
return (reinterpret_tensor(buf2, (4, 9, 9, 4), (324, 36, 4, 1), 0), primals_2, reinterpret_tensor(primals_1, (4, 4, 4, 4), (64, 1, 16, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.utils
class Conv2d(nn.Module):
def __init__(self, C_in, C_out, kernel_size, padding):
super(Conv2d, self).__init__()
self.conv = nn.Conv2d(C_in, C_out, kernel_size=kernel_size, stride=
1, padding=padding)
def forward(self, x):
return self.conv(x.permute(0, 3, 1, 2)).permute(0, 2, 3, 1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'C_in': 4, 'C_out': 4, 'kernel_size': 4, 'padding': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
import torch.utils
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_0(in_ptr0, out_ptr0, ynumel, xnumel,
YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = yindex // 4
tmp0 = tl.load(in_ptr0 + (x2 + 16 * y3), xmask & ymask)
tl.store(out_ptr0 + (y0 + 4 * x2 + 64 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 1296
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 1, 16, 4), torch.float32)
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(16, 16)](primals_2, buf0, 16,
16, XBLOCK=16, YBLOCK=16, num_warps=4, num_stages=1)
buf1 = extern_kernels.convolution(reinterpret_tensor(primals_1, (4,
4, 4, 4), (64, 1, 16, 4), 0), buf0, stride=(1, 1), padding=(4,
4), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 9, 9), (324, 1, 36, 4))
del buf0
buf2 = buf1
del buf1
triton_poi_fused_convolution_1[grid(1296)](buf2, primals_3, 1296,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_3
return reinterpret_tensor(buf2, (4, 9, 9, 4), (324, 36, 4, 1), 0
), primals_2, reinterpret_tensor(primals_1, (4, 4, 4, 4), (64, 1,
16, 4), 0)
class Conv2dNew(nn.Module):
def __init__(self, C_in, C_out, kernel_size, padding):
super(Conv2dNew, self).__init__()
self.conv = nn.Conv2d(C_in, C_out, kernel_size=kernel_size, stride=
1, padding=padding)
def forward(self, input_0):
primals_1 = self.conv.weight
primals_3 = self.conv.bias
primals_2 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| lorylei/DARTS-et | Conv2d | false | 7,137 | [
"Apache-2.0"
] | 1 | f22cfd53c14afd6ba602b8ecfbff9cdf77fc2ff8 | https://github.com/lorylei/DARTS-et/tree/f22cfd53c14afd6ba602b8ecfbff9cdf77fc2ff8 | import torch
import torch.nn as nn
import torch.utils
class Model(nn.Module):
def __init__(self, C_in, C_out, kernel_size, padding):
super().__init__()
self.conv = nn.Conv2d(C_in, C_out, kernel_size=kernel_size, stride=
1, padding=padding)
def forward(self, x):
return self.conv(x.permute(0, 3, 1, 2)).permute(0, 2, 3, 1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4, 4, 4]
|
GeneralizedMeanPooling | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/jo/cjokg6rcdy3skvr2km4dzpdafioonte45gjnhuttcbrzjku4p4ly.py
# Topologically Sorted Source Nodes: [clamp, x, adaptive_avg_pool2d, pow_2], Original ATen: [aten.clamp, aten.pow, aten.mean]
# Source node to ATen node mapping:
# adaptive_avg_pool2d => mean
# clamp => clamp_min
# pow_2 => pow_2
# x => pow_1
# Graph fragment:
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%arg0_1, 1e-06), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%clamp_min, 3.0), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%pow_1, [-1, -2], True), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%mean, 0.3333333333333333), kwargs = {})
triton_per_fused_clamp_mean_pow_0 = async_compile.triton('triton_per_fused_clamp_mean_pow_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[16, 16],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_clamp_mean_pow_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_clamp_mean_pow_0(in_out_ptr0, in_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 16
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (16*x0)), xmask, other=0.0)
tmp1 = 1e-06
tmp2 = triton_helpers.maximum(tmp0, tmp1)
tmp3 = tmp2 * tmp2
tmp4 = tmp3 * tmp2
tmp5 = tl.broadcast_to(tmp4, [XBLOCK, RBLOCK])
tmp7 = tl.where(xmask, tmp5, 0)
tmp8 = tl.sum(tmp7, 1)[:, None]
tmp9 = 16.0
tmp10 = tmp8 / tmp9
tmp11 = 0.3333333333333333
tmp12 = libdevice.pow(tmp10, tmp11)
tl.debug_barrier()
tl.store(in_out_ptr0 + (x0), tmp12, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 1, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [clamp, x, adaptive_avg_pool2d, pow_2], Original ATen: [aten.clamp, aten.pow, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_clamp_mean_pow_0.run(buf1, arg0_1, 16, 16, grid=grid(16), stream=stream0)
del arg0_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
from torch.optim.lr_scheduler import *
from torch.optim import *
class GeneralizedMeanPooling(nn.Module):
"""Applies a 2D power-average adaptive pooling over an input signal composed of several input planes.
The function computed is: :math:`f(X) = pow(sum(pow(X, p)), 1/p)`
- At p = infinity, one gets Max Pooling
- At p = 1, one gets Average Pooling
The output is of size H x W, for any input size.
The number of output features is equal to the number of input planes.
Args:
output_size: the target output size of the image of the form H x W.
Can be a tuple (H, W) or a single H for a square image H x H
H and W can be either a ``int``, or ``None`` which means the size will
be the same as that of the input.
"""
def __init__(self, norm=3, output_size=1, eps=1e-06):
super(GeneralizedMeanPooling, self).__init__()
assert norm > 0
self.p = float(norm)
self.output_size = output_size
self.eps = eps
def forward(self, x):
x = x.clamp(min=self.eps).pow(self.p)
return torch.nn.functional.adaptive_avg_pool2d(x, self.output_size
).pow(1.0 / self.p)
def __repr__(self):
return self.__class__.__name__ + '(' + str(self.p
) + ', ' + 'output_size=' + str(self.output_size) + ')'
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
from torch.optim.lr_scheduler import *
from torch.optim import *
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_per_fused_clamp_mean_pow_0(in_out_ptr0, in_ptr0, xnumel, rnumel,
XBLOCK: tl.constexpr):
xnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 16 * x0), xmask, other=0.0)
tmp1 = 1e-06
tmp2 = triton_helpers.maximum(tmp0, tmp1)
tmp3 = tmp2 * tmp2
tmp4 = tmp3 * tmp2
tmp5 = tl.broadcast_to(tmp4, [XBLOCK, RBLOCK])
tmp7 = tl.where(xmask, tmp5, 0)
tmp8 = tl.sum(tmp7, 1)[:, None]
tmp9 = 16.0
tmp10 = tmp8 / tmp9
tmp11 = 0.3333333333333333
tmp12 = libdevice.pow(tmp10, tmp11)
tl.debug_barrier()
tl.store(in_out_ptr0 + x0, tmp12, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 1, 1), 0)
del buf0
get_raw_stream(0)
triton_per_fused_clamp_mean_pow_0[grid(16)](buf1, arg0_1, 16, 16,
XBLOCK=8, num_warps=2, num_stages=1)
del arg0_1
return buf1,
class GeneralizedMeanPoolingNew(nn.Module):
"""Applies a 2D power-average adaptive pooling over an input signal composed of several input planes.
The function computed is: :math:`f(X) = pow(sum(pow(X, p)), 1/p)`
- At p = infinity, one gets Max Pooling
- At p = 1, one gets Average Pooling
The output is of size H x W, for any input size.
The number of output features is equal to the number of input planes.
Args:
output_size: the target output size of the image of the form H x W.
Can be a tuple (H, W) or a single H for a square image H x H
H and W can be either a ``int``, or ``None`` which means the size will
be the same as that of the input.
"""
def __init__(self, norm=3, output_size=1, eps=1e-06):
super(GeneralizedMeanPoolingNew, self).__init__()
assert norm > 0
self.p = float(norm)
self.output_size = output_size
self.eps = eps
def __repr__(self):
return self.__class__.__name__ + '(' + str(self.p
) + ', ' + 'output_size=' + str(self.output_size) + ')'
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| lxc86739795/fast-reid | GeneralizedMeanPooling | false | 7,138 | [
"Apache-2.0"
] | 1 | 29178d70c591ef64021f10767eb606f3053156b9 | https://github.com/lxc86739795/fast-reid/tree/29178d70c591ef64021f10767eb606f3053156b9 | import torch
from torch import nn
from torch.optim.lr_scheduler import *
from torch.optim import *
class Model(nn.Module):
"""Applies a 2D power-average adaptive pooling over an input signal composed of several input planes.
The function computed is: :math:`f(X) = pow(sum(pow(X, p)), 1/p)`
- At p = infinity, one gets Max Pooling
- At p = 1, one gets Average Pooling
The output is of size H x W, for any input size.
The number of output features is equal to the number of input planes.
Args:
output_size: the target output size of the image of the form H x W.
Can be a tuple (H, W) or a single H for a square image H x H
H and W can be either a ``int``, or ``None`` which means the size will
be the same as that of the input.
"""
def __init__(self, norm=3, output_size=1, eps=1e-06):
super().__init__()
assert norm > 0
self.p = float(norm)
self.output_size = output_size
self.eps = eps
def forward(self, x):
x = x.clamp(min=self.eps).pow(self.p)
return torch.nn.functional.adaptive_avg_pool2d(x, self.output_size
).pow(1.0 / self.p)
def __repr__(self):
return self.__class__.__name__ + '(' + str(self.p
) + ', ' + 'output_size=' + str(self.output_size) + ')'
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
net2 | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/fp/cfp5jrxxyxrvhcpoq5tio3p5tkhj5ugdrpyur3x4v6meatzih7jn.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_0 = async_compile.triton('triton_poi_fused_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = (yindex // 4)
tmp0 = tl.load(in_ptr0 + (x2 + (16*y3)), xmask & ymask)
tl.store(out_ptr0 + (y0 + (4*x2) + (64*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/f4/cf435ljkvsbmsfqmb2jfkuq4aor2tpycnclm7llj5efimpnzxiwz.py
# Topologically Sorted Source Nodes: [conv2d, x], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d => convolution
# x => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_1 = async_compile.triton('triton_poi_fused_convolution_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4096
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/j3/cj3tcccv2au6fwrk7zrqv6aadd7rde25xioifxhrg3rnq277xw2z.py
# Topologically Sorted Source Nodes: [conv2d_1, x_1], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_1 => convolution_1
# x_1 => relu_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_1,), kwargs = {})
triton_poi_fused_convolution_relu_2 = async_compile.triton('triton_poi_fused_convolution_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16384],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16384
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 256
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/mv/cmvvxenutty4sz3eynxns45xagw56ec2ufreiud5s5acbv345vt5.py
# Topologically Sorted Source Nodes: [conv2d_2, x_2], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_2 => convolution_2
# x_2 => relu_2
# Graph fragment:
# %convolution_2 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_1, %primals_6, %primals_7, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_2,), kwargs = {})
triton_poi_fused_convolution_relu_3 = async_compile.triton('triton_poi_fused_convolution_relu_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8192],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 8192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 128
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/gb/cgbyyfi5lll3xaoryj23lises4aplhm5a4rcgchfqja6bfmq4vbx.py
# Topologically Sorted Source Nodes: [conv2d_3, x_3, x_4], Original ATen: [aten.convolution, aten.relu, aten._log_softmax, aten.threshold_backward]
# Source node to ATen node mapping:
# conv2d_3 => convolution_3
# x_3 => relu_3
# x_4 => amax, sub
# Graph fragment:
# %convolution_3 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_2, %primals_8, %primals_9, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_3 : [num_users=3] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_3,), kwargs = {})
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%relu_3, [1], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%relu_3, %amax), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_3, 0), kwargs = {})
triton_poi_fused__log_softmax_convolution_relu_threshold_backward_4 = async_compile.triton('triton_poi_fused__log_softmax_convolution_relu_threshold_backward_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*i1', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_convolution_relu_threshold_backward_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_convolution_relu_threshold_backward_4(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
x1 = (xindex // 2)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (2*x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr1 + (0))
tmp7 = tl.broadcast_to(tmp6, [XBLOCK])
tmp10 = tl.load(in_ptr0 + (1 + (2*x1)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr1 + (1))
tmp12 = tl.broadcast_to(tmp11, [XBLOCK])
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp8 = tmp5 + tmp7
tmp9 = triton_helpers.maximum(tmp3, tmp8)
tmp13 = tmp10 + tmp12
tmp14 = triton_helpers.maximum(tmp3, tmp13)
tmp15 = triton_helpers.maximum(tmp9, tmp14)
tmp16 = tmp4 - tmp15
tmp17 = 0.0
tmp18 = tmp4 <= tmp17
tl.store(out_ptr0 + (x2), tmp16, xmask)
tl.store(out_ptr1 + (x2), tmp18, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/f6/cf62mfpqwv5pcofy3ro6w6ycq3f2iwdol4c5h7ameyvaum2rwiui.py
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# x_4 => exp, log, sub_1, sum_1
# Graph fragment:
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %log), kwargs = {})
triton_poi_fused__log_softmax_5 = async_compile.triton('triton_poi_fused__log_softmax_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_5(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 8
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 2
y1 = (yindex // 2)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (2*x2) + (32*y1)), xmask & ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + ((2*x2) + (32*y1)), xmask & ymask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (2*x2) + (32*y1)), xmask & ymask, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp6 = tl_math.log(tmp5)
tmp7 = tmp0 - tmp6
tl.store(out_ptr0 + (x2 + (16*y3)), tmp7, xmask & ymask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (64, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_2, (64, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (256, 64, 1, 1), (64, 1, 1, 1))
assert_size_stride(primals_5, (256, ), (1, ))
assert_size_stride(primals_6, (128, 256, 1, 1), (256, 1, 1, 1))
assert_size_stride(primals_7, (128, ), (1, ))
assert_size_stride(primals_8, (2, 128, 1, 1), (128, 1, 1, 1))
assert_size_stride(primals_9, (2, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 1, 16, 4), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
stream0 = get_raw_stream(0)
triton_poi_fused_0.run(primals_3, buf0, 16, 16, grid=grid(16, 16), stream=stream0)
del primals_3
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_1, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 64, 4, 4), (1024, 1, 256, 64))
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [conv2d, x], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_1.run(buf2, primals_2, 4096, grid=grid(4096), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(buf2, primals_4, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 256, 4, 4), (4096, 1, 1024, 256))
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [conv2d_1, x_1], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_2.run(buf4, primals_5, 16384, grid=grid(16384), stream=stream0)
del primals_5
# Topologically Sorted Source Nodes: [conv2d_2], Original ATen: [aten.convolution]
buf5 = extern_kernels.convolution(buf4, primals_6, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf5, (4, 128, 4, 4), (2048, 1, 512, 128))
buf6 = buf5; del buf5 # reuse
# Topologically Sorted Source Nodes: [conv2d_2, x_2], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_3.run(buf6, primals_7, 8192, grid=grid(8192), stream=stream0)
del primals_7
# Topologically Sorted Source Nodes: [conv2d_3], Original ATen: [aten.convolution]
buf7 = extern_kernels.convolution(buf6, primals_8, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf7, (4, 2, 4, 4), (32, 1, 8, 2))
buf8 = empty_strided_cuda((4, 2, 4, 4), (32, 1, 8, 2), torch.float32)
buf10 = empty_strided_cuda((4, 2, 4, 4), (32, 1, 8, 2), torch.bool)
# Topologically Sorted Source Nodes: [conv2d_3, x_3, x_4], Original ATen: [aten.convolution, aten.relu, aten._log_softmax, aten.threshold_backward]
triton_poi_fused__log_softmax_convolution_relu_threshold_backward_4.run(buf7, primals_9, buf8, buf10, 128, grid=grid(128), stream=stream0)
del primals_9
buf9 = reinterpret_tensor(buf7, (4, 2, 4, 4), (32, 16, 4, 1), 0); del buf7 # reuse
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten._log_softmax]
triton_poi_fused__log_softmax_5.run(buf8, buf9, 8, 16, grid=grid(8, 16), stream=stream0)
del buf8
return (buf9, primals_1, buf0, primals_4, primals_6, primals_8, buf2, buf4, buf6, buf9, buf10, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((64, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((256, 64, 1, 1), (64, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((128, 256, 1, 1), (256, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((2, 128, 1, 1), (128, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class net2(nn.Module):
"""
"""
def __init__(self, n_classes=2):
super(net2, self).__init__()
if torch.cuda.is_available():
torch.device('cuda')
else:
torch.device('cpu')
self.n_classes = n_classes
self.conv1 = nn.Conv2d(4, 64, 1)
self.conv2 = nn.Conv2d(64, 256, 1)
self.conv3 = nn.Conv2d(256, 128, 1)
self.conv4 = nn.Conv2d(128, self.n_classes, 1)
self.logsoftmax = nn.LogSoftmax(dim=1)
def forward(self, x):
x = F.relu(self.conv1(x))
x = F.relu(self.conv2(x))
x = F.relu(self.conv3(x))
x = F.relu(self.conv4(x))
x = self.logsoftmax(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = yindex // 4
tmp0 = tl.load(in_ptr0 + (x2 + 16 * y3), xmask & ymask)
tl.store(out_ptr0 + (y0 + 4 * x2 + 64 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_convolution_relu_1(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused_convolution_relu_2(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 256
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused_convolution_relu_3(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 128
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused__log_softmax_convolution_relu_threshold_backward_4(in_ptr0
, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
x1 = xindex // 2
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + 2 * x1, xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr1 + 0)
tmp7 = tl.broadcast_to(tmp6, [XBLOCK])
tmp10 = tl.load(in_ptr0 + (1 + 2 * x1), xmask, eviction_policy='evict_last'
)
tmp11 = tl.load(in_ptr1 + 1)
tmp12 = tl.broadcast_to(tmp11, [XBLOCK])
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp8 = tmp5 + tmp7
tmp9 = triton_helpers.maximum(tmp3, tmp8)
tmp13 = tmp10 + tmp12
tmp14 = triton_helpers.maximum(tmp3, tmp13)
tmp15 = triton_helpers.maximum(tmp9, tmp14)
tmp16 = tmp4 - tmp15
tmp17 = 0.0
tmp18 = tmp4 <= tmp17
tl.store(out_ptr0 + x2, tmp16, xmask)
tl.store(out_ptr1 + x2, tmp18, xmask)
@triton.jit
def triton_poi_fused__log_softmax_5(in_ptr0, out_ptr0, ynumel, xnumel,
YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 8
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 2
y1 = yindex // 2
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 2 * x2 + 32 * y1), xmask & ymask,
eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (2 * x2 + 32 * y1), xmask & ymask,
eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 2 * x2 + 32 * y1), xmask & ymask,
eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp6 = tl_math.log(tmp5)
tmp7 = tmp0 - tmp6
tl.store(out_ptr0 + (x2 + 16 * y3), tmp7, xmask & ymask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (64, 4, 1, 1), (4, 1, 1, 1))
assert_size_stride(primals_2, (64,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (256, 64, 1, 1), (64, 1, 1, 1))
assert_size_stride(primals_5, (256,), (1,))
assert_size_stride(primals_6, (128, 256, 1, 1), (256, 1, 1, 1))
assert_size_stride(primals_7, (128,), (1,))
assert_size_stride(primals_8, (2, 128, 1, 1), (128, 1, 1, 1))
assert_size_stride(primals_9, (2,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 1, 16, 4), torch.float32)
get_raw_stream(0)
triton_poi_fused_0[grid(16, 16)](primals_3, buf0, 16, 16, XBLOCK=16,
YBLOCK=16, num_warps=4, num_stages=1)
del primals_3
buf1 = extern_kernels.convolution(buf0, primals_1, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 64, 4, 4), (1024, 1, 256, 64))
buf2 = buf1
del buf1
triton_poi_fused_convolution_relu_1[grid(4096)](buf2, primals_2,
4096, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf3 = extern_kernels.convolution(buf2, primals_4, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 256, 4, 4), (4096, 1, 1024, 256))
buf4 = buf3
del buf3
triton_poi_fused_convolution_relu_2[grid(16384)](buf4, primals_5,
16384, XBLOCK=256, num_warps=4, num_stages=1)
del primals_5
buf5 = extern_kernels.convolution(buf4, primals_6, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf5, (4, 128, 4, 4), (2048, 1, 512, 128))
buf6 = buf5
del buf5
triton_poi_fused_convolution_relu_3[grid(8192)](buf6, primals_7,
8192, XBLOCK=128, num_warps=4, num_stages=1)
del primals_7
buf7 = extern_kernels.convolution(buf6, primals_8, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf7, (4, 2, 4, 4), (32, 1, 8, 2))
buf8 = empty_strided_cuda((4, 2, 4, 4), (32, 1, 8, 2), torch.float32)
buf10 = empty_strided_cuda((4, 2, 4, 4), (32, 1, 8, 2), torch.bool)
triton_poi_fused__log_softmax_convolution_relu_threshold_backward_4[
grid(128)](buf7, primals_9, buf8, buf10, 128, XBLOCK=128,
num_warps=4, num_stages=1)
del primals_9
buf9 = reinterpret_tensor(buf7, (4, 2, 4, 4), (32, 16, 4, 1), 0)
del buf7
triton_poi_fused__log_softmax_5[grid(8, 16)](buf8, buf9, 8, 16,
XBLOCK=16, YBLOCK=8, num_warps=4, num_stages=1)
del buf8
return (buf9, primals_1, buf0, primals_4, primals_6, primals_8, buf2,
buf4, buf6, buf9, buf10)
class net2New(nn.Module):
"""
"""
def __init__(self, n_classes=2):
super(net2New, self).__init__()
if torch.cuda.is_available():
torch.device('cuda')
else:
torch.device('cpu')
self.n_classes = n_classes
self.conv1 = nn.Conv2d(4, 64, 1)
self.conv2 = nn.Conv2d(64, 256, 1)
self.conv3 = nn.Conv2d(256, 128, 1)
self.conv4 = nn.Conv2d(128, self.n_classes, 1)
self.logsoftmax = nn.LogSoftmax(dim=1)
def forward(self, input_0):
primals_1 = self.conv1.weight
primals_2 = self.conv1.bias
primals_4 = self.conv2.weight
primals_5 = self.conv2.bias
primals_6 = self.conv3.weight
primals_7 = self.conv3.bias
primals_8 = self.conv4.weight
primals_9 = self.conv4.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
| luisesanmartin/dwelling-recognition | net2 | false | 7,139 | [
"MIT"
] | 1 | b2437b64088a26746947c1c88077c96332e7b9c6 | https://github.com/luisesanmartin/dwelling-recognition/tree/b2437b64088a26746947c1c88077c96332e7b9c6 | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
"""
"""
def __init__(self, n_classes=2):
super().__init__()
if torch.cuda.is_available():
torch.device('cuda')
else:
torch.device('cpu')
self.n_classes = n_classes
self.conv1 = nn.Conv2d(4, 64, 1)
self.conv2 = nn.Conv2d(64, 256, 1)
self.conv3 = nn.Conv2d(256, 128, 1)
self.conv4 = nn.Conv2d(128, self.n_classes, 1)
self.logsoftmax = nn.LogSoftmax(dim=1)
def forward(self, x):
x = F.relu(self.conv1(x))
x = F.relu(self.conv2(x))
x = F.relu(self.conv3(x))
x = F.relu(self.conv4(x))
x = self.logsoftmax(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
LSTM | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ms/cmsuzohbg5nq52jnvirovzkvykrzzko5xomu7zyu5e5u2lhegppw.py
# Topologically Sorted Source Nodes: [combined], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# combined => cat
# Graph fragment:
# %cat : [num_users=2] = call_function[target=torch.ops.aten.cat.default](args = ([%primals_1, %primals_2], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 8
x1 = (xindex // 8)
x2 = xindex
tmp0 = x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + ((4*x1) + x0), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr1 + ((4*x1) + ((-4) + x0)), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + (x2), tmp10, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/bf/cbf5h5s73n3ugovaw7yqfdle2h2bfbw3r52thdqgaloet277mdg6.py
# Topologically Sorted Source Nodes: [f_gate, cell_sub, mul, mul_1, cell, tanh_1, hidden], Original ATen: [aten.sigmoid, aten.tanh, aten.mul, aten.add]
# Source node to ATen node mapping:
# cell => add
# cell_sub => tanh
# f_gate => sigmoid
# hidden => mul_2
# mul => mul
# mul_1 => mul_1
# tanh_1 => tanh_1
# Graph fragment:
# %sigmoid : [num_users=3] = call_function[target=torch.ops.aten.sigmoid.default](args = (%addmm,), kwargs = {})
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%addmm,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_5, %sigmoid), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%tanh, %sigmoid), kwargs = {})
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
# %tanh_1 : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%add,), kwargs = {})
# %mul_2 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%tanh_1, %sigmoid), kwargs = {})
triton_poi_fused_add_mul_sigmoid_tanh_1 = async_compile.triton('triton_poi_fused_add_mul_sigmoid_tanh_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_sigmoid_tanh_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_sigmoid_tanh_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask)
tmp2 = tl.sigmoid(tmp1)
tmp3 = tmp0 * tmp2
tmp4 = libdevice.tanh(tmp1)
tmp5 = tmp4 * tmp2
tmp6 = tmp3 + tmp5
tmp7 = libdevice.tanh(tmp6)
tmp8 = tmp7 * tmp2
tl.store(out_ptr0 + (x0), tmp6, xmask)
tl.store(out_ptr1 + (x0), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/b7/cb7iq44xucvx4o4uio3etz5hrrkllxx5igr3vjyglpwcku6mi232.py
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.sigmoid]
# Source node to ATen node mapping:
# output => sigmoid_3
# Graph fragment:
# %add_tensor : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default, %primals_7), kwargs = {})
# %sigmoid_3 : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%add_tensor,), kwargs = {})
triton_poi_fused_sigmoid_2 = async_compile.triton('triton_poi_fused_sigmoid_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sigmoid_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sigmoid_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (2, 8), (8, 1))
assert_size_stride(primals_4, (2, ), (1, ))
assert_size_stride(primals_5, (4, 2), (2, 1))
assert_size_stride(primals_6, (1, 2), (2, 1))
assert_size_stride(primals_7, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [combined], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(primals_1, primals_2, buf0, 32, grid=grid(32), stream=stream0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_4, buf0, reinterpret_tensor(primals_3, (8, 2), (1, 8), 0), alpha=1, beta=1, out=buf1)
del primals_3
del primals_4
buf2 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
buf3 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [f_gate, cell_sub, mul, mul_1, cell, tanh_1, hidden], Original ATen: [aten.sigmoid, aten.tanh, aten.mul, aten.add]
triton_poi_fused_add_mul_sigmoid_tanh_1.run(primals_5, buf1, buf2, buf3, 8, grid=grid(8), stream=stream0)
buf4 = empty_strided_cuda((4, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf3, reinterpret_tensor(primals_6, (2, 1), (1, 2), 0), out=buf4)
buf5 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.sigmoid]
triton_poi_fused_sigmoid_2.run(buf5, primals_7, 4, grid=grid(4), stream=stream0)
del primals_7
return (buf5, buf3, buf2, primals_5, buf0, buf1, buf2, buf3, buf5, primals_6, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((2, 8), (8, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 2), (2, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((1, 2), (2, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.utils.data
import torch.nn
import torch.optim
import torch.nn as nn
from torch.autograd import Variable
class LSTM(nn.Module):
def __init__(self, input_size, hidden_size, output_size=1, cell_size=2):
super(LSTM, self).__init__()
self.hidden_size = hidden_size
self.cell_size = cell_size
self.gate = nn.Linear(input_size + hidden_size, cell_size)
self.output = nn.Linear(cell_size, output_size)
self.sigmoid = nn.Sigmoid()
self.tanh = nn.Tanh()
self.softmax = nn.LogSoftmax()
def forward(self, input, hidden, cell):
combined = torch.cat((input, hidden), 1)
f_gate = self.sigmoid(self.gate(combined))
i_gate = self.sigmoid(self.gate(combined))
o_gate = self.sigmoid(self.gate(combined))
cell_sub = self.tanh(self.gate(combined))
cell = torch.add(torch.mul(cell, f_gate), torch.mul(cell_sub, i_gate))
hidden = torch.mul(self.tanh(cell), o_gate)
output = self.sigmoid(self.output(hidden))
return output, hidden, cell
def initHidden(self, dim_num):
return Variable(torch.zeros(dim_num, self.hidden_size))
def initCell(self, dim_num):
return Variable(torch.zeros(dim_num, self.cell_size))
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4]), torch.rand([4, 2])]
def get_init_inputs():
return [[], {'input_size': 4, 'hidden_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.utils.data
import torch.nn
import torch.optim
import torch.nn as nn
from torch.autograd import Variable
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 8
x1 = xindex // 8
x2 = xindex
tmp0 = x0
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (4 * x1 + x0), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1], 8, tl.int64)
tmp9 = tl.load(in_ptr1 + (4 * x1 + (-4 + x0)), tmp6 & xmask,
eviction_policy='evict_last', other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + x2, tmp10, xmask)
@triton.jit
def triton_poi_fused_add_mul_sigmoid_tanh_1(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask)
tmp2 = tl.sigmoid(tmp1)
tmp3 = tmp0 * tmp2
tmp4 = libdevice.tanh(tmp1)
tmp5 = tmp4 * tmp2
tmp6 = tmp3 + tmp5
tmp7 = libdevice.tanh(tmp6)
tmp8 = tmp7 * tmp2
tl.store(out_ptr0 + x0, tmp6, xmask)
tl.store(out_ptr1 + x0, tmp8, xmask)
@triton.jit
def triton_poi_fused_sigmoid_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + x0, tmp4, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (2, 8), (8, 1))
assert_size_stride(primals_4, (2,), (1,))
assert_size_stride(primals_5, (4, 2), (2, 1))
assert_size_stride(primals_6, (1, 2), (2, 1))
assert_size_stride(primals_7, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 8), (8, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(32)](primals_1, primals_2, buf0, 32,
XBLOCK=32, num_warps=1, num_stages=1)
del primals_1
del primals_2
buf1 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
extern_kernels.addmm(primals_4, buf0, reinterpret_tensor(primals_3,
(8, 2), (1, 8), 0), alpha=1, beta=1, out=buf1)
del primals_3
del primals_4
buf2 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
buf3 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
triton_poi_fused_add_mul_sigmoid_tanh_1[grid(8)](primals_5, buf1,
buf2, buf3, 8, XBLOCK=8, num_warps=1, num_stages=1)
buf4 = empty_strided_cuda((4, 1), (1, 1), torch.float32)
extern_kernels.mm(buf3, reinterpret_tensor(primals_6, (2, 1), (1, 2
), 0), out=buf4)
buf5 = buf4
del buf4
triton_poi_fused_sigmoid_2[grid(4)](buf5, primals_7, 4, XBLOCK=4,
num_warps=1, num_stages=1)
del primals_7
return buf5, buf3, buf2, primals_5, buf0, buf1, buf2, buf3, buf5, primals_6
class LSTMNew(nn.Module):
def __init__(self, input_size, hidden_size, output_size=1, cell_size=2):
super(LSTMNew, self).__init__()
self.hidden_size = hidden_size
self.cell_size = cell_size
self.gate = nn.Linear(input_size + hidden_size, cell_size)
self.output = nn.Linear(cell_size, output_size)
self.sigmoid = nn.Sigmoid()
self.tanh = nn.Tanh()
self.softmax = nn.LogSoftmax()
def initHidden(self, dim_num):
return Variable(torch.zeros(dim_num, self.hidden_size))
def initCell(self, dim_num):
return Variable(torch.zeros(dim_num, self.cell_size))
def forward(self, input_0, input_1, input_2):
primals_3 = self.gate.weight
primals_4 = self.gate.bias
primals_6 = self.output.weight
primals_7 = self.output.bias
primals_1 = input_0
primals_2 = input_1
primals_5 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0], output[1], output[2]
| lwaekfjlk/Light-the-Torch | LSTM | false | 7,140 | [
"MIT"
] | 1 | eed1df3d28016aee86385959b5e94e2108ee0571 | https://github.com/lwaekfjlk/Light-the-Torch/tree/eed1df3d28016aee86385959b5e94e2108ee0571 | import torch
import torch.utils.data
import torch.nn
import torch.optim
import torch.nn as nn
from torch.autograd import Variable
class Model(nn.Module):
def __init__(self, input_size, hidden_size, output_size=1, cell_size=2):
super().__init__()
self.hidden_size = hidden_size
self.cell_size = cell_size
self.gate = nn.Linear(input_size + hidden_size, cell_size)
self.output = nn.Linear(cell_size, output_size)
self.sigmoid = nn.Sigmoid()
self.tanh = nn.Tanh()
self.softmax = nn.LogSoftmax()
def forward(self, input, hidden, cell):
combined = torch.cat((input, hidden), 1)
f_gate = self.sigmoid(self.gate(combined))
i_gate = self.sigmoid(self.gate(combined))
o_gate = self.sigmoid(self.gate(combined))
cell_sub = self.tanh(self.gate(combined))
cell = torch.add(torch.mul(cell, f_gate), torch.mul(cell_sub, i_gate))
hidden = torch.mul(self.tanh(cell), o_gate)
output = self.sigmoid(self.output(hidden))
return output, hidden, cell
def initHidden(self, dim_num):
return Variable(torch.zeros(dim_num, self.hidden_size))
def initCell(self, dim_num):
return Variable(torch.zeros(dim_num, self.cell_size))
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4]), torch.rand([4, 2])]
def get_init_inputs():
return [4, 4]
|
Attloss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/af/caf6j66yckvq7w2p5nlwms3itxkh7ciedjybplrzlphkhgszee2a.py
# Topologically Sorted Source Nodes: [sub_1, pow_2, loss_att_1, loss_att_2, mul_1], Original ATen: [aten.sub, aten.pow, aten.mean, aten.clamp, aten.mul]
# Source node to ATen node mapping:
# loss_att_1 => mean_1
# loss_att_2 => clamp_max
# mul_1 => mul_1
# pow_2 => pow_2
# sub_1 => sub_1
# Graph fragment:
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg1_1, %arg2_1), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_1, 2), kwargs = {})
# %mean_1 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%pow_2,), kwargs = {})
# %clamp_max : [num_users=1] = call_function[target=torch.ops.aten.clamp_max.default](args = (%mean_1, 30), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%clamp_max, 10), kwargs = {})
triton_per_fused_clamp_mean_mul_pow_sub_0 = async_compile.triton('triton_per_fused_clamp_mean_mul_pow_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_clamp_mean_mul_pow_sub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_clamp_mean_mul_pow_sub_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = tl.load(in_ptr1 + (r0), None)
tmp2 = tmp0 - tmp1
tmp3 = tmp2 * tmp2
tmp4 = tl.broadcast_to(tmp3, [RBLOCK])
tmp6 = triton_helpers.promote_to_tensor(tl.sum(tmp4, 0))
tmp7 = 256.0
tmp8 = tmp6 / tmp7
tmp9 = 30.0
tmp10 = triton_helpers.minimum(tmp8, tmp9)
tmp11 = 10.0
tmp12 = tmp10 * tmp11
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp12, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [sub_1, pow_2, loss_att_1, loss_att_2, mul_1], Original ATen: [aten.sub, aten.pow, aten.mean, aten.clamp, aten.mul]
stream0 = get_raw_stream(0)
triton_per_fused_clamp_mean_mul_pow_sub_0.run(buf1, arg1_1, arg2_1, 1, 256, grid=grid(1), stream=stream0)
del arg1_1
del arg2_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional
class Attloss(nn.Module):
def __init__(self):
super(Attloss, self).__init__()
self.bce = nn.BCEWithLogitsLoss()
def forward(self, x_org, y_mask, att):
loss_att = ((x_org * y_mask[:, 1, ...].unsqueeze(dim=1) - att) ** 2
).mean()
loss_att = ((x_org - att) ** 2).mean()
loss_att = torch.clamp(loss_att, max=30)
return 10 * loss_att
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.nn.functional
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_clamp_mean_mul_pow_sub_0(in_out_ptr0, in_ptr0, in_ptr1,
xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = tl.load(in_ptr1 + r0, None)
tmp2 = tmp0 - tmp1
tmp3 = tmp2 * tmp2
tmp4 = tl.broadcast_to(tmp3, [RBLOCK])
tmp6 = triton_helpers.promote_to_tensor(tl.sum(tmp4, 0))
tmp7 = 256.0
tmp8 = tmp6 / tmp7
tmp9 = 30.0
tmp10 = triton_helpers.minimum(tmp8, tmp9)
tmp11 = 10.0
tmp12 = tmp10 * tmp11
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp12, None)
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_per_fused_clamp_mean_mul_pow_sub_0[grid(1)](buf1, arg1_1,
arg2_1, 1, 256, num_warps=2, num_stages=1)
del arg1_1
del arg2_1
return buf1,
class AttlossNew(nn.Module):
def __init__(self):
super(AttlossNew, self).__init__()
self.bce = nn.BCEWithLogitsLoss()
def forward(self, input_0, input_1, input_2):
arg0_1 = input_0
arg1_1 = input_1
arg2_1 = input_2
output = call([arg0_1, arg1_1, arg2_1])
return output[0]
| lvxiuwang/ferattention | Attloss | false | 7,141 | [
"MIT"
] | 1 | 02e97df4a12129ed6706bddf0d2109650eae8765 | https://github.com/lvxiuwang/ferattention/tree/02e97df4a12129ed6706bddf0d2109650eae8765 | import torch
import torch.nn as nn
import torch.nn.functional
class Model(nn.Module):
def __init__(self):
super().__init__()
self.bce = nn.BCEWithLogitsLoss()
def forward(self, x_org, y_mask, att):
loss_att = ((x_org * y_mask[:, 1, ...].unsqueeze(dim=1) - att) ** 2
).mean()
loss_att = ((x_org - att) ** 2).mean()
loss_att = torch.clamp(loss_att, max=30)
return 10 * loss_att
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return []
|
SOSLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/l7/cl7r5hd5kpfmxo3gnbc4qcrsuttbgppzbejaxa4raarshdtjz4qc.py
# Topologically Sorted Source Nodes: [sub, pow_1, dist_an, sub_1, pow_2, dist_pn, sub_2, pow_3, sum_3, pow_4, truediv], Original ATen: [aten.sub, aten.pow, aten.sum, aten.div]
# Source node to ATen node mapping:
# dist_an => sum_1
# dist_pn => sum_2
# pow_1 => pow_1
# pow_2 => pow_2
# pow_3 => pow_3
# pow_4 => pow_4
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# sum_3 => sum_3
# truediv => div
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %arg1_1), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub, 2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [1]), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg2_1, %arg1_1), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_1, 2), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_2, [1]), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sum_1, %sum_2), kwargs = {})
# %pow_3 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_2, 2), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%pow_3,), kwargs = {})
# %pow_4 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sum_3, 0.5), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%pow_4, 4), kwargs = {})
triton_per_fused_div_pow_sub_sum_0 = async_compile.triton('triton_per_fused_div_pow_sub_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {4: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 5), equal_to_1=(4,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_div_pow_sub_sum_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 12, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_div_pow_sub_sum_0(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex % 16
r1 = (rindex // 16)
r2 = rindex
tmp0 = tl.load(in_ptr0 + (r0 + (64*r1)), None)
tmp1 = tl.load(in_ptr1 + (r0 + (64*r1)), None)
tmp4 = tl.load(in_ptr0 + (16 + r0 + (64*r1)), None)
tmp5 = tl.load(in_ptr1 + (16 + r0 + (64*r1)), None)
tmp9 = tl.load(in_ptr0 + (32 + r0 + (64*r1)), None)
tmp10 = tl.load(in_ptr1 + (32 + r0 + (64*r1)), None)
tmp14 = tl.load(in_ptr0 + (48 + r0 + (64*r1)), None)
tmp15 = tl.load(in_ptr1 + (48 + r0 + (64*r1)), None)
tmp19 = tl.load(in_ptr2 + (r0 + (64*r1)), None)
tmp22 = tl.load(in_ptr2 + (16 + r0 + (64*r1)), None)
tmp26 = tl.load(in_ptr2 + (32 + r0 + (64*r1)), None)
tmp30 = tl.load(in_ptr2 + (48 + r0 + (64*r1)), None)
tmp2 = tmp0 - tmp1
tmp3 = tmp2 * tmp2
tmp6 = tmp4 - tmp5
tmp7 = tmp6 * tmp6
tmp8 = tmp3 + tmp7
tmp11 = tmp9 - tmp10
tmp12 = tmp11 * tmp11
tmp13 = tmp8 + tmp12
tmp16 = tmp14 - tmp15
tmp17 = tmp16 * tmp16
tmp18 = tmp13 + tmp17
tmp20 = tmp19 - tmp1
tmp21 = tmp20 * tmp20
tmp23 = tmp22 - tmp5
tmp24 = tmp23 * tmp23
tmp25 = tmp21 + tmp24
tmp27 = tmp26 - tmp10
tmp28 = tmp27 * tmp27
tmp29 = tmp25 + tmp28
tmp31 = tmp30 - tmp15
tmp32 = tmp31 * tmp31
tmp33 = tmp29 + tmp32
tmp34 = tmp18 - tmp33
tmp35 = tmp34 * tmp34
tmp36 = tl.broadcast_to(tmp35, [XBLOCK, RBLOCK])
tmp38 = tl.sum(tmp36, 1)[:, None]
tmp39 = libdevice.sqrt(tmp38)
tmp40 = 0.25
tmp41 = tmp39 * tmp40
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp41, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [sub, pow_1, dist_an, sub_1, pow_2, dist_pn, sub_2, pow_3, sum_3, pow_4, truediv], Original ATen: [aten.sub, aten.pow, aten.sum, aten.div]
stream0 = get_raw_stream(0)
triton_per_fused_div_pow_sub_sum_0.run(buf2, arg0_1, arg1_1, arg2_1, 1, 64, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
del arg2_1
return (buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
class SOSLoss(nn.Module):
def __init__(self):
super().__init__()
def forward(self, anchors, positives, negatives):
dist_an = torch.sum(torch.pow(anchors - negatives, 2), dim=1)
dist_pn = torch.sum(torch.pow(positives - negatives, 2), dim=1)
nq = anchors.size(dim=0)
return torch.sum(torch.pow(dist_an - dist_pn, 2)) ** 0.5 / nq
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_div_pow_sub_sum_0(in_out_ptr0, in_ptr0, in_ptr1,
in_ptr2, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex % 16
r1 = rindex // 16
tmp0 = tl.load(in_ptr0 + (r0 + 64 * r1), None)
tmp1 = tl.load(in_ptr1 + (r0 + 64 * r1), None)
tmp4 = tl.load(in_ptr0 + (16 + r0 + 64 * r1), None)
tmp5 = tl.load(in_ptr1 + (16 + r0 + 64 * r1), None)
tmp9 = tl.load(in_ptr0 + (32 + r0 + 64 * r1), None)
tmp10 = tl.load(in_ptr1 + (32 + r0 + 64 * r1), None)
tmp14 = tl.load(in_ptr0 + (48 + r0 + 64 * r1), None)
tmp15 = tl.load(in_ptr1 + (48 + r0 + 64 * r1), None)
tmp19 = tl.load(in_ptr2 + (r0 + 64 * r1), None)
tmp22 = tl.load(in_ptr2 + (16 + r0 + 64 * r1), None)
tmp26 = tl.load(in_ptr2 + (32 + r0 + 64 * r1), None)
tmp30 = tl.load(in_ptr2 + (48 + r0 + 64 * r1), None)
tmp2 = tmp0 - tmp1
tmp3 = tmp2 * tmp2
tmp6 = tmp4 - tmp5
tmp7 = tmp6 * tmp6
tmp8 = tmp3 + tmp7
tmp11 = tmp9 - tmp10
tmp12 = tmp11 * tmp11
tmp13 = tmp8 + tmp12
tmp16 = tmp14 - tmp15
tmp17 = tmp16 * tmp16
tmp18 = tmp13 + tmp17
tmp20 = tmp19 - tmp1
tmp21 = tmp20 * tmp20
tmp23 = tmp22 - tmp5
tmp24 = tmp23 * tmp23
tmp25 = tmp21 + tmp24
tmp27 = tmp26 - tmp10
tmp28 = tmp27 * tmp27
tmp29 = tmp25 + tmp28
tmp31 = tmp30 - tmp15
tmp32 = tmp31 * tmp31
tmp33 = tmp29 + tmp32
tmp34 = tmp18 - tmp33
tmp35 = tmp34 * tmp34
tmp36 = tl.broadcast_to(tmp35, [XBLOCK, RBLOCK])
tmp38 = tl.sum(tmp36, 1)[:, None]
tmp39 = libdevice.sqrt(tmp38)
tmp40 = 0.25
tmp41 = tmp39 * tmp40
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp41, None)
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1
del buf1
get_raw_stream(0)
triton_per_fused_div_pow_sub_sum_0[grid(1)](buf2, arg0_1, arg1_1,
arg2_1, 1, 64, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
del arg2_1
return buf2,
class SOSLossNew(nn.Module):
def __init__(self):
super().__init__()
def forward(self, input_0, input_1, input_2):
arg0_1 = input_0
arg1_1 = input_1
arg2_1 = input_2
output = call([arg0_1, arg1_1, arg2_1])
return output[0]
| lulor/project_vg | SOSLoss | false | 7,142 | [
"MIT"
] | 1 | 27b0c3b3038c5a666dde516a0a265ae8ddf2059f | https://github.com/lulor/project_vg/tree/27b0c3b3038c5a666dde516a0a265ae8ddf2059f | import torch
from torch import nn
class Model(nn.Module):
def __init__(self):
super().__init__()
def forward(self, anchors, positives, negatives):
dist_an = torch.sum(torch.pow(anchors - negatives, 2), dim=1)
dist_pn = torch.sum(torch.pow(positives - negatives, 2), dim=1)
nq = anchors.size(dim=0)
return torch.sum(torch.pow(dist_an - dist_pn, 2)) ** 0.5 / nq
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return []
|
AttentionLayer | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ut/cuthdky2gmlcllypdu6te7qddvqxmdfttriaxjae3jm7vigvse2t.py
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# linear_1 => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x2 = (xindex // 16)
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (4*x2)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x3), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/7b/c7bf34fgn2dhohe7ejneqlees25vyq6sbe4c5lfvoehzliak2nz6.py
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# linear_1 => add
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_3, %primals_6), kwargs = {})
triton_poi_fused_add_1 = async_compile.triton('triton_poi_fused_add_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/rh/crhb3fabztrl26dryvz44mbpy6ti3grcwx6xt22njk7wba7qwjsr.py
# Topologically Sorted Source Nodes: [add, x], Original ATen: [aten.add, aten.tanh]
# Source node to ATen node mapping:
# add => add_1
# x => tanh
# Graph fragment:
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, %getitem_2), kwargs = {})
# %tanh : [num_users=2] = call_function[target=torch.ops.aten.tanh.default](args = (%add_1,), kwargs = {})
triton_poi_fused_add_tanh_2 = async_compile.triton('triton_poi_fused_add_tanh_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_tanh_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_tanh_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask)
tmp2 = tmp0 + tmp1
tmp3 = libdevice.tanh(tmp2)
tl.store(in_out_ptr0 + (x0), tmp3, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/lt/cltwbpokq7b7gvah2tjf27qlzw6vpmwfuzs3xfk7mhbxym753kvi.py
# Topologically Sorted Source Nodes: [a], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# a => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%squeeze, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%squeeze, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_3 = async_compile.triton('triton_poi_fused__softmax_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/rr/crrmj7r54x5uk325xkhuskxp4m5prz3fpx53yc2st4o5pwbhq32p.py
# Topologically Sorted Source Nodes: [a], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# a => div, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_4 = async_compile.triton('triton_poi_fused__softmax_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, ), (1, ))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
assert_size_stride(primals_7, (1, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_4, reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_3
del primals_4
# Topologically Sorted Source Nodes: [dropout], Original ATen: [aten.native_dropout]
buf1 = torch.ops.aten.native_dropout.default(reinterpret_tensor(buf0, (4, 4, 4), (16, 4, 1), 0), 0.5, True)
buf2 = buf1[0]
buf3 = buf1[1]
del buf1
buf4 = reinterpret_tensor(buf0, (4, 4, 4), (16, 4, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(primals_2, buf4, 64, grid=grid(64), stream=stream0)
del primals_2
buf5 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf4, (16, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf5)
del primals_5
buf6 = reinterpret_tensor(buf5, (4, 4, 4), (16, 4, 1), 0); del buf5 # reuse
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.add]
triton_poi_fused_add_1.run(buf6, primals_6, 64, grid=grid(64), stream=stream0)
del primals_6
# Topologically Sorted Source Nodes: [linear_1, dropout_1], Original ATen: [aten.add, aten.native_dropout]
buf7 = torch.ops.aten.native_dropout.default(buf6, 0.5, True)
del buf6
buf8 = buf7[0]
buf9 = buf7[1]
del buf7
buf10 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [add, x], Original ATen: [aten.add, aten.tanh]
triton_poi_fused_add_tanh_2.run(buf10, buf8, 64, grid=grid(64), stream=stream0)
del buf8
buf11 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_2], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf10, (16, 4), (4, 1), 0), reinterpret_tensor(primals_7, (4, 1), (1, 4), 0), out=buf11)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.native_dropout]
buf12 = torch.ops.aten.native_dropout.default(reinterpret_tensor(buf11, (4, 4, 1), (4, 1, 1), 0), 0.5, True)
buf13 = buf12[0]
buf14 = buf12[1]
del buf12
buf15 = reinterpret_tensor(buf11, (4, 4), (4, 1), 0); del buf11 # reuse
# Topologically Sorted Source Nodes: [a], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf13, buf15, 16, grid=grid(16), stream=stream0)
buf16 = reinterpret_tensor(buf13, (4, 4), (4, 1), 0); del buf13 # reuse
# Topologically Sorted Source Nodes: [a], Original ATen: [aten._softmax]
triton_poi_fused__softmax_4.run(buf15, buf16, 16, grid=grid(16), stream=stream0)
del buf15
return (buf16, reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), buf3, reinterpret_tensor(buf4, (16, 4), (4, 1), 0), buf9, buf10, buf14, buf16, primals_7, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((1, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class AttentionLayer(nn.Module):
def __init__(self, hidden_dim_en, hidden_dim_de, projected_size):
super(AttentionLayer, self).__init__()
self.linear1 = nn.Linear(hidden_dim_en, projected_size)
self.linear2 = nn.Linear(hidden_dim_de, projected_size)
self.linear3 = nn.Linear(projected_size, 1, False)
def forward(self, out_e, h):
"""
out_e: batch_size * num_frames * en_hidden_dim
h : batch_size * de_hidden_dim
"""
assert out_e.size(0) == h.size(0)
batch_size, num_frames, _ = out_e.size()
hidden_dim = h.size(1)
h_att = h.unsqueeze(1).expand(batch_size, num_frames, hidden_dim)
x = F.tanh(F.dropout(self.linear1(out_e)) + F.dropout(self.linear2(
h_att)))
x = F.dropout(self.linear3(x))
a = F.softmax(x.squeeze(2))
return a
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'hidden_dim_en': 4, 'hidden_dim_de': 4, 'projected_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x2 = xindex // 16
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 4 * x2), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + x3, tmp0, xmask)
@triton.jit
def triton_poi_fused_add_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_add_tanh_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask)
tmp2 = tmp0 + tmp1
tmp3 = libdevice.tanh(tmp2)
tl.store(in_out_ptr0 + x0, tmp3, xmask)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_4(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4,), (1,))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4,), (1,))
assert_size_stride(primals_7, (1, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_4, reinterpret_tensor(primals_1, (16,
4), (4, 1), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
del primals_3
del primals_4
buf1 = torch.ops.aten.native_dropout.default(reinterpret_tensor(
buf0, (4, 4, 4), (16, 4, 1), 0), 0.5, True)
buf2 = buf1[0]
buf3 = buf1[1]
del buf1
buf4 = reinterpret_tensor(buf0, (4, 4, 4), (16, 4, 1), 0)
del buf0
get_raw_stream(0)
triton_poi_fused_clone_0[grid(64)](primals_2, buf4, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_2
buf5 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf4, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf5)
del primals_5
buf6 = reinterpret_tensor(buf5, (4, 4, 4), (16, 4, 1), 0)
del buf5
triton_poi_fused_add_1[grid(64)](buf6, primals_6, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_6
buf7 = torch.ops.aten.native_dropout.default(buf6, 0.5, True)
del buf6
buf8 = buf7[0]
buf9 = buf7[1]
del buf7
buf10 = buf2
del buf2
triton_poi_fused_add_tanh_2[grid(64)](buf10, buf8, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf8
buf11 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf10, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_7, (4, 1), (1, 4), 0), out=buf11)
buf12 = torch.ops.aten.native_dropout.default(reinterpret_tensor(
buf11, (4, 4, 1), (4, 1, 1), 0), 0.5, True)
buf13 = buf12[0]
buf14 = buf12[1]
del buf12
buf15 = reinterpret_tensor(buf11, (4, 4), (4, 1), 0)
del buf11
triton_poi_fused__softmax_3[grid(16)](buf13, buf15, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf16 = reinterpret_tensor(buf13, (4, 4), (4, 1), 0)
del buf13
triton_poi_fused__softmax_4[grid(16)](buf15, buf16, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del buf15
return buf16, reinterpret_tensor(primals_1, (16, 4), (4, 1), 0
), buf3, reinterpret_tensor(buf4, (16, 4), (4, 1), 0
), buf9, buf10, buf14, buf16, primals_7
class AttentionLayerNew(nn.Module):
def __init__(self, hidden_dim_en, hidden_dim_de, projected_size):
super(AttentionLayerNew, self).__init__()
self.linear1 = nn.Linear(hidden_dim_en, projected_size)
self.linear2 = nn.Linear(hidden_dim_de, projected_size)
self.linear3 = nn.Linear(projected_size, 1, False)
def forward(self, input_0, input_1):
primals_2 = self.linear1.weight
primals_4 = self.linear1.bias
primals_3 = self.linear2.weight
primals_6 = self.linear2.bias
primals_7 = self.linear3.weight
primals_1 = input_0
primals_5 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0]
| lost-person/AREL | AttentionLayer | false | 7,143 | [
"MIT"
] | 1 | cee8bc542a2226f41fcbf65ed805fd585512689d | https://github.com/lost-person/AREL/tree/cee8bc542a2226f41fcbf65ed805fd585512689d | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self, hidden_dim_en, hidden_dim_de, projected_size):
super().__init__()
self.linear1 = nn.Linear(hidden_dim_en, projected_size)
self.linear2 = nn.Linear(hidden_dim_de, projected_size)
self.linear3 = nn.Linear(projected_size, 1, False)
def forward(self, out_e, h):
"""
out_e: batch_size * num_frames * en_hidden_dim
h : batch_size * de_hidden_dim
"""
assert out_e.size(0) == h.size(0)
batch_size, num_frames, _ = out_e.size()
hidden_dim = h.size(1)
h_att = h.unsqueeze(1).expand(batch_size, num_frames, hidden_dim)
x = F.tanh(F.dropout(self.linear1(out_e)) + F.dropout(self.linear2(
h_att)))
x = F.dropout(self.linear3(x))
a = F.softmax(x.squeeze(2))
return a
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [4, 4, 4]
|
MultiHeadAttention | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/yd/cydbtjoq352gcolmflbvu2nqkda7xg7q5hnvltb47jsg5dbmubym.py
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (16*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/s2/cs2rk3o3kmhydx4oijp6rsdb5atcrq5axy4adadrpl7gkt7scies.py
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# atte_weights => exp
# Graph fragment:
# %mul_tensor : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_11, 1), kwargs = {})
# %amax_default : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_tensor, [-1], True), kwargs = {})
# %sub_tensor : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_tensor, %amax_default), kwargs = {})
# %div_tensor : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_tensor, 1.0), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%div_tensor,), kwargs = {})
triton_poi_fused__softmax_1 = async_compile.triton('triton_poi_fused__softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp3 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp14 * tmp1
tmp16 = tl_math.exp(tmp15)
tl.store(out_ptr0 + (x2), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/3f/c3fx6bzkalkw7u7askqdnz4rzlcoyqiec4r434sjc5x3axxgkrmr.py
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# atte_weights => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_2 = async_compile.triton('triton_poi_fused__softmax_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/tc/ctcugqu2nbqlxcf2thnspnnypxifbalbzmclmutd5vaxdes2oyyk.py
# Topologically Sorted Source Nodes: [query_5, query_6], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# query_5 => add
# query_6 => var_mean
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_17, %primals_1), kwargs = {})
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add, [2]), kwargs = {correction: 0, keepdim: True})
triton_poi_fused_add_native_layer_norm_3 = async_compile.triton('triton_poi_fused_add_native_layer_norm_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_3(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + (x0), tmp16, xmask)
tl.store(out_ptr1 + (x0), tmp28, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ki/ckiemul6tf32aklelekgb7nubdcitlamu5mdffv4u2b2odledpcc.py
# Topologically Sorted Source Nodes: [query_5, query_6], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# query_5 => add
# query_6 => add_1, add_2, mul, mul_1, rsqrt, sub_1
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_17, %primals_1), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-08), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %getitem_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_6), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_7), kwargs = {})
triton_poi_fused_add_native_layer_norm_4 = async_compile.triton('triton_poi_fused_add_native_layer_norm_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_4(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-08
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + (x2), tmp13, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
assert_size_stride(primals_7, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [query], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf0)
del primals_2
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [key], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf1)
del primals_3
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [value], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf2)
del primals_4
buf3 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(buf0, buf3, 16, 4, grid=grid(16, 4), stream=stream0)
buf4 = reinterpret_tensor(buf0, (4, 4, 1, 4), (16, 4, 4, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(buf1, buf4, 16, 4, grid=grid(16, 4), stream=stream0)
buf5 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf4, (16, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf5, buf6, 256, grid=grid(256), stream=stream0)
buf7 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf5 # reuse
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf6, buf7, 256, grid=grid(256), stream=stream0)
del buf6
buf8 = reinterpret_tensor(buf1, (4, 4, 4, 1), (16, 4, 1, 1), 0); del buf1 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(buf2, buf8, 16, 4, grid=grid(16, 4), stream=stream0)
buf9 = reinterpret_tensor(buf2, (16, 4, 1), (4, 1, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf7, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf8, (16, 4, 1), (4, 1, 0), 0), out=buf9)
buf10 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(buf9, buf10, 16, 4, grid=grid(16, 4), stream=stream0)
buf11 = reinterpret_tensor(buf9, (16, 4), (4, 1), 0); del buf9 # reuse
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf10, (16, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf11)
buf12 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf13 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
# Topologically Sorted Source Nodes: [query_5, query_6], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_3.run(buf11, primals_1, buf12, buf13, 16, grid=grid(16), stream=stream0)
buf14 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [query_5, query_6], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_4.run(buf11, primals_1, buf12, buf13, primals_6, primals_7, buf14, 64, grid=grid(64), stream=stream0)
del buf12
del buf13
del primals_7
return (buf14, buf7, primals_1, primals_6, buf7, reinterpret_tensor(buf10, (16, 4), (4, 1), 0), buf11, primals_5, reinterpret_tensor(buf8, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf3, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super(ScaledDotProductAttention, self).__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(MultiHeadAttention, self).__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'embedding_dim': 4, 'reduced_dim': 4, 'n_head': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 16 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp3 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp14 * tmp1
tmp16 = tl_math.exp(tmp15)
tl.store(out_ptr0 + x2, tmp16, xmask)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_3(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp12 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + x0, tmp16, xmask)
tl.store(out_ptr1 + x0, tmp28, xmask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_4(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-08
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + x2, tmp13, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4,), (1,))
assert_size_stride(primals_7, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf0)
del primals_2
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf1)
del primals_3
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf2)
del primals_4
buf3 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(16, 4)](buf0, buf3, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf4 = reinterpret_tensor(buf0, (4, 4, 1, 4), (16, 4, 4, 1), 0)
del buf0
triton_poi_fused_clone_0[grid(16, 4)](buf1, buf4, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf5 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf4, (16, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_1[grid(256)](buf5, buf6, 256, XBLOCK=256,
num_warps=4, num_stages=1)
buf7 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf5
triton_poi_fused__softmax_2[grid(256)](buf6, buf7, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del buf6
buf8 = reinterpret_tensor(buf1, (4, 4, 4, 1), (16, 4, 1, 1), 0)
del buf1
triton_poi_fused_clone_0[grid(16, 4)](buf2, buf8, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf9 = reinterpret_tensor(buf2, (16, 4, 1), (4, 1, 1), 0)
del buf2
extern_kernels.bmm(reinterpret_tensor(buf7, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf8, (16, 4, 1), (4, 1, 0), 0), out=buf9)
buf10 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_clone_0[grid(16, 4)](buf9, buf10, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf11 = reinterpret_tensor(buf9, (16, 4), (4, 1), 0)
del buf9
extern_kernels.mm(reinterpret_tensor(buf10, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf11)
buf12 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf13 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
triton_poi_fused_add_native_layer_norm_3[grid(16)](buf11, primals_1,
buf12, buf13, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf14 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_add_native_layer_norm_4[grid(64)](buf11, primals_1,
buf12, buf13, primals_6, primals_7, buf14, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf12
del buf13
del primals_7
return buf14, buf7, primals_1, primals_6, buf7, reinterpret_tensor(buf10,
(16, 4), (4, 1), 0), buf11, primals_5, reinterpret_tensor(buf8, (16,
1, 4), (4, 1, 1), 0), reinterpret_tensor(buf3, (16, 1, 4), (4, 1, 1), 0
), reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 4), 0)
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super(ScaledDotProductAttention, self).__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttentionNew(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(MultiHeadAttentionNew, self).__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, input_0):
primals_2 = self.Wq.weight
primals_3 = self.Wk.weight
primals_4 = self.Wv.weight
primals_5 = self.Wo.weight
primals_6 = self.ln.weight
primals_7 = self.ln.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0], output[1]
| luyu-fan/LRCM | MultiHeadAttention | false | 7,144 | [
"MIT"
] | 1 | 6b0e4d7998bc4969afa764eb753077e3f858f1ba | https://github.com/luyu-fan/LRCM/tree/6b0e4d7998bc4969afa764eb753077e3f858f1ba | import math
import torch
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super().__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class Model(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super().__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [4, 4, 4]
|
ComprehensionLayer_step3 | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/yd/cydbtjoq352gcolmflbvu2nqkda7xg7q5hnvltb47jsg5dbmubym.py
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (16*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/s2/cs2rk3o3kmhydx4oijp6rsdb5atcrq5axy4adadrpl7gkt7scies.py
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# atte_weights => exp
# Graph fragment:
# %mul_tensor : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_11, 1), kwargs = {})
# %amax_default : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_tensor, [-1], True), kwargs = {})
# %sub_tensor : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_tensor, %amax_default), kwargs = {})
# %div_tensor : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_tensor, 1.0), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%div_tensor,), kwargs = {})
triton_poi_fused__softmax_1 = async_compile.triton('triton_poi_fused__softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp3 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp14 * tmp1
tmp16 = tl_math.exp(tmp15)
tl.store(out_ptr0 + (x2), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/3f/c3fx6bzkalkw7u7askqdnz4rzlcoyqiec4r434sjc5x3axxgkrmr.py
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# atte_weights => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_2 = async_compile.triton('triton_poi_fused__softmax_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/tc/ctcugqu2nbqlxcf2thnspnnypxifbalbzmclmutd5vaxdes2oyyk.py
# Topologically Sorted Source Nodes: [hig_vectors_1, hig_vectors_2], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# hig_vectors_1 => add
# hig_vectors_2 => var_mean
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %view_17), kwargs = {})
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add, [2]), kwargs = {correction: 0, keepdim: True})
triton_poi_fused_add_native_layer_norm_3 = async_compile.triton('triton_poi_fused_add_native_layer_norm_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_3(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + (x0), tmp16, xmask)
tl.store(out_ptr1 + (x0), tmp28, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ki/ckiemul6tf32aklelekgb7nubdcitlamu5mdffv4u2b2odledpcc.py
# Topologically Sorted Source Nodes: [hig_vectors_1, hig_vectors_2], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# hig_vectors_1 => add
# hig_vectors_2 => add_1, add_2, mul, mul_1, rsqrt, sub_1
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %view_17), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-08), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %getitem_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_7), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_8), kwargs = {})
triton_poi_fused_add_native_layer_norm_4 = async_compile.triton('triton_poi_fused_add_native_layer_norm_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_4(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-08
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + (x2), tmp13, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [hig_query], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf0)
del primals_3
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [low_key], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [low_value], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(buf0, buf3, 16, 4, grid=grid(16, 4), stream=stream0)
buf4 = reinterpret_tensor(buf0, (4, 4, 1, 4), (16, 4, 4, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(buf1, buf4, 16, 4, grid=grid(16, 4), stream=stream0)
buf5 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf4, (16, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf5, buf6, 256, grid=grid(256), stream=stream0)
buf7 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf5 # reuse
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf6, buf7, 256, grid=grid(256), stream=stream0)
del buf6
buf8 = reinterpret_tensor(buf1, (4, 4, 4, 1), (16, 4, 1, 1), 0); del buf1 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(buf2, buf8, 16, 4, grid=grid(16, 4), stream=stream0)
buf9 = reinterpret_tensor(buf2, (16, 4, 1), (4, 1, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf7, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf8, (16, 4, 1), (4, 1, 0), 0), out=buf9)
buf10 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(buf9, buf10, 16, 4, grid=grid(16, 4), stream=stream0)
buf11 = reinterpret_tensor(buf9, (16, 4), (4, 1), 0); del buf9 # reuse
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf10, (16, 4), (4, 1), 0), reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf11)
buf12 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf13 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
# Topologically Sorted Source Nodes: [hig_vectors_1, hig_vectors_2], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_3.run(primals_2, buf11, buf12, buf13, 16, grid=grid(16), stream=stream0)
buf14 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [hig_vectors_1, hig_vectors_2], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_4.run(primals_2, buf11, buf12, buf13, primals_7, primals_8, buf14, 64, grid=grid(64), stream=stream0)
del buf12
del buf13
del primals_8
return (buf14, buf7, primals_2, primals_7, reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), buf7, reinterpret_tensor(buf10, (16, 4), (4, 1), 0), buf11, primals_6, reinterpret_tensor(buf8, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf3, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super(ScaledDotProductAttention, self).__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(MultiHeadAttention, self).__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class ComprehensionLayer_step3(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(ComprehensionLayer_step3, self).__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, low_vectors, hig_vectors):
b = low_vectors.size()[0]
low_num, hig_num = low_vectors.size()[1], hig_vectors.size()[1]
hig_residual = hig_vectors
hig_query = self.Wq(hig_vectors)
low_key = self.Wk(low_vectors)
low_value = self.Wv(low_vectors)
hig_query = hig_query.reshape(b, hig_num, self.n_head, self.
reduced_dim // self.n_head)
low_key = low_key.reshape(b, low_num, self.n_head, self.reduced_dim //
self.n_head)
low_value = low_value.reshape(b, low_num, self.n_head, self.
reduced_dim // self.n_head)
hig_query = hig_query.transpose(1, 2)
low_key = low_key.transpose(1, 2)
low_value = low_value.transpose(1, 2)
hig_query, hig_low_weights = self.inner_attention(hig_query,
low_key, low_value)
hig_query = hig_query.transpose(1, 2).reshape(b, hig_num, self.
reduced_dim)
hig_vectors = self.dropout(self.Wo(hig_query))
hig_vectors = hig_residual + hig_vectors
hig_vectors = self.hig_ln(hig_vectors)
return hig_vectors, hig_low_weights
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'embedding_dim': 4, 'reduced_dim': 4, 'n_head': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 16 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp3 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp14 * tmp1
tmp16 = tl_math.exp(tmp15)
tl.store(out_ptr0 + x2, tmp16, xmask)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_3(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp12 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + x0, tmp16, xmask)
tl.store(out_ptr1 + x0, tmp28, xmask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_4(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-08
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + x2, tmp13, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf0)
del primals_3
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(16, 4)](buf0, buf3, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf4 = reinterpret_tensor(buf0, (4, 4, 1, 4), (16, 4, 4, 1), 0)
del buf0
triton_poi_fused_clone_0[grid(16, 4)](buf1, buf4, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf5 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf4, (16, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_1[grid(256)](buf5, buf6, 256, XBLOCK=256,
num_warps=4, num_stages=1)
buf7 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf5
triton_poi_fused__softmax_2[grid(256)](buf6, buf7, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del buf6
buf8 = reinterpret_tensor(buf1, (4, 4, 4, 1), (16, 4, 1, 1), 0)
del buf1
triton_poi_fused_clone_0[grid(16, 4)](buf2, buf8, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf9 = reinterpret_tensor(buf2, (16, 4, 1), (4, 1, 1), 0)
del buf2
extern_kernels.bmm(reinterpret_tensor(buf7, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf8, (16, 4, 1), (4, 1, 0), 0), out=buf9)
buf10 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_clone_0[grid(16, 4)](buf9, buf10, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf11 = reinterpret_tensor(buf9, (16, 4), (4, 1), 0)
del buf9
extern_kernels.mm(reinterpret_tensor(buf10, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf11)
buf12 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf13 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
triton_poi_fused_add_native_layer_norm_3[grid(16)](primals_2, buf11,
buf12, buf13, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf14 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_add_native_layer_norm_4[grid(64)](primals_2, buf11,
buf12, buf13, primals_7, primals_8, buf14, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf12
del buf13
del primals_8
return buf14, buf7, primals_2, primals_7, reinterpret_tensor(primals_1,
(16, 4), (4, 1), 0), buf7, reinterpret_tensor(buf10, (16, 4), (4, 1), 0
), buf11, primals_6, reinterpret_tensor(buf8, (16, 1, 4), (4, 1, 1), 0
), reinterpret_tensor(buf3, (16, 1, 4), (4, 1, 1), 0
), reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 4), 0)
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super(ScaledDotProductAttention, self).__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(MultiHeadAttention, self).__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class ComprehensionLayer_step3New(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(ComprehensionLayer_step3New, self).__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, input_0, input_1):
primals_3 = self.Wq.weight
primals_4 = self.Wk.weight
primals_5 = self.Wv.weight
primals_6 = self.Wo.weight
primals_7 = self.hig_ln.weight
primals_8 = self.hig_ln.bias
primals_1 = input_0
primals_2 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8])
return output[0], output[1]
| luyu-fan/LRCM | ComprehensionLayer_step3 | false | 7,145 | [
"MIT"
] | 1 | 6b0e4d7998bc4969afa764eb753077e3f858f1ba | https://github.com/luyu-fan/LRCM/tree/6b0e4d7998bc4969afa764eb753077e3f858f1ba | import math
import torch
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super().__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super().__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class Model(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super().__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, low_vectors, hig_vectors):
b = low_vectors.size()[0]
low_num, hig_num = low_vectors.size()[1], hig_vectors.size()[1]
hig_residual = hig_vectors
hig_query = self.Wq(hig_vectors)
low_key = self.Wk(low_vectors)
low_value = self.Wv(low_vectors)
hig_query = hig_query.reshape(b, hig_num, self.n_head, self.
reduced_dim // self.n_head)
low_key = low_key.reshape(b, low_num, self.n_head, self.reduced_dim //
self.n_head)
low_value = low_value.reshape(b, low_num, self.n_head, self.
reduced_dim // self.n_head)
hig_query = hig_query.transpose(1, 2)
low_key = low_key.transpose(1, 2)
low_value = low_value.transpose(1, 2)
hig_query, hig_low_weights = self.inner_attention(hig_query,
low_key, low_value)
hig_query = hig_query.transpose(1, 2).reshape(b, hig_num, self.
reduced_dim)
hig_vectors = self.dropout(self.Wo(hig_query))
hig_vectors = hig_residual + hig_vectors
hig_vectors = self.hig_ln(hig_vectors)
return hig_vectors, hig_low_weights
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4])]
def get_init_inputs():
return [4, 4, 4]
|
Tanh | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/dn/cdnhr6ixjduuhci57kobqjnehjrl22mcyjqzuuhvtxxshy437diy.py
# Topologically Sorted Source Nodes: [tanh], Original ATen: [aten.tanh]
# Source node to ATen node mapping:
# tanh => tanh
# Graph fragment:
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%arg0_1,), kwargs = {})
triton_poi_fused_tanh_0 = async_compile.triton('triton_poi_fused_tanh_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_tanh_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_tanh_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = libdevice.tanh(tmp0)
tl.store(out_ptr0 + (x0), tmp1, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [tanh], Original ATen: [aten.tanh]
stream0 = get_raw_stream(0)
triton_poi_fused_tanh_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.utils.data
import torch.nn as nn
import torch._utils
from torch import optim as optim
import torch.nn.parallel
class Tanh(nn.Module):
def __init__(self, inplace=False):
super(Tanh, self).__init__()
self.inplace = inplace
def forward(self, x):
return x.tanh_() if self.inplace else x.tanh()
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.utils.data
import torch.nn as nn
import torch._utils
from torch import optim as optim
import torch.nn.parallel
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_tanh_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = libdevice.tanh(tmp0)
tl.store(out_ptr0 + x0, tmp1, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_tanh_0[grid(256)](arg0_1, buf0, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del arg0_1
return buf0,
class TanhNew(nn.Module):
def __init__(self, inplace=False):
super(TanhNew, self).__init__()
self.inplace = inplace
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| lovelinability/pytorch_image_models | Tanh | false | 7,146 | [
"Apache-2.0"
] | 1 | 7c54200f3de7611ab1222a37088eb7f66ae2858f | https://github.com/lovelinability/pytorch_image_models/tree/7c54200f3de7611ab1222a37088eb7f66ae2858f | import torch
import torch.utils.data
import torch.nn as nn
import torch._utils
from torch import optim as optim
import torch.nn.parallel
class Model(nn.Module):
def __init__(self, inplace=False):
super().__init__()
self.inplace = inplace
def forward(self, x):
return x.tanh_() if self.inplace else x.tanh()
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
LayerNormalization | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/dx/cdxvfu4asrdqxv3hebtdi7k2mpsqdnfv2swotbko2wrdw43mle4b.py
# Topologically Sorted Source Nodes: [sub, add, ln_out, mul, ln_out_1], Original ATen: [aten.sub, aten.add, aten.div, aten.mul]
# Source node to ATen node mapping:
# add => add
# ln_out => div
# ln_out_1 => add_1
# mul => mul
# sub => sub
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %expand), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%expand_1, 0.001), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %add), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%expand_2, %div), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %expand_3), kwargs = {})
triton_poi_fused_add_div_mul_sub_0 = async_compile.triton('triton_poi_fused_add_div_mul_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mul_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 7, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mul_sub_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp2 = tl.load(in_ptr1 + (4*x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr1 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp30 = tl.load(in_ptr2 + (x0), xmask, eviction_policy='evict_last')
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp8 = tmp6 + tmp7
tmp9 = 4.0
tmp10 = tmp8 / tmp9
tmp11 = tmp1 - tmp10
tmp12 = tmp2 - tmp10
tmp13 = tmp12 * tmp12
tmp14 = tmp3 - tmp10
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp10
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp7 - tmp10
tmp21 = tmp20 * tmp20
tmp22 = tmp19 + tmp21
tmp23 = 3.0
tmp24 = tmp22 / tmp23
tmp25 = libdevice.sqrt(tmp24)
tmp26 = 0.001
tmp27 = tmp25 + tmp26
tmp28 = tmp11 / tmp27
tmp29 = tmp0 * tmp28
tmp31 = tmp29 + tmp30
tl.store(out_ptr0 + (x2), tmp31, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [sub, add, ln_out, mul, ln_out_1], Original ATen: [aten.sub, aten.add, aten.div, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_mul_sub_0.run(primals_2, primals_1, primals_3, buf0, 256, grid=grid(256), stream=stream0)
del primals_2
del primals_3
return (buf0, primals_1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class LayerNormalization(nn.Module):
def __init__(self, d_hid, eps=0.001):
super(LayerNormalization, self).__init__()
self.gamma = nn.Parameter(torch.ones(d_hid), requires_grad=True)
self.beta = nn.Parameter(torch.zeros(d_hid), requires_grad=True)
self.eps = eps
def forward(self, z):
mean = z.mean(dim=-1, keepdim=True)
std = z.std(dim=-1, keepdim=True)
ln_out = (z - mean.expand_as(z)) / (std.expand_as(z) + self.eps)
ln_out = self.gamma.expand_as(ln_out) * ln_out + self.beta.expand_as(
ln_out)
return ln_out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'d_hid': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_mul_sub_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp2 = tl.load(in_ptr1 + 4 * x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr1 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp30 = tl.load(in_ptr2 + x0, xmask, eviction_policy='evict_last')
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp8 = tmp6 + tmp7
tmp9 = 4.0
tmp10 = tmp8 / tmp9
tmp11 = tmp1 - tmp10
tmp12 = tmp2 - tmp10
tmp13 = tmp12 * tmp12
tmp14 = tmp3 - tmp10
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp10
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp7 - tmp10
tmp21 = tmp20 * tmp20
tmp22 = tmp19 + tmp21
tmp23 = 3.0
tmp24 = tmp22 / tmp23
tmp25 = libdevice.sqrt(tmp24)
tmp26 = 0.001
tmp27 = tmp25 + tmp26
tmp28 = tmp11 / tmp27
tmp29 = tmp0 * tmp28
tmp31 = tmp29 + tmp30
tl.store(out_ptr0 + x2, tmp31, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_mul_sub_0[grid(256)](primals_2, primals_1,
primals_3, buf0, 256, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
del primals_3
return buf0, primals_1
class LayerNormalizationNew(nn.Module):
def __init__(self, d_hid, eps=0.001):
super(LayerNormalizationNew, self).__init__()
self.gamma = nn.Parameter(torch.ones(d_hid), requires_grad=True)
self.beta = nn.Parameter(torch.zeros(d_hid), requires_grad=True)
self.eps = eps
def forward(self, input_0):
primals_2 = self.gamma
primals_3 = self.beta
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| lz-chen/ner-bert | LayerNormalization | false | 7,147 | [
"MIT"
] | 1 | 86e73c1e7124a4fb6ee65d42b72333573841fe5b | https://github.com/lz-chen/ner-bert/tree/86e73c1e7124a4fb6ee65d42b72333573841fe5b | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self, d_hid, eps=0.001):
super().__init__()
self.gamma = nn.Parameter(torch.ones(d_hid), requires_grad=True)
self.beta = nn.Parameter(torch.zeros(d_hid), requires_grad=True)
self.eps = eps
def forward(self, z):
mean = z.mean(dim=-1, keepdim=True)
std = z.std(dim=-1, keepdim=True)
ln_out = (z - mean.expand_as(z)) / (std.expand_as(z) + self.eps)
ln_out = self.gamma.expand_as(ln_out) * ln_out + self.beta.expand_as(
ln_out)
return ln_out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
Conv2dUntiedBias | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/et/cet6mje25ge6ywvlm34oad3xuwmemw7aw7m3cvpkzt6yzrofl26r.py
# Topologically Sorted Source Nodes: [repeat, output_1], Original ATen: [aten.repeat, aten.add]
# Source node to ATen node mapping:
# output_1 => add
# repeat => repeat
# Graph fragment:
# %repeat : [num_users=1] = call_function[target=torch.ops.aten.repeat.default](args = (%unsqueeze, [4, 1, 1, 1]), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convolution, %repeat), kwargs = {})
triton_poi_fused_add_repeat_0 = async_compile.triton('triton_poi_fused_add_repeat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_repeat_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_repeat_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, 1, 1), (1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_2, primals_1, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 1, 1), (4, 1, 1, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [repeat, output_1], Original ATen: [aten.repeat, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_repeat_0.run(buf1, primals_3, 16, grid=grid(16), stream=stream0)
del primals_3
return (buf1, primals_1, primals_2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 1, 1), (1, 1, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn.modules.utils import _pair
class Conv2dUntiedBias(nn.Module):
def __init__(self, in_channels, out_channels, kernel_size, input_len,
stride=1, padding=0, dilation=1, groups=1):
super(Conv2dUntiedBias, self).__init__()
kernel_size = _pair(kernel_size)
stride = _pair(stride)
padding = _pair(padding)
dilation = _pair(dilation)
if in_channels % groups != 0:
raise ValueError('in_channels must be divisible by groups')
if out_channels % groups != 0:
raise ValueError('out_channels must be divisible by groups')
self.in_channels = in_channels
self.out_channels = out_channels
self.kernel_size = kernel_size
self.stride = stride
self.padding = padding
self.dilation = dilation
self.groups = groups
self.weight = nn.Parameter(torch.Tensor(out_channels, in_channels //
groups, *kernel_size))
height = 1
width = self.calc_output_width(input_len, kernel_size)
self.bias = nn.Parameter(torch.Tensor(out_channels, height, width))
self.reset_parameters()
def calc_output_width(self, input_length, kernel_size, stride=1):
return (input_length - kernel_size[-1] + stride) // stride
def reset_parameters(self):
n = self.in_channels
for k in self.kernel_size:
n *= k
stdv = 1.0 / math.sqrt(n)
self.weight.data.uniform_(-stdv, stdv)
self.bias.data.uniform_(-stdv, stdv)
def forward(self, input):
output = F.conv2d(input, self.weight, None, self.stride, self.
padding, self.dilation, self.groups)
output += self.bias.unsqueeze(0).repeat(input.size(0), 1, 1, 1)
return output
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'kernel_size': 4,
'input_len': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import math
import torch.nn as nn
from torch.nn.modules.utils import _pair
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
@triton.jit
def triton_poi_fused_add_repeat_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, 1, 1), (1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_2, primals_1, stride=(1,
1), padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 1, 1), (4, 1, 1, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_add_repeat_0[grid(16)](buf1, primals_3, 16, XBLOCK
=16, num_warps=1, num_stages=1)
del primals_3
return buf1, primals_1, primals_2
class Conv2dUntiedBiasNew(nn.Module):
def __init__(self, in_channels, out_channels, kernel_size, input_len,
stride=1, padding=0, dilation=1, groups=1):
super(Conv2dUntiedBiasNew, self).__init__()
kernel_size = _pair(kernel_size)
stride = _pair(stride)
padding = _pair(padding)
dilation = _pair(dilation)
if in_channels % groups != 0:
raise ValueError('in_channels must be divisible by groups')
if out_channels % groups != 0:
raise ValueError('out_channels must be divisible by groups')
self.in_channels = in_channels
self.out_channels = out_channels
self.kernel_size = kernel_size
self.stride = stride
self.padding = padding
self.dilation = dilation
self.groups = groups
self.weight = nn.Parameter(torch.Tensor(out_channels, in_channels //
groups, *kernel_size))
height = 1
width = self.calc_output_width(input_len, kernel_size)
self.bias = nn.Parameter(torch.Tensor(out_channels, height, width))
self.reset_parameters()
def calc_output_width(self, input_length, kernel_size, stride=1):
return (input_length - kernel_size[-1] + stride) // stride
def reset_parameters(self):
n = self.in_channels
for k in self.kernel_size:
n *= k
stdv = 1.0 / math.sqrt(n)
self.weight.data.uniform_(-stdv, stdv)
self.bias.data.uniform_(-stdv, stdv)
def forward(self, input_0):
primals_1 = self.weight
primals_3 = self.bias
primals_2 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| lzamparo/SeqDemote | Conv2dUntiedBias | false | 7,148 | [
"MIT"
] | 1 | 3eaf18e88c9dc6a3d1a69444ecdba9f9b5d9682a | https://github.com/lzamparo/SeqDemote/tree/3eaf18e88c9dc6a3d1a69444ecdba9f9b5d9682a | import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn.modules.utils import _pair
class Model(nn.Module):
def __init__(self, in_channels, out_channels, kernel_size, input_len,
stride=1, padding=0, dilation=1, groups=1):
super().__init__()
kernel_size = _pair(kernel_size)
stride = _pair(stride)
padding = _pair(padding)
dilation = _pair(dilation)
if in_channels % groups != 0:
raise ValueError('in_channels must be divisible by groups')
if out_channels % groups != 0:
raise ValueError('out_channels must be divisible by groups')
self.in_channels = in_channels
self.out_channels = out_channels
self.kernel_size = kernel_size
self.stride = stride
self.padding = padding
self.dilation = dilation
self.groups = groups
self.weight = nn.Parameter(torch.Tensor(out_channels, in_channels //
groups, *kernel_size))
height = 1
width = self.calc_output_width(input_len, kernel_size)
self.bias = nn.Parameter(torch.Tensor(out_channels, height, width))
self.reset_parameters()
def calc_output_width(self, input_length, kernel_size, stride=1):
return (input_length - kernel_size[-1] + stride) // stride
def reset_parameters(self):
n = self.in_channels
for k in self.kernel_size:
n *= k
stdv = 1.0 / math.sqrt(n)
self.weight.data.uniform_(-stdv, stdv)
self.bias.data.uniform_(-stdv, stdv)
def forward(self, input):
output = F.conv2d(input, self.weight, None, self.stride, self.
padding, self.dilation, self.groups)
output += self.bias.unsqueeze(0).repeat(input.size(0), 1, 1, 1)
return output
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'kernel_size': 4,
'input_len': 4}]
|
FocalLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/tq/ctqbufosv5x2yg6jklh4fqpslk7be7rbqcnu7eipvb6b4pemujk4.py
# Topologically Sorted Source Nodes: [binary_cross_entropy_with_logits, mul_2, sub_2, mul_3, alpha_t, p, mul, sub, sub_1, mul_1, p_t, sub_3, pow_1, w], Original ATen: [aten.binary_cross_entropy_with_logits, aten.mul, aten.rsub, aten.add, aten.sigmoid, aten.pow]
# Source node to ATen node mapping:
# alpha_t => add_1
# binary_cross_entropy_with_logits => abs_1, exp, full_default, log1p, minimum, mul_5, mul_6, neg, sub_4, sub_5, sub_6, sum_1
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# mul_3 => mul_3
# p => sigmoid
# p_t => add
# pow_1 => pow_1
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# sub_3 => sub_3
# w => mul_4
# Graph fragment:
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %arg1_1), kwargs = {})
# %mul_5 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_4, %arg0_1), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %minimum : [num_users=1] = call_function[target=torch.ops.aten.minimum.default](args = (%full_default, %arg0_1), kwargs = {})
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%arg0_1,), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%abs_1,), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%neg,), kwargs = {})
# %log1p : [num_users=1] = call_function[target=torch.ops.aten.log1p.default](args = (%exp,), kwargs = {})
# %sub_5 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%minimum, %log1p), kwargs = {})
# %sub_6 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_5, %sub_5), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg1_1, 0.7), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %arg1_1), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_2, 0.30000000000000004), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_2, %mul_3), kwargs = {})
# %sigmoid : [num_users=2] = call_function[target=torch.ops.aten.sigmoid.default](args = (%arg0_1,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid, %arg1_1), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %sigmoid), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %arg1_1), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %sub_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %add), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_3, 1.5), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_1, %pow_1), kwargs = {})
# %mul_6 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_6, %mul_4), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%mul_6,), kwargs = {})
triton_per_fused_add_binary_cross_entropy_with_logits_mul_pow_rsub_sigmoid_0 = async_compile.triton('triton_per_fused_add_binary_cross_entropy_with_logits_mul_pow_rsub_sigmoid_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_binary_cross_entropy_with_logits_mul_pow_rsub_sigmoid_0', 'mutated_arg_names': [], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_binary_cross_entropy_with_logits_mul_pow_rsub_sigmoid_0(in_ptr0, in_ptr1, out_ptr0, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp3 = tl.load(in_ptr1 + (r0), None)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp2 * tmp3
tmp5 = 0.0
tmp6 = triton_helpers.minimum(tmp5, tmp3)
tmp7 = tl_math.abs(tmp3)
tmp8 = -tmp7
tmp9 = tl_math.exp(tmp8)
tmp10 = libdevice.log1p(tmp9)
tmp11 = tmp6 - tmp10
tmp12 = tmp4 - tmp11
tmp13 = 0.7
tmp14 = tmp0 * tmp13
tmp15 = 0.30000000000000004
tmp16 = tmp2 * tmp15
tmp17 = tmp14 + tmp16
tmp18 = tl.sigmoid(tmp3)
tmp19 = tmp18 * tmp0
tmp20 = tmp1 - tmp18
tmp21 = tmp20 * tmp2
tmp22 = tmp19 + tmp21
tmp23 = tmp1 - tmp22
tmp24 = 1.5
tmp25 = libdevice.pow(tmp23, tmp24)
tmp26 = tmp17 * tmp25
tmp27 = tmp12 * tmp26
tmp28 = tl.broadcast_to(tmp27, [RBLOCK])
tmp30 = triton_helpers.promote_to_tensor(tl.sum(tmp28, 0))
tl.store(out_ptr0 + (tl.full([1], 0, tl.int32)), tmp30, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
# Topologically Sorted Source Nodes: [binary_cross_entropy_with_logits, mul_2, sub_2, mul_3, alpha_t, p, mul, sub, sub_1, mul_1, p_t, sub_3, pow_1, w], Original ATen: [aten.binary_cross_entropy_with_logits, aten.mul, aten.rsub, aten.add, aten.sigmoid, aten.pow]
stream0 = get_raw_stream(0)
triton_per_fused_add_binary_cross_entropy_with_logits_mul_pow_rsub_sigmoid_0.run(arg1_1, arg0_1, buf0, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class FocalLoss(nn.Module):
def __init__(self, reduce=True, gamma=1.5, alpha=0.7):
super(FocalLoss, self).__init__()
self.reduce = reduce
self.gamma = gamma
self.alpha = alpha
def _get_weights(self, x, t):
"""
Helper to get the weights for focal loss calculation
"""
p = nn.functional.sigmoid(x)
p_t = p * t + (1 - p) * (1 - t)
alpha_t = self.alpha * t + (1 - self.alpha) * (1 - t)
w = alpha_t * (1 - p_t).pow(self.gamma)
return w
def focal_loss(self, x, t):
"""
Focal Loss cf. arXiv:1708.02002
Args:
x: (tensor) output from last layer of network
t: (tensor) targets in [0,1]
alpha: (float) class imbalance correction weight \\in (0,1)
gamma: (float) amplification factor for uncertain classification
Return:
(tensor) focal loss.
"""
weights = self._get_weights(x, t)
return nn.functional.binary_cross_entropy_with_logits(x, t, weights,
size_average=False, reduce=self.reduce)
def forward(self, input, target):
return self.focal_loss(input, target)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_add_binary_cross_entropy_with_logits_mul_pow_rsub_sigmoid_0(
in_ptr0, in_ptr1, out_ptr0, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp3 = tl.load(in_ptr1 + r0, None)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp2 * tmp3
tmp5 = 0.0
tmp6 = triton_helpers.minimum(tmp5, tmp3)
tmp7 = tl_math.abs(tmp3)
tmp8 = -tmp7
tmp9 = tl_math.exp(tmp8)
tmp10 = libdevice.log1p(tmp9)
tmp11 = tmp6 - tmp10
tmp12 = tmp4 - tmp11
tmp13 = 0.7
tmp14 = tmp0 * tmp13
tmp15 = 0.30000000000000004
tmp16 = tmp2 * tmp15
tmp17 = tmp14 + tmp16
tmp18 = tl.sigmoid(tmp3)
tmp19 = tmp18 * tmp0
tmp20 = tmp1 - tmp18
tmp21 = tmp20 * tmp2
tmp22 = tmp19 + tmp21
tmp23 = tmp1 - tmp22
tmp24 = 1.5
tmp25 = libdevice.pow(tmp23, tmp24)
tmp26 = tmp17 * tmp25
tmp27 = tmp12 * tmp26
tmp28 = tl.broadcast_to(tmp27, [RBLOCK])
tmp30 = triton_helpers.promote_to_tensor(tl.sum(tmp28, 0))
tl.store(out_ptr0 + tl.full([1], 0, tl.int32), tmp30, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
get_raw_stream(0)
triton_per_fused_add_binary_cross_entropy_with_logits_mul_pow_rsub_sigmoid_0[
grid(1)](arg1_1, arg0_1, buf0, 1, 256, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf0,
class FocalLossNew(nn.Module):
def __init__(self, reduce=True, gamma=1.5, alpha=0.7):
super(FocalLossNew, self).__init__()
self.reduce = reduce
self.gamma = gamma
self.alpha = alpha
def _get_weights(self, x, t):
"""
Helper to get the weights for focal loss calculation
"""
p = nn.functional.sigmoid(x)
p_t = p * t + (1 - p) * (1 - t)
alpha_t = self.alpha * t + (1 - self.alpha) * (1 - t)
w = alpha_t * (1 - p_t).pow(self.gamma)
return w
def focal_loss(self, x, t):
"""
Focal Loss cf. arXiv:1708.02002
Args:
x: (tensor) output from last layer of network
t: (tensor) targets in [0,1]
alpha: (float) class imbalance correction weight \\in (0,1)
gamma: (float) amplification factor for uncertain classification
Return:
(tensor) focal loss.
"""
weights = self._get_weights(x, t)
return nn.functional.binary_cross_entropy_with_logits(x, t, weights,
size_average=False, reduce=self.reduce)
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| lzamparo/SeqDemote | FocalLoss | false | 7,149 | [
"MIT"
] | 1 | 3eaf18e88c9dc6a3d1a69444ecdba9f9b5d9682a | https://github.com/lzamparo/SeqDemote/tree/3eaf18e88c9dc6a3d1a69444ecdba9f9b5d9682a | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self, reduce=True, gamma=1.5, alpha=0.7):
super().__init__()
self.reduce = reduce
self.gamma = gamma
self.alpha = alpha
def _get_weights(self, x, t):
"""
Helper to get the weights for focal loss calculation
"""
p = nn.functional.sigmoid(x)
p_t = p * t + (1 - p) * (1 - t)
alpha_t = self.alpha * t + (1 - self.alpha) * (1 - t)
w = alpha_t * (1 - p_t).pow(self.gamma)
return w
def focal_loss(self, x, t):
"""
Focal Loss cf. arXiv:1708.02002
Args:
x: (tensor) output from last layer of network
t: (tensor) targets in [0,1]
alpha: (float) class imbalance correction weight \\in (0,1)
gamma: (float) amplification factor for uncertain classification
Return:
(tensor) focal loss.
"""
weights = self._get_weights(x, t)
return nn.functional.binary_cross_entropy_with_logits(x, t, weights,
size_average=False, reduce=self.reduce)
def forward(self, input, target):
return self.focal_loss(input, target)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
SoftTargetCrossEntropy | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/nr/cnrkptzsuv7qm3ss6i6xgoxkou23z76h2vmwqkwz2zkgpdbxhedc.py
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# log_softmax => amax, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%arg1_1, [-1], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg1_1, %amax), kwargs = {})
triton_poi_fused__log_softmax_0 = async_compile.triton('triton_poi_fused__log_softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/7e/c7eos52pj4trwrwevfplxacwgfirtfuiycj3hrmzuhm4mq7vguud.py
# Topologically Sorted Source Nodes: [neg, log_softmax, mul, loss, mean], Original ATen: [aten.neg, aten._log_softmax, aten.mul, aten.sum, aten.mean]
# Source node to ATen node mapping:
# log_softmax => exp, log, sub_1, sum_1
# loss => sum_2
# mean => mean
# mul => mul
# neg => neg
# Graph fragment:
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%arg0_1,), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %log), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%neg, %sub_1), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [-1]), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sum_2,), kwargs = {})
triton_per_fused__log_softmax_mean_mul_neg_sum_1 = async_compile.triton('triton_per_fused__log_softmax_mean_mul_neg_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused__log_softmax_mean_mul_neg_sum_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused__log_softmax_mean_mul_neg_sum_1(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (4*r0), None, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr1 + (4*r0), None, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*r0)), None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (2 + (4*r0)), None, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr1 + (3 + (4*r0)), None, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr0 + (1 + (4*r0)), None, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (2 + (4*r0)), None, eviction_policy='evict_last')
tmp26 = tl.load(in_ptr0 + (3 + (4*r0)), None, eviction_policy='evict_last')
tmp1 = -tmp0
tmp3 = tl_math.exp(tmp2)
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp3 + tmp5
tmp8 = tl_math.exp(tmp7)
tmp9 = tmp6 + tmp8
tmp11 = tl_math.exp(tmp10)
tmp12 = tmp9 + tmp11
tmp13 = tl_math.log(tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp1 * tmp14
tmp17 = -tmp16
tmp18 = tmp4 - tmp13
tmp19 = tmp17 * tmp18
tmp20 = tmp15 + tmp19
tmp22 = -tmp21
tmp23 = tmp7 - tmp13
tmp24 = tmp22 * tmp23
tmp25 = tmp20 + tmp24
tmp27 = -tmp26
tmp28 = tmp10 - tmp13
tmp29 = tmp27 * tmp28
tmp30 = tmp25 + tmp29
tmp31 = tl.broadcast_to(tmp30, [XBLOCK, RBLOCK])
tmp33 = tl.sum(tmp31, 1)[:, None]
tmp34 = 64.0
tmp35 = tmp33 / tmp34
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp35, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__log_softmax_0.run(arg1_1, buf0, 256, grid=grid(256), stream=stream0)
del arg1_1
buf2 = empty_strided_cuda((), (), torch.float32)
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [neg, log_softmax, mul, loss, mean], Original ATen: [aten.neg, aten._log_softmax, aten.mul, aten.sum, aten.mean]
triton_per_fused__log_softmax_mean_mul_neg_sum_1.run(buf3, arg0_1, buf0, 1, 64, grid=grid(1), stream=stream0)
del arg0_1
del buf0
return (buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.utils.data
import torchvision.transforms.functional as F
import torch.nn as nn
import torch.nn.functional as F
import torch._utils
from torch import optim as optim
import torch.nn.parallel
class SoftTargetCrossEntropy(nn.Module):
def __init__(self):
super(SoftTargetCrossEntropy, self).__init__()
def forward(self, x, target):
loss = torch.sum(-target * F.log_softmax(x, dim=-1), dim=-1)
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.utils.data
import torch.nn as nn
import torch._utils
from torch import optim as optim
import torch.nn.parallel
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused__log_softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_per_fused__log_softmax_mean_mul_neg_sum_1(in_out_ptr0, in_ptr0,
in_ptr1, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + 4 * r0, None, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr1 + 4 * r0, None, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * r0), None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (2 + 4 * r0), None, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr1 + (3 + 4 * r0), None, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr0 + (1 + 4 * r0), None, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (2 + 4 * r0), None, eviction_policy='evict_last')
tmp26 = tl.load(in_ptr0 + (3 + 4 * r0), None, eviction_policy='evict_last')
tmp1 = -tmp0
tmp3 = tl_math.exp(tmp2)
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp3 + tmp5
tmp8 = tl_math.exp(tmp7)
tmp9 = tmp6 + tmp8
tmp11 = tl_math.exp(tmp10)
tmp12 = tmp9 + tmp11
tmp13 = tl_math.log(tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp1 * tmp14
tmp17 = -tmp16
tmp18 = tmp4 - tmp13
tmp19 = tmp17 * tmp18
tmp20 = tmp15 + tmp19
tmp22 = -tmp21
tmp23 = tmp7 - tmp13
tmp24 = tmp22 * tmp23
tmp25 = tmp20 + tmp24
tmp27 = -tmp26
tmp28 = tmp10 - tmp13
tmp29 = tmp27 * tmp28
tmp30 = tmp25 + tmp29
tmp31 = tl.broadcast_to(tmp30, [XBLOCK, RBLOCK])
tmp33 = tl.sum(tmp31, 1)[:, None]
tmp34 = 64.0
tmp35 = tmp33 / tmp34
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp35, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__log_softmax_0[grid(256)](arg1_1, buf0, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del arg1_1
buf2 = empty_strided_cuda((), (), torch.float32)
buf3 = buf2
del buf2
triton_per_fused__log_softmax_mean_mul_neg_sum_1[grid(1)](buf3,
arg0_1, buf0, 1, 64, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
del buf0
return buf3,
class SoftTargetCrossEntropyNew(nn.Module):
def __init__(self):
super(SoftTargetCrossEntropyNew, self).__init__()
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| lovelinability/pytorch_image_models | SoftTargetCrossEntropy | false | 7,150 | [
"Apache-2.0"
] | 1 | 7c54200f3de7611ab1222a37088eb7f66ae2858f | https://github.com/lovelinability/pytorch_image_models/tree/7c54200f3de7611ab1222a37088eb7f66ae2858f | import torch
import torch.utils.data
import torchvision.transforms.functional as F
import torch.nn as nn
import torch.nn.functional as F
import torch._utils
from torch import optim as optim
import torch.nn.parallel
class Model(nn.Module):
def __init__(self):
super().__init__()
def forward(self, x, target):
loss = torch.sum(-target * F.log_softmax(x, dim=-1), dim=-1)
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
SimpleConv | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/d2/cd2w2tjdjbgdeh65hso2rcz7l72h2h6i23qyutv2xumb52zz6lm5.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_0 = async_compile.triton('triton_poi_fused_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256, 32], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 150
xnumel = 25
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 3
y1 = (yindex // 3)
tmp0 = tl.load(in_ptr0 + (x2 + (25*y3)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (3*x2) + (75*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/r5/cr5let6yjntwkbf53jd4yeoeuy4icu7yjffx67o3n62pxqegfu3x.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_1 = async_compile.triton('triton_poi_fused_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 16384], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 12
xnumel = 9216
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 3
y1 = (yindex // 3)
tmp0 = tl.load(in_ptr0 + (x2 + (9216*y3)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (3*x2) + (27648*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/iy/ciycfbumlz4rjxsscjylf3pnlz3qughkqw7azbbgnqrujpqehgmi.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_2 = async_compile.triton('triton_poi_fused_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8192, 32], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 5000
xnumel = 25
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 50
y1 = (yindex // 50)
tmp0 = tl.load(in_ptr0 + (x2 + (25*y3)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (50*x2) + (1250*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/h2/ch2pot2ltn735r5nwss3elqbmedml6qb25b3cxvbbljtevwwf3w7.py
# Topologically Sorted Source Nodes: [conv2d, x], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d => convolution
# x => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_3 = async_compile.triton('triton_poi_fused_convolution_relu_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2097152],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1692800
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 50
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/fa/cfa5nl26eiorzxnqty6cf4z2hlhgp66sfgehgi3dwxwws5xri2pz.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# x_1 => getitem, getitem_1
# Graph fragment:
# %getitem : [num_users=2] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 0), kwargs = {})
# %getitem_1 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_4 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[524288],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i8', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_4(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 423200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 50
x1 = (xindex // 50) % 46
x2 = (xindex // 2300)
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (100*x1) + (9200*x2)), xmask)
tmp1 = tl.load(in_ptr0 + (50 + x0 + (100*x1) + (9200*x2)), xmask)
tmp3 = tl.load(in_ptr0 + (4600 + x0 + (100*x1) + (9200*x2)), xmask)
tmp5 = tl.load(in_ptr0 + (4650 + x0 + (100*x1) + (9200*x2)), xmask)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + (x3), tmp6, xmask)
tl.store(out_ptr1 + (x3), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/sk/cskatsridi3flhgsxedhur3sq32zkbjdjktoidxvtka7equyu4wb.py
# Topologically Sorted Source Nodes: [conv2d_1, x_2], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_1 => convolution_1
# x_2 => relu_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%getitem, %primals_4, %primals_5, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_1,), kwargs = {})
triton_poi_fused_convolution_relu_5 = async_compile.triton('triton_poi_fused_convolution_relu_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1048576],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_5', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_5(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 705600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 100
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/6z/c6zka5uja52wkvzy4tf572dgdlrjzureqfwhrf6ur5j2vk377n25.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# x_3 => _low_memory_max_pool2d_with_offsets_1, getitem_3
# Graph fragment:
# %_low_memory_max_pool2d_with_offsets_1 : [num_users=2] = call_function[target=torch.ops.prims._low_memory_max_pool2d_with_offsets.default](args = (%relu_1, [2, 2], [2, 2], [0, 0], [1, 1], False), kwargs = {})
# %getitem_3 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_1, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_6 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048, 128], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i8', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_6(in_ptr0, out_ptr0, out_ptr1, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 1764
xnumel = 100
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 21
y1 = (yindex // 21)
y5 = yindex
y4 = (yindex // 441)
y6 = yindex % 441
tmp0 = tl.load(in_ptr0 + (x2 + (200*y0) + (8400*y1)), xmask & ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (100 + x2 + (200*y0) + (8400*y1)), xmask & ymask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (4200 + x2 + (200*y0) + (8400*y1)), xmask & ymask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (4300 + x2 + (200*y0) + (8400*y1)), xmask & ymask, eviction_policy='evict_last')
tmp2 = tmp1 > tmp0
tmp3 = tl.full([1, 1], 1, tl.int8)
tmp4 = tl.full([1, 1], 0, tl.int8)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp6 = triton_helpers.maximum(tmp1, tmp0)
tmp8 = tmp7 > tmp6
tmp9 = tl.full([1, 1], 2, tl.int8)
tmp10 = tl.where(tmp8, tmp9, tmp5)
tmp11 = triton_helpers.maximum(tmp7, tmp6)
tmp13 = tmp12 > tmp11
tmp14 = tl.full([1, 1], 3, tl.int8)
tmp15 = tl.where(tmp13, tmp14, tmp10)
tmp16 = triton_helpers.maximum(tmp12, tmp11)
tl.store(out_ptr0 + (x2 + (100*y5)), tmp15, xmask & ymask)
tl.store(out_ptr1 + (y6 + (441*x2) + (44128*y4)), tmp16, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/5e/c5emgzg5fbprnf3wyj2jnx6u2wyjtrvngahoogdt67b7b6hvmp75.py
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# x_5 => relu_2
# Graph fragment:
# %add_tensor_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default_1, %primals_7), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor_1,), kwargs = {})
triton_poi_fused_relu_7 = async_compile.triton('triton_poi_fused_relu_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8192],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_7', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_7(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 6400
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 1600
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/zf/czf2s4ziy43zvvw4yd6flzvdpvo7xa6mkywfzqdujwrysbntx3ma.py
# Topologically Sorted Source Nodes: [o], Original ATen: [aten.sigmoid]
# Source node to ATen node mapping:
# o => sigmoid
# Graph fragment:
# %add_tensor : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default, %primals_9), kwargs = {})
# %sigmoid : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%add_tensor,), kwargs = {})
triton_poi_fused_sigmoid_8 = async_compile.triton('triton_poi_fused_sigmoid_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sigmoid_8', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sigmoid_8(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (50, 3, 5, 5), (75, 25, 5, 1))
assert_size_stride(primals_2, (50, ), (1, ))
assert_size_stride(primals_3, (4, 3, 96, 96), (27648, 9216, 96, 1))
assert_size_stride(primals_4, (100, 50, 5, 5), (1250, 25, 5, 1))
assert_size_stride(primals_5, (100, ), (1, ))
assert_size_stride(primals_6, (1600, 44100), (44100, 1))
assert_size_stride(primals_7, (1600, ), (1, ))
assert_size_stride(primals_8, (1, 1600), (1600, 1))
assert_size_stride(primals_9, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((50, 3, 5, 5), (75, 1, 15, 3), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
stream0 = get_raw_stream(0)
triton_poi_fused_0.run(primals_1, buf0, 150, 25, grid=grid(150, 25), stream=stream0)
del primals_1
buf1 = empty_strided_cuda((4, 3, 96, 96), (27648, 1, 288, 3), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_1.run(primals_3, buf1, 12, 9216, grid=grid(12, 9216), stream=stream0)
del primals_3
buf2 = empty_strided_cuda((100, 50, 5, 5), (1250, 1, 250, 50), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_2.run(primals_4, buf2, 5000, 25, grid=grid(5000, 25), stream=stream0)
del primals_4
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(buf1, buf0, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 50, 92, 92), (423200, 1, 4600, 50))
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [conv2d, x], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_3.run(buf4, primals_2, 1692800, grid=grid(1692800), stream=stream0)
del primals_2
buf5 = empty_strided_cuda((4, 50, 46, 46), (105800, 1, 2300, 50), torch.float32)
buf6 = empty_strided_cuda((4, 50, 46, 46), (105800, 1, 2300, 50), torch.int8)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_4.run(buf4, buf5, buf6, 423200, grid=grid(423200), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf7 = extern_kernels.convolution(buf5, buf2, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf7, (4, 100, 42, 42), (176400, 1, 4200, 100))
buf8 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [conv2d_1, x_2], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_5.run(buf8, primals_5, 705600, grid=grid(705600), stream=stream0)
del primals_5
buf9 = empty_strided_cuda((4, 100, 21, 21), (44100, 1, 2100, 100), torch.int8)
buf10 = empty_strided_cuda((4, 100, 21, 21), (44128, 441, 21, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_6.run(buf8, buf9, buf10, 1764, 100, grid=grid(1764, 100), stream=stream0)
buf11 = empty_strided_cuda((4, 1600), (1600, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf10, (4, 44100), (44128, 1), 0), reinterpret_tensor(primals_6, (44100, 1600), (1, 44100), 0), out=buf11)
buf12 = buf11; del buf11 # reuse
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.relu]
triton_poi_fused_relu_7.run(buf12, primals_7, 6400, grid=grid(6400), stream=stream0)
del primals_7
buf13 = empty_strided_cuda((4, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf12, reinterpret_tensor(primals_8, (1600, 1), (1, 1600), 0), out=buf13)
buf14 = buf13; del buf13 # reuse
# Topologically Sorted Source Nodes: [o], Original ATen: [aten.sigmoid]
triton_poi_fused_sigmoid_8.run(buf14, primals_9, 4, grid=grid(4), stream=stream0)
del primals_9
return (buf14, buf0, buf1, buf2, buf4, buf5, buf6, buf8, buf9, reinterpret_tensor(buf10, (4, 44100), (44128, 1), 0), buf12, buf14, primals_8, primals_6, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((50, 3, 5, 5), (75, 25, 5, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((50, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 3, 96, 96), (27648, 9216, 96, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((100, 50, 5, 5), (1250, 25, 5, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((100, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((1600, 44100), (44100, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((1600, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((1, 1600), (1600, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class SimpleConv(nn.Module):
def __init__(self):
super(SimpleConv, self).__init__()
self.conv1 = nn.Conv2d(3, 50, 5, 1)
self.conv2 = nn.Conv2d(50, 100, 5, 1)
self.fc1 = nn.Linear(21 * 21 * 100, 1600)
self.fc2 = nn.Linear(1600, 1)
def forward(self, x):
x = F.relu(self.conv1(x))
x = F.max_pool2d(x, 2, 2)
x = F.relu(self.conv2(x))
x = F.max_pool2d(x, 2, 2)
x = x.view(-1, 21 * 21 * 100)
x = F.relu(self.fc1(x))
x = self.fc2(x)
o = F.sigmoid(x)
return o
def get_inputs():
return [torch.rand([4, 3, 96, 96])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 150
xnumel = 25
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 3
y1 = yindex // 3
tmp0 = tl.load(in_ptr0 + (x2 + 25 * y3), xmask & ymask, eviction_policy
='evict_last')
tl.store(out_ptr0 + (y0 + 3 * x2 + 75 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 12
xnumel = 9216
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 3
y1 = yindex // 3
tmp0 = tl.load(in_ptr0 + (x2 + 9216 * y3), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + 3 * x2 + 27648 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 5000
xnumel = 25
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 50
y1 = yindex // 50
tmp0 = tl.load(in_ptr0 + (x2 + 25 * y3), xmask & ymask, eviction_policy
='evict_last')
tl.store(out_ptr0 + (y0 + 50 * x2 + 1250 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_convolution_relu_3(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 1692800
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 50
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_4(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 423200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 50
x1 = xindex // 50 % 46
x2 = xindex // 2300
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 100 * x1 + 9200 * x2), xmask)
tmp1 = tl.load(in_ptr0 + (50 + x0 + 100 * x1 + 9200 * x2), xmask)
tmp3 = tl.load(in_ptr0 + (4600 + x0 + 100 * x1 + 9200 * x2), xmask)
tmp5 = tl.load(in_ptr0 + (4650 + x0 + 100 * x1 + 9200 * x2), xmask)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + x3, tmp6, xmask)
tl.store(out_ptr1 + x3, tmp16, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_5(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 705600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 100
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_6(in_ptr0, out_ptr0, out_ptr1,
ynumel, xnumel, YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 1764
xnumel = 100
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 21
y1 = yindex // 21
y5 = yindex
y4 = yindex // 441
y6 = yindex % 441
tmp0 = tl.load(in_ptr0 + (x2 + 200 * y0 + 8400 * y1), xmask & ymask,
eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (100 + x2 + 200 * y0 + 8400 * y1), xmask &
ymask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (4200 + x2 + 200 * y0 + 8400 * y1), xmask &
ymask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (4300 + x2 + 200 * y0 + 8400 * y1), xmask &
ymask, eviction_policy='evict_last')
tmp2 = tmp1 > tmp0
tmp3 = tl.full([1, 1], 1, tl.int8)
tmp4 = tl.full([1, 1], 0, tl.int8)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp6 = triton_helpers.maximum(tmp1, tmp0)
tmp8 = tmp7 > tmp6
tmp9 = tl.full([1, 1], 2, tl.int8)
tmp10 = tl.where(tmp8, tmp9, tmp5)
tmp11 = triton_helpers.maximum(tmp7, tmp6)
tmp13 = tmp12 > tmp11
tmp14 = tl.full([1, 1], 3, tl.int8)
tmp15 = tl.where(tmp13, tmp14, tmp10)
tmp16 = triton_helpers.maximum(tmp12, tmp11)
tl.store(out_ptr0 + (x2 + 100 * y5), tmp15, xmask & ymask)
tl.store(out_ptr1 + (y6 + 441 * x2 + 44128 * y4), tmp16, xmask & ymask)
@triton.jit
def triton_poi_fused_relu_7(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 6400
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 1600
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_sigmoid_8(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + x0, tmp4, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (50, 3, 5, 5), (75, 25, 5, 1))
assert_size_stride(primals_2, (50,), (1,))
assert_size_stride(primals_3, (4, 3, 96, 96), (27648, 9216, 96, 1))
assert_size_stride(primals_4, (100, 50, 5, 5), (1250, 25, 5, 1))
assert_size_stride(primals_5, (100,), (1,))
assert_size_stride(primals_6, (1600, 44100), (44100, 1))
assert_size_stride(primals_7, (1600,), (1,))
assert_size_stride(primals_8, (1, 1600), (1600, 1))
assert_size_stride(primals_9, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((50, 3, 5, 5), (75, 1, 15, 3), torch.float32)
get_raw_stream(0)
triton_poi_fused_0[grid(150, 25)](primals_1, buf0, 150, 25, XBLOCK=
32, YBLOCK=32, num_warps=4, num_stages=1)
del primals_1
buf1 = empty_strided_cuda((4, 3, 96, 96), (27648, 1, 288, 3), torch
.float32)
triton_poi_fused_1[grid(12, 9216)](primals_3, buf1, 12, 9216,
XBLOCK=64, YBLOCK=16, num_warps=4, num_stages=1)
del primals_3
buf2 = empty_strided_cuda((100, 50, 5, 5), (1250, 1, 250, 50),
torch.float32)
triton_poi_fused_2[grid(5000, 25)](primals_4, buf2, 5000, 25,
XBLOCK=32, YBLOCK=32, num_warps=4, num_stages=1)
del primals_4
buf3 = extern_kernels.convolution(buf1, buf0, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 50, 92, 92), (423200, 1, 4600, 50))
buf4 = buf3
del buf3
triton_poi_fused_convolution_relu_3[grid(1692800)](buf4, primals_2,
1692800, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_2
buf5 = empty_strided_cuda((4, 50, 46, 46), (105800, 1, 2300, 50),
torch.float32)
buf6 = empty_strided_cuda((4, 50, 46, 46), (105800, 1, 2300, 50),
torch.int8)
triton_poi_fused_max_pool2d_with_indices_4[grid(423200)](buf4, buf5,
buf6, 423200, XBLOCK=512, num_warps=8, num_stages=1)
buf7 = extern_kernels.convolution(buf5, buf2, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf7, (4, 100, 42, 42), (176400, 1, 4200, 100))
buf8 = buf7
del buf7
triton_poi_fused_convolution_relu_5[grid(705600)](buf8, primals_5,
705600, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_5
buf9 = empty_strided_cuda((4, 100, 21, 21), (44100, 1, 2100, 100),
torch.int8)
buf10 = empty_strided_cuda((4, 100, 21, 21), (44128, 441, 21, 1),
torch.float32)
triton_poi_fused_max_pool2d_with_indices_6[grid(1764, 100)](buf8,
buf9, buf10, 1764, 100, XBLOCK=128, YBLOCK=8, num_warps=4,
num_stages=1)
buf11 = empty_strided_cuda((4, 1600), (1600, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf10, (4, 44100), (44128, 1),
0), reinterpret_tensor(primals_6, (44100, 1600), (1, 44100), 0),
out=buf11)
buf12 = buf11
del buf11
triton_poi_fused_relu_7[grid(6400)](buf12, primals_7, 6400, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_7
buf13 = empty_strided_cuda((4, 1), (1, 1), torch.float32)
extern_kernels.mm(buf12, reinterpret_tensor(primals_8, (1600, 1), (
1, 1600), 0), out=buf13)
buf14 = buf13
del buf13
triton_poi_fused_sigmoid_8[grid(4)](buf14, primals_9, 4, XBLOCK=4,
num_warps=1, num_stages=1)
del primals_9
return (buf14, buf0, buf1, buf2, buf4, buf5, buf6, buf8, buf9,
reinterpret_tensor(buf10, (4, 44100), (44128, 1), 0), buf12, buf14,
primals_8, primals_6)
class SimpleConvNew(nn.Module):
def __init__(self):
super(SimpleConvNew, self).__init__()
self.conv1 = nn.Conv2d(3, 50, 5, 1)
self.conv2 = nn.Conv2d(50, 100, 5, 1)
self.fc1 = nn.Linear(21 * 21 * 100, 1600)
self.fc2 = nn.Linear(1600, 1)
def forward(self, input_0):
primals_1 = self.conv1.weight
primals_2 = self.conv1.bias
primals_4 = self.conv2.weight
primals_5 = self.conv2.bias
primals_6 = self.fc1.weight
primals_7 = self.fc1.bias
primals_8 = self.fc2.weight
primals_9 = self.fc2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
| junoon53/pcam_challenge | SimpleConv | false | 7,151 | [
"MIT"
] | 1 | 283c98b2d2e211424cdcb56d8230a7a29dc5af46 | https://github.com/junoon53/pcam_challenge/tree/283c98b2d2e211424cdcb56d8230a7a29dc5af46 | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self):
super().__init__()
self.conv1 = nn.Conv2d(3, 50, 5, 1)
self.conv2 = nn.Conv2d(50, 100, 5, 1)
self.fc1 = nn.Linear(21 * 21 * 100, 1600)
self.fc2 = nn.Linear(1600, 1)
def forward(self, x):
x = F.relu(self.conv1(x))
x = F.max_pool2d(x, 2, 2)
x = F.relu(self.conv2(x))
x = F.max_pool2d(x, 2, 2)
x = x.view(-1, 21 * 21 * 100)
x = F.relu(self.fc1(x))
x = self.fc2(x)
o = F.sigmoid(x)
return o
def get_inputs():
return [torch.rand([4, 3, 96, 96])]
def get_init_inputs():
return []
|
BilinearConvLayer | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/q7/cq7qwv755rskgi3fxmqbrnzfm6sxg6uprg2cozcqvgaiyr3e5jdv.py
# Topologically Sorted Source Nodes: [y], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# y => convolution
# Graph fragment:
# %convolution : [num_users=3] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[512],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 8
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/z3/cz3wzaxpnjzxw2tlwd44hdoijks7pwssvfnwqddof4cwhxcvud6q.py
# Topologically Sorted Source Nodes: [z], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# z => mul
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%slice_4, %slice_6), kwargs = {})
triton_poi_fused_mul_1 = async_compile.triton('triton_poi_fused_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 64
x1 = (xindex // 64)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (128*x1)), xmask)
tmp1 = tl.load(in_ptr0 + (64 + x0 + (128*x1)), xmask)
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (8, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_2, (8, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [y], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 8, 4, 4), (128, 16, 4, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [y], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(buf1, primals_2, 512, grid=grid(512), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [z], Original ATen: [aten.mul]
triton_poi_fused_mul_1.run(buf1, buf2, 256, grid=grid(256), stream=stream0)
return (buf2, primals_1, primals_3, buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((8, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((8, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
def setup_conv(in_channels, out_channels, kernel_size, bias, padding_mode,
stride=1, Conv=torch.nn.Conv2d):
return Conv(in_channels=in_channels, out_channels=out_channels,
kernel_size=kernel_size, padding=(kernel_size - 1) // 2, stride=
stride, bias=bias)
class BilinearConvLayer(torch.nn.Module):
def __init__(self, input_channels, output_channels, bilin_channels=None,
padding_mode='zeros', Conv=torch.nn.Conv2d, nonlinearity=torch.nn.
Identity(), norm=torch.nn.Identity(), kernel_size=3):
super(BilinearConvLayer, self).__init__()
bilin_channels = (output_channels if bilin_channels is None else
bilin_channels)
self.chgrp1 = max(0, output_channels - bilin_channels)
self.chgrp2 = bilin_channels
self.layer1 = setup_conv(in_channels=input_channels, out_channels=
self.chgrp1 + 2 * self.chgrp2, kernel_size=kernel_size, bias=
True, padding_mode=padding_mode, stride=1, Conv=Conv)
self.norm = norm
self.nonlinearity = nonlinearity
def forward(self, x):
y = self.nonlinearity(self.norm(self.layer1(x)))
mid = self.chgrp1 + self.chgrp2
y1, y2, y3 = y[:, :self.chgrp1], y[:, self.chgrp1:mid], y[:, mid:]
z = y2 * y3
out = torch.cat((y1, z), dim=1)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'input_channels': 4, 'output_channels': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 8
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
@triton.jit
def triton_poi_fused_mul_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 64
x1 = xindex // 64
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 128 * x1), xmask)
tmp1 = tl.load(in_ptr0 + (64 + x0 + 128 * x1), xmask)
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (8, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_2, (8,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,
1), padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 8, 4, 4), (128, 16, 4, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(512)](buf1, primals_2, 512,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_mul_1[grid(256)](buf1, buf2, 256, XBLOCK=128,
num_warps=4, num_stages=1)
return buf2, primals_1, primals_3, buf1
def setup_conv(in_channels, out_channels, kernel_size, bias, padding_mode,
stride=1, Conv=torch.nn.Conv2d):
return Conv(in_channels=in_channels, out_channels=out_channels,
kernel_size=kernel_size, padding=(kernel_size - 1) // 2, stride=
stride, bias=bias)
class BilinearConvLayerNew(torch.nn.Module):
def __init__(self, input_channels, output_channels, bilin_channels=None,
padding_mode='zeros', Conv=torch.nn.Conv2d, nonlinearity=torch.nn.
Identity(), norm=torch.nn.Identity(), kernel_size=3):
super(BilinearConvLayerNew, self).__init__()
bilin_channels = (output_channels if bilin_channels is None else
bilin_channels)
self.chgrp1 = max(0, output_channels - bilin_channels)
self.chgrp2 = bilin_channels
self.layer1 = setup_conv(in_channels=input_channels, out_channels=
self.chgrp1 + 2 * self.chgrp2, kernel_size=kernel_size, bias=
True, padding_mode=padding_mode, stride=1, Conv=Conv)
self.norm = norm
self.nonlinearity = nonlinearity
def forward(self, input_0):
primals_1 = self.layer1.weight
primals_2 = self.layer1.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| m-dml/lil2021swe | BilinearConvLayer | false | 7,152 | [
"Apache-2.0"
] | 1 | 45352f214ec28c9f91dd24ed3669f492d8b68382 | https://github.com/m-dml/lil2021swe/tree/45352f214ec28c9f91dd24ed3669f492d8b68382 | import torch
def setup_conv(in_channels, out_channels, kernel_size, bias, padding_mode,
stride=1, Conv=torch.nn.Conv2d):
return Conv(in_channels=in_channels, out_channels=out_channels,
kernel_size=kernel_size, padding=(kernel_size - 1) // 2, stride=
stride, bias=bias)
class Model(torch.nn.Module):
def __init__(self, input_channels, output_channels, bilin_channels=None,
padding_mode='zeros', Conv=torch.nn.Conv2d, nonlinearity=torch.nn.
Identity(), norm=torch.nn.Identity(), kernel_size=3):
super().__init__()
bilin_channels = (output_channels if bilin_channels is None else
bilin_channels)
self.chgrp1 = max(0, output_channels - bilin_channels)
self.chgrp2 = bilin_channels
self.layer1 = setup_conv(in_channels=input_channels, out_channels=
self.chgrp1 + 2 * self.chgrp2, kernel_size=kernel_size, bias=
True, padding_mode=padding_mode, stride=1, Conv=Conv)
self.norm = norm
self.nonlinearity = nonlinearity
def forward(self, x):
y = self.nonlinearity(self.norm(self.layer1(x)))
mid = self.chgrp1 + self.chgrp2
y1, y2, y3 = y[:, :self.chgrp1], y[:, self.chgrp1:mid], y[:, mid:]
z = y2 * y3
out = torch.cat((y1, z), dim=1)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
FFN | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/z5/cz5cv4otoqqg7gisuc6bx7rlau6zkmkldsfxasbc2l2hvtqjgjou.py
# Topologically Sorted Source Nodes: [cated_vectors], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cated_vectors => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%primals_1, %primals_2, %primals_3], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 768
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 16) % 12
x0 = xindex % 16
x2 = (xindex // 192)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (16*x1) + (64*x2)), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + (16*((-4) + x1)) + (64*x2)), tmp9 & xmask, other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 12, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tl.load(in_ptr2 + (x0 + (16*((-8) + x1)) + (64*x2)), tmp11 & xmask, other=0.0)
tmp15 = tl.where(tmp9, tmp10, tmp14)
tmp16 = tl.where(tmp4, tmp5, tmp15)
tl.store(out_ptr0 + (x3), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/lk/clkvwvqcst6t4cucaxdngnnuhangq7kuawiiub5xqxykvb3eih2u.py
# Topologically Sorted Source Nodes: [mul, pow_1, mul_1, add, mul_2, tanh, add_1, mul_3], Original ATen: [aten.mul, aten.pow, aten.add, aten.tanh]
# Source node to ATen node mapping:
# add => add
# add_1 => add_1
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# mul_3 => mul_3
# pow_1 => pow_1
# tanh => tanh
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_1, 0.5), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%view_1, 3), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%pow_1, 0.044715), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_1, %mul_1), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add, 0.7978845608028654), kwargs = {})
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%mul_2,), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%tanh, 1), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %add_1), kwargs = {})
triton_poi_fused_add_mul_pow_tanh_1 = async_compile.triton('triton_poi_fused_add_mul_pow_tanh_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_pow_tanh_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_pow_tanh_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 768
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tmp3 = tmp0 * tmp0
tmp4 = tmp3 * tmp0
tmp5 = 0.044715
tmp6 = tmp4 * tmp5
tmp7 = tmp0 + tmp6
tmp8 = 0.7978845608028654
tmp9 = tmp7 * tmp8
tmp10 = libdevice.tanh(tmp9)
tmp11 = 1.0
tmp12 = tmp10 + tmp11
tmp13 = tmp2 * tmp12
tl.store(out_ptr0 + (x0), tmp13, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/qk/cqk4fc6sh5nxra5g2ze56qihubkj2kgf6kgs6cvrl7zew3sie6gm.py
# Topologically Sorted Source Nodes: [low_vectors_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# low_vectors_1 => add_2
# Graph fragment:
# %add_2 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_1, %getitem), kwargs = {})
triton_poi_fused_add_2 = async_compile.triton('triton_poi_fused_add_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
x1 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0 + (192*x1)), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/3o/c3ohg5u3kgv4nzwhohtfisgwxfppk7dvczwpyrw3fsxa3lspzcrm.py
# Topologically Sorted Source Nodes: [mid_vectors_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# mid_vectors_1 => add_3
# Graph fragment:
# %add_3 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %getitem_1), kwargs = {})
triton_poi_fused_add_3 = async_compile.triton('triton_poi_fused_add_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_3(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
x1 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (64 + x0 + (192*x1)), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ci/ccicvllvu6lhrojla6mefablakbyu4tzhjfeosnnd47v7p7yde2j.py
# Topologically Sorted Source Nodes: [hig_vectors_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# hig_vectors_1 => add_4
# Graph fragment:
# %add_4 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_3, %getitem_2), kwargs = {})
triton_poi_fused_add_4 = async_compile.triton('triton_poi_fused_add_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_4(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
x1 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (128 + x0 + (192*x1)), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/we/cwehzh3z7s3nstmx7epm33gsgvy4mbg6xttmec2livc3ylvoisbm.py
# Topologically Sorted Source Nodes: [low_vectors_2], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# low_vectors_2 => add_5, rsqrt, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add_2, [3]), kwargs = {correction: 0, keepdim: True})
# %add_5 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_3, 1e-08), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_5,), kwargs = {})
triton_poi_fused_native_layer_norm_5 = async_compile.triton('triton_poi_fused_native_layer_norm_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_5(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-08
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x0), tmp8, xmask)
tl.store(out_ptr1 + (x0), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/gm/cgmzdl6z4u4iu5bafof4nvofri54twnq7gifufkslpwxjmyguzb3.py
# Topologically Sorted Source Nodes: [low_vectors_2], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# low_vectors_2 => add_5, add_6, mul_4, mul_5, rsqrt, sub, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add_2, [3]), kwargs = {correction: 0, keepdim: True})
# %add_5 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_3, 1e-08), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_5,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_2, %getitem_4), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %rsqrt), kwargs = {})
# %mul_5 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_4, %primals_6), kwargs = {})
# %add_6 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_5, %primals_7), kwargs = {})
triton_poi_fused_native_layer_norm_6 = async_compile.triton('triton_poi_fused_native_layer_norm_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_6(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (4, ), (1, ))
assert_size_stride(primals_9, (4, ), (1, ))
assert_size_stride(primals_10, (4, ), (1, ))
assert_size_stride(primals_11, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 12, 4, 4), (192, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cated_vectors], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(primals_1, primals_2, primals_3, buf0, 768, grid=grid(768), stream=stream0)
buf1 = empty_strided_cuda((192, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf0, (192, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((4, 12, 4, 4), (192, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, pow_1, mul_1, add, mul_2, tanh, add_1, mul_3], Original ATen: [aten.mul, aten.pow, aten.add, aten.tanh]
triton_poi_fused_add_mul_pow_tanh_1.run(buf1, buf2, 768, grid=grid(768), stream=stream0)
buf3 = empty_strided_cuda((192, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf2, (192, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [low_vectors_1], Original ATen: [aten.add]
triton_poi_fused_add_2.run(primals_1, buf3, buf4, 256, grid=grid(256), stream=stream0)
del primals_1
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mid_vectors_1], Original ATen: [aten.add]
triton_poi_fused_add_3.run(primals_2, buf3, buf5, 256, grid=grid(256), stream=stream0)
del primals_2
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [hig_vectors_1], Original ATen: [aten.add]
triton_poi_fused_add_4.run(primals_3, buf3, buf6, 256, grid=grid(256), stream=stream0)
del buf3
del primals_3
buf7 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf8 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
# Topologically Sorted Source Nodes: [low_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_5.run(buf4, buf7, buf8, 64, grid=grid(64), stream=stream0)
buf9 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [low_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_6.run(buf4, buf7, buf8, primals_6, primals_7, buf9, 256, grid=grid(256), stream=stream0)
del primals_7
buf10 = buf8; del buf8 # reuse
buf11 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [mid_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_5.run(buf5, buf10, buf11, 64, grid=grid(64), stream=stream0)
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mid_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_6.run(buf5, buf10, buf11, primals_8, primals_9, buf12, 256, grid=grid(256), stream=stream0)
del primals_9
buf13 = buf11; del buf11 # reuse
buf14 = buf10; del buf10 # reuse
# Topologically Sorted Source Nodes: [hig_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_5.run(buf6, buf13, buf14, 64, grid=grid(64), stream=stream0)
buf15 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [hig_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_6.run(buf6, buf13, buf14, primals_10, primals_11, buf15, 256, grid=grid(256), stream=stream0)
del buf13
del buf14
del primals_11
return (buf9, buf12, buf15, primals_6, primals_8, primals_10, reinterpret_tensor(buf0, (192, 4), (4, 1), 0), buf1, reinterpret_tensor(buf2, (192, 4), (4, 1), 0), buf4, buf5, buf6, primals_5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import torch.nn as nn
class GELU(nn.Module):
"""
Paper Section 3.4, last paragraph notice that BERT used the GELU instead of RELU
came from : https://github.com/codertimo/BERT-pytorch/blob/master/bert_pytorch/model/utils/gelu.py
"""
def __init__(self):
super(GELU, self).__init__()
def forward(self, x):
return 0.5 * x * (1 + torch.tanh(math.sqrt(2 / math.pi) * (x +
0.044715 * torch.pow(x, 3))))
class FFN(nn.Module):
def __init__(self, embedding_dim, hidden_unit, dropout=0.0, eps=1e-08):
super(FFN, self).__init__()
self.fc1 = nn.Linear(embedding_dim, hidden_unit, bias=False)
self.fc2 = nn.Linear(hidden_unit, embedding_dim, bias=False)
self.dropout1 = nn.Dropout(dropout)
self.dropout2 = nn.Dropout(dropout)
self.low_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.act = GELU()
def forward(self, low_vectors, mid_vectors, hig_vectors):
low_num, mid_num, hig_num = low_vectors.size()[1], mid_vectors.size()[1
], hig_vectors.size()[1]
low_residual = low_vectors
mid_residual = mid_vectors
hig_residual = hig_vectors
cated_vectors = torch.cat((low_vectors, mid_vectors, hig_vectors),
dim=1)
output = self.dropout2(self.fc2(self.dropout1(self.act(self.fc1(
cated_vectors)))))
low_vectors, mid_vectors, hig_vectors = torch.split(output, [
low_num, mid_num, hig_num], dim=1)
low_vectors = low_residual + low_vectors
mid_vectors = mid_residual + mid_vectors
hig_vectors = hig_residual + hig_vectors
low_vectors = self.low_ln(low_vectors)
mid_vectors = self.mid_ln(mid_vectors)
hig_vectors = self.hig_ln(hig_vectors)
return low_vectors, mid_vectors, hig_vectors
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [[], {'embedding_dim': 4, 'hidden_unit': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 768
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 16 % 12
x0 = xindex % 16
x2 = xindex // 192
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 16 * x1 + 64 * x2), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + 16 * (-4 + x1) + 64 * x2), tmp9 & xmask,
other=0.0)
tmp11 = tmp0 >= tmp7
tl.full([1], 12, tl.int64)
tmp14 = tl.load(in_ptr2 + (x0 + 16 * (-8 + x1) + 64 * x2), tmp11 &
xmask, other=0.0)
tmp15 = tl.where(tmp9, tmp10, tmp14)
tmp16 = tl.where(tmp4, tmp5, tmp15)
tl.store(out_ptr0 + x3, tmp16, xmask)
@triton.jit
def triton_poi_fused_add_mul_pow_tanh_1(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 768
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.5
tmp2 = tmp0 * tmp1
tmp3 = tmp0 * tmp0
tmp4 = tmp3 * tmp0
tmp5 = 0.044715
tmp6 = tmp4 * tmp5
tmp7 = tmp0 + tmp6
tmp8 = 0.7978845608028654
tmp9 = tmp7 * tmp8
tmp10 = libdevice.tanh(tmp9)
tmp11 = 1.0
tmp12 = tmp10 + tmp11
tmp13 = tmp2 * tmp12
tl.store(out_ptr0 + x0, tmp13, xmask)
@triton.jit
def triton_poi_fused_add_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
x1 = xindex // 64
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (x0 + 192 * x1), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_add_3(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
x1 = xindex // 64
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (64 + x0 + 192 * x1), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_add_4(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 64
x1 = xindex // 64
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (128 + x0 + 192 * x1), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_5(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-08
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x0, tmp8, xmask)
tl.store(out_ptr1 + x0, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_6(in_ptr0, in_ptr1, in_ptr2, in_ptr3,
in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4,), (1,))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (4,), (1,))
assert_size_stride(primals_9, (4,), (1,))
assert_size_stride(primals_10, (4,), (1,))
assert_size_stride(primals_11, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 12, 4, 4), (192, 16, 4, 1), torch.float32
)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(768)](primals_1, primals_2, primals_3,
buf0, 768, XBLOCK=128, num_warps=4, num_stages=1)
buf1 = empty_strided_cuda((192, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf0, (192, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((4, 12, 4, 4), (192, 16, 4, 1), torch.float32
)
triton_poi_fused_add_mul_pow_tanh_1[grid(768)](buf1, buf2, 768,
XBLOCK=256, num_warps=4, num_stages=1)
buf3 = empty_strided_cuda((192, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf2, (192, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_add_2[grid(256)](primals_1, buf3, buf4, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_1
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_add_3[grid(256)](primals_2, buf3, buf5, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_2
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_add_4[grid(256)](primals_3, buf3, buf6, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del buf3
del primals_3
buf7 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf8 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
triton_poi_fused_native_layer_norm_5[grid(64)](buf4, buf7, buf8, 64,
XBLOCK=64, num_warps=1, num_stages=1)
buf9 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_6[grid(256)](buf4, buf7, buf8,
primals_6, primals_7, buf9, 256, XBLOCK=256, num_warps=4,
num_stages=1)
del primals_7
buf10 = buf8
del buf8
buf11 = buf7
del buf7
triton_poi_fused_native_layer_norm_5[grid(64)](buf5, buf10, buf11,
64, XBLOCK=64, num_warps=1, num_stages=1)
buf12 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_6[grid(256)](buf5, buf10, buf11,
primals_8, primals_9, buf12, 256, XBLOCK=256, num_warps=4,
num_stages=1)
del primals_9
buf13 = buf11
del buf11
buf14 = buf10
del buf10
triton_poi_fused_native_layer_norm_5[grid(64)](buf6, buf13, buf14,
64, XBLOCK=64, num_warps=1, num_stages=1)
buf15 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_6[grid(256)](buf6, buf13, buf14,
primals_10, primals_11, buf15, 256, XBLOCK=256, num_warps=4,
num_stages=1)
del buf13
del buf14
del primals_11
return (buf9, buf12, buf15, primals_6, primals_8, primals_10,
reinterpret_tensor(buf0, (192, 4), (4, 1), 0), buf1,
reinterpret_tensor(buf2, (192, 4), (4, 1), 0), buf4, buf5, buf6,
primals_5)
class GELU(nn.Module):
"""
Paper Section 3.4, last paragraph notice that BERT used the GELU instead of RELU
came from : https://github.com/codertimo/BERT-pytorch/blob/master/bert_pytorch/model/utils/gelu.py
"""
def __init__(self):
super(GELU, self).__init__()
def forward(self, x):
return 0.5 * x * (1 + torch.tanh(math.sqrt(2 / math.pi) * (x +
0.044715 * torch.pow(x, 3))))
class FFNNew(nn.Module):
def __init__(self, embedding_dim, hidden_unit, dropout=0.0, eps=1e-08):
super(FFNNew, self).__init__()
self.fc1 = nn.Linear(embedding_dim, hidden_unit, bias=False)
self.fc2 = nn.Linear(hidden_unit, embedding_dim, bias=False)
self.dropout1 = nn.Dropout(dropout)
self.dropout2 = nn.Dropout(dropout)
self.low_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.act = GELU()
def forward(self, input_0, input_1, input_2):
primals_4 = self.fc1.weight
primals_5 = self.fc2.weight
primals_6 = self.low_ln.weight
primals_7 = self.low_ln.bias
primals_8 = self.mid_ln.weight
primals_9 = self.mid_ln.bias
primals_10 = self.hig_ln.weight
primals_11 = self.hig_ln.bias
primals_1 = input_0
primals_2 = input_1
primals_3 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11])
return output[0], output[1], output[2]
| luyu-fan/LRCM | FFN | false | 7,153 | [
"MIT"
] | 1 | 6b0e4d7998bc4969afa764eb753077e3f858f1ba | https://github.com/luyu-fan/LRCM/tree/6b0e4d7998bc4969afa764eb753077e3f858f1ba | import math
import torch
import torch.nn as nn
class GELU(nn.Module):
"""
Paper Section 3.4, last paragraph notice that BERT used the GELU instead of RELU
came from : https://github.com/codertimo/BERT-pytorch/blob/master/bert_pytorch/model/utils/gelu.py
"""
def __init__(self):
super().__init__()
def forward(self, x):
return 0.5 * x * (1 + torch.tanh(math.sqrt(2 / math.pi) * (x +
0.044715 * torch.pow(x, 3))))
class Model(nn.Module):
def __init__(self, embedding_dim, hidden_unit, dropout=0.0, eps=1e-08):
super().__init__()
self.fc1 = nn.Linear(embedding_dim, hidden_unit, bias=False)
self.fc2 = nn.Linear(hidden_unit, embedding_dim, bias=False)
self.dropout1 = nn.Dropout(dropout)
self.dropout2 = nn.Dropout(dropout)
self.low_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.act = GELU()
def forward(self, low_vectors, mid_vectors, hig_vectors):
low_num, mid_num, hig_num = low_vectors.size()[1], mid_vectors.size()[1
], hig_vectors.size()[1]
low_residual = low_vectors
mid_residual = mid_vectors
hig_residual = hig_vectors
cated_vectors = torch.cat((low_vectors, mid_vectors, hig_vectors),
dim=1)
output = self.dropout2(self.fc2(self.dropout1(self.act(self.fc1(
cated_vectors)))))
low_vectors, mid_vectors, hig_vectors = torch.split(output, [
low_num, mid_num, hig_num], dim=1)
low_vectors = low_residual + low_vectors
mid_vectors = mid_residual + mid_vectors
hig_vectors = hig_residual + hig_vectors
low_vectors = self.low_ln(low_vectors)
mid_vectors = self.mid_ln(mid_vectors)
hig_vectors = self.hig_ln(hig_vectors)
return low_vectors, mid_vectors, hig_vectors
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
CRN | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/4d/c4d7os35bf4bckecmik4nlyqqsirmteh4sh3yxnab5lmuntnmwk2.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_0 = async_compile.triton('triton_poi_fused_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 128
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = (yindex // 4)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (4*x2) + (36*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/lf/clfcn5gfhri4nhvee4viezhjoojfy5qn3nygre4bx75qrms4qd3n.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_1 = async_compile.triton('triton_poi_fused_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128, 32], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 128
xnumel = 25
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = (yindex // 4)
tmp0 = tl.load(in_ptr0 + (x2 + (25*y3)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (4*x2) + (100*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/oa/coaftdaxqdmymqkwsf3jue6wikwr6e2kd4b3crc5ijigktwyfnc7.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_2 = async_compile.triton('triton_poi_fused_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128, 64], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 80
xnumel = 49
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = (yindex // 4)
tmp0 = tl.load(in_ptr0 + (x2 + (49*y3)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (4*x2) + (196*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/bq/cbqzqjopcg7rquo5pcfwokk5u2cqvyw4oleohxzzvguydmlqt543.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten._adaptive_avg_pool2d]
# Source node to ATen node mapping:
# x => _adaptive_avg_pool2d
# Graph fragment:
# %_adaptive_avg_pool2d : [num_users=4] = call_function[target=torch.ops.aten._adaptive_avg_pool2d.default](args = (%primals_1, [13, 13]), kwargs = {})
triton_poi_fused__adaptive_avg_pool2d_3 = async_compile.triton('triton_poi_fused__adaptive_avg_pool2d_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__adaptive_avg_pool2d_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__adaptive_avg_pool2d_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 2704
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = (xindex // 52) % 13
x1 = (xindex // 4) % 13
x0 = xindex % 4
x3 = (xindex // 676)
x6 = xindex
tmp0 = ((4*x2) // 13)
tmp1 = ((16 + (4*x2)) // 13)
tmp2 = tmp0 < tmp1
tmp3 = ((4*x1) // 13)
tmp4 = ((16 + (4*x1)) // 13)
tmp5 = tmp3 < tmp4
tmp6 = tmp2 & tmp5
tmp7 = tl.load(in_ptr0 + ((4*((4*x2) // 13)) + (16*x0) + (64*x3) + ((4*x1) // 13)), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp8 = 1 + ((4*x1) // 13)
tmp9 = tmp8 < tmp4
tmp10 = tmp2 & tmp9
tmp11 = tl.load(in_ptr0 + (1 + (4*((4*x2) // 13)) + (16*x0) + (64*x3) + ((4*x1) // 13)), tmp10 & xmask, eviction_policy='evict_last', other=0.0)
tmp12 = tmp11 + tmp7
tmp13 = 1 + ((4*x2) // 13)
tmp14 = tmp13 < tmp1
tmp15 = tmp14 & tmp5
tmp16 = tl.load(in_ptr0 + (4 + (4*((4*x2) // 13)) + (16*x0) + (64*x3) + ((4*x1) // 13)), tmp15 & xmask, eviction_policy='evict_last', other=0.0)
tmp17 = tmp16 + tmp12
tmp18 = tmp14 & tmp9
tmp19 = tl.load(in_ptr0 + (5 + (4*((4*x2) // 13)) + (16*x0) + (64*x3) + ((4*x1) // 13)), tmp18 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tmp19 + tmp17
tmp21 = 1.0
tmp22 = tl.full(tmp21.shape, 0.0, tmp21.dtype)
tmp23 = tl.where(tmp6, tmp21, tmp22)
tmp24 = tl.where(tmp10, tmp21, tmp22)
tmp25 = tmp24 + tmp23
tmp26 = tl.where(tmp15, tmp21, tmp22)
tmp27 = tmp26 + tmp25
tmp28 = tl.where(tmp18, tmp21, tmp22)
tmp29 = tmp28 + tmp27
tmp30 = tmp20 / tmp29
tl.store(out_ptr0 + (x6), tmp30, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/cz/cczxbirje7zmag4quxpbquw63c24dvrazbdkwdkw4fdmkxoba5sd.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# x_1 => cat
# Graph fragment:
# %cat : [num_users=2] = call_function[target=torch.ops.aten.cat.default](args = ([%relu, %relu_1, %relu_2], 1), kwargs = {})
triton_poi_fused_cat_4 = async_compile.triton('triton_poi_fused_cat_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[65536],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_4(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 56784
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 84
x1 = (xindex // 84)
x2 = xindex
tmp0 = x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 32, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + ((32*x1) + x0), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tl.load(in_ptr1 + (x0), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp7 = tmp5 + tmp6
tmp8 = tl.full([1], 0, tl.int32)
tmp9 = triton_helpers.maximum(tmp8, tmp7)
tmp10 = tl.full(tmp9.shape, 0.0, tmp9.dtype)
tmp11 = tl.where(tmp4, tmp9, tmp10)
tmp12 = tmp0 >= tmp3
tmp13 = tl.full([1], 64, tl.int64)
tmp14 = tmp0 < tmp13
tmp15 = tmp12 & tmp14
tmp16 = tl.load(in_ptr2 + ((32*x1) + ((-32) + x0)), tmp15 & xmask, eviction_policy='evict_last', other=0.0)
tmp17 = tl.load(in_ptr3 + ((-32) + x0), tmp15 & xmask, eviction_policy='evict_last', other=0.0)
tmp18 = tmp16 + tmp17
tmp19 = triton_helpers.maximum(tmp8, tmp18)
tmp20 = tl.full(tmp19.shape, 0.0, tmp19.dtype)
tmp21 = tl.where(tmp15, tmp19, tmp20)
tmp22 = tmp0 >= tmp13
tmp23 = tl.full([1], 84, tl.int64)
tmp24 = tmp0 < tmp23
tmp25 = tl.load(in_ptr4 + ((20*x1) + ((-64) + x0)), tmp22 & xmask, eviction_policy='evict_last', other=0.0)
tmp26 = tl.load(in_ptr5 + ((-64) + x0), tmp22 & xmask, eviction_policy='evict_last', other=0.0)
tmp27 = tmp25 + tmp26
tmp28 = triton_helpers.maximum(tmp8, tmp27)
tmp29 = tl.full(tmp28.shape, 0.0, tmp28.dtype)
tmp30 = tl.where(tmp22, tmp28, tmp29)
tmp31 = tl.where(tmp15, tmp21, tmp30)
tmp32 = tl.where(tmp4, tmp11, tmp31)
tl.store(out_ptr0 + (x2), tmp32, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/c7/cc7vg3vp6rt7i5czyvtxc4u4srtr22u26lqgxea6dagdtwkjxhde.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.arange, aten.add, aten.mul, aten._to_copy]
# Source node to ATen node mapping:
# x_3 => add, add_1, convert_element_type, convert_element_type_1, iota, mul, mul_1
# Graph fragment:
# %iota : [num_users=1] = call_function[target=torch.ops.prims.iota.default](args = (4,), kwargs = {start: 0, step: 1, dtype: torch.int64, device: cuda:0, requires_grad: False})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%iota, 1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, 0), kwargs = {})
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%add, torch.float32), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convert_element_type, 0.0), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_1, 3.25), kwargs = {})
# %convert_element_type_1 : [num_users=3] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%mul_1, torch.int64), kwargs = {})
triton_poi_fused__to_copy_add_arange_mul_5 = async_compile.triton('triton_poi_fused__to_copy_add_arange_mul_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0,), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_add_arange_mul_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_add_arange_mul_5(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 3.25
tmp3 = tmp1 * tmp2
tmp4 = tmp3.to(tl.int32)
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ji/cjixe3kw6mcg4wg7ou63cb26cboath5eiifzkkwdha5rwajcikos.py
# Topologically Sorted Source Nodes: [conv2d_3, x_2, x_3], Original ATen: [aten.convolution, aten.relu, aten._unsafe_index]
# Source node to ATen node mapping:
# conv2d_3 => convolution_3
# x_2 => relu_3
# x_3 => _unsafe_index
# Graph fragment:
# %convolution_3 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%cat, %primals_8, %primals_9, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_3 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_3,), kwargs = {})
# %_unsafe_index : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_3, [None, None, %unsqueeze, %convert_element_type_1]), kwargs = {})
triton_poi_fused__unsafe_index_convolution_relu_6 = async_compile.triton('triton_poi_fused__unsafe_index_convolution_relu_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__unsafe_index_convolution_relu_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__unsafe_index_convolution_relu_6(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x0 = xindex % 4
x2 = (xindex // 16)
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr2 + (0))
tmp11 = tl.broadcast_to(tmp10, [XBLOCK])
tmp1 = tl.full([XBLOCK], 13, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tmp6 = tmp5 + tmp1
tmp7 = tmp5 < 0
tmp8 = tl.where(tmp7, tmp6, tmp5)
tmp9 = tl.load(in_ptr1 + (tmp8 + (13*tmp4) + (169*x2)), xmask, eviction_policy='evict_last')
tmp12 = tmp9 + tmp11
tmp13 = tl.full([1], 0, tl.int32)
tmp14 = triton_helpers.maximum(tmp13, tmp12)
tl.store(out_ptr0 + (x4), tmp14, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/3h/c3hfbj6mk3kpl2k7vav7la4davtsxxy2gheyxcoxcfqfqj7pf7ep.py
# Topologically Sorted Source Nodes: [conv2d_3, x_2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# conv2d_3 => convolution_3
# x_2 => relu_3
# Graph fragment:
# %convolution_3 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%cat, %primals_8, %primals_9, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_3 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_3,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_3, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_7 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_7(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 676
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr1 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.full([1], 0, tl.int32)
tmp5 = triton_helpers.maximum(tmp4, tmp3)
tmp6 = 0.0
tmp7 = tmp5 <= tmp6
tl.store(out_ptr0 + (x0), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/we/cweie3u3hadgguqvfwjj6rti75pfxgznxvyfzvtvunbesaswu536.py
# Topologically Sorted Source Nodes: [conv2d_2, x3], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# conv2d_2 => convolution_2
# x3 => relu_2
# Graph fragment:
# %convolution_2 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%_adaptive_avg_pool2d, %primals_6, %primals_7, [1, 1], [3, 3], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_2,), kwargs = {})
# %le_1 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_2, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_8 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16384],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_8', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_8(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 13520
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 20
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/xl/cxl4x74l4avjy5s34pyd6st75c3nb76ytdignjkuu2wy34l4ompp.py
# Topologically Sorted Source Nodes: [conv2d_1, x2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# conv2d_1 => convolution_1
# x2 => relu_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%_adaptive_avg_pool2d, %primals_4, %primals_5, [1, 1], [2, 2], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_1,), kwargs = {})
# %le_2 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_1, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_9 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_9', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32768],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_9', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_9(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 21632
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 32
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (32, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_3, (32, ), (1, ))
assert_size_stride(primals_4, (32, 4, 5, 5), (100, 25, 5, 1))
assert_size_stride(primals_5, (32, ), (1, ))
assert_size_stride(primals_6, (20, 4, 7, 7), (196, 49, 7, 1))
assert_size_stride(primals_7, (20, ), (1, ))
assert_size_stride(primals_8, (1, 84, 1, 1), (84, 1, 1, 1))
assert_size_stride(primals_9, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((32, 4, 3, 3), (36, 1, 12, 4), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
stream0 = get_raw_stream(0)
triton_poi_fused_0.run(primals_2, buf0, 128, 9, grid=grid(128, 9), stream=stream0)
del primals_2
buf1 = empty_strided_cuda((32, 4, 5, 5), (100, 1, 20, 4), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_1.run(primals_4, buf1, 128, 25, grid=grid(128, 25), stream=stream0)
del primals_4
buf2 = empty_strided_cuda((20, 4, 7, 7), (196, 1, 28, 4), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_2.run(primals_6, buf2, 80, 49, grid=grid(80, 49), stream=stream0)
del primals_6
buf3 = empty_strided_cuda((4, 4, 13, 13), (676, 1, 52, 4), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten._adaptive_avg_pool2d]
triton_poi_fused__adaptive_avg_pool2d_3.run(primals_1, buf3, 2704, grid=grid(2704), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(buf3, buf0, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 32, 13, 13), (5408, 1, 416, 32))
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf5 = extern_kernels.convolution(buf3, buf1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf5, (4, 32, 13, 13), (5408, 1, 416, 32))
# Topologically Sorted Source Nodes: [conv2d_2], Original ATen: [aten.convolution]
buf6 = extern_kernels.convolution(buf3, buf2, stride=(1, 1), padding=(3, 3), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf6, (4, 20, 13, 13), (3380, 1, 260, 20))
buf7 = empty_strided_cuda((4, 84, 13, 13), (14196, 1, 1092, 84), torch.float32)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.cat]
triton_poi_fused_cat_4.run(buf4, primals_3, buf5, primals_5, buf6, primals_7, buf7, 56784, grid=grid(56784), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_3], Original ATen: [aten.convolution]
buf8 = extern_kernels.convolution(buf7, primals_8, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf8, (4, 1, 13, 13), (169, 1, 13, 1))
buf9 = empty_strided_cuda((4, ), (1, ), torch.int64)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.arange, aten.add, aten.mul, aten._to_copy]
triton_poi_fused__to_copy_add_arange_mul_5.run(buf9, 4, grid=grid(4), stream=stream0)
buf10 = empty_strided_cuda((4, 1, 4, 4), (16, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [conv2d_3, x_2, x_3], Original ATen: [aten.convolution, aten.relu, aten._unsafe_index]
triton_poi_fused__unsafe_index_convolution_relu_6.run(buf9, buf8, primals_9, buf10, 64, grid=grid(64), stream=stream0)
buf11 = empty_strided_cuda((4, 1, 13, 13), (169, 1, 13, 1), torch.bool)
# Topologically Sorted Source Nodes: [conv2d_3, x_2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_7.run(buf8, primals_9, buf11, 676, grid=grid(676), stream=stream0)
del buf8
del primals_9
buf12 = empty_strided_cuda((4, 20, 13, 13), (3380, 1, 260, 20), torch.bool)
# Topologically Sorted Source Nodes: [conv2d_2, x3], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_8.run(buf6, primals_7, buf12, 13520, grid=grid(13520), stream=stream0)
del buf6
del primals_7
buf13 = empty_strided_cuda((4, 32, 13, 13), (5408, 1, 416, 32), torch.bool)
# Topologically Sorted Source Nodes: [conv2d_1, x2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_9.run(buf5, primals_5, buf13, 21632, grid=grid(21632), stream=stream0)
del buf5
del primals_5
buf14 = empty_strided_cuda((4, 32, 13, 13), (5408, 1, 416, 32), torch.bool)
# Topologically Sorted Source Nodes: [conv2d, x1], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_9.run(buf4, primals_3, buf14, 21632, grid=grid(21632), stream=stream0)
del buf4
del primals_3
return (buf10, buf0, buf1, buf2, primals_8, buf3, buf7, buf9, buf11, buf12, buf13, buf14, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((32, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((32, 4, 5, 5), (100, 25, 5, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((20, 4, 7, 7), (196, 49, 7, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((20, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((1, 84, 1, 1), (84, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn.functional as F
class CRN(torch.nn.Module):
def __init__(self, dim):
super(CRN, self).__init__()
self.h_w = 13, 13
self.downsample = torch.nn.AdaptiveAvgPool2d(self.h_w)
n_filters = [32, 32, 20]
self.conv1 = torch.nn.Conv2d(dim, n_filters[0], 3, padding=1)
self.conv2 = torch.nn.Conv2d(dim, n_filters[1], 5, padding=2)
self.conv3 = torch.nn.Conv2d(dim, n_filters[2], 7, padding=3)
self.conv_accum = torch.nn.Conv2d(sum(n_filters), 1, 1)
for m in self.modules():
if isinstance(m, torch.nn.Conv2d):
torch.nn.init.kaiming_normal_(m.weight, mode='fan_out',
nonlinearity='relu')
if m.bias is not None:
torch.nn.init.zeros_(m.bias)
def forward(self, x):
input_h_w = x.shape[2:]
x = self.downsample(x)
x1 = F.relu(self.conv1(x))
x2 = F.relu(self.conv2(x))
x3 = F.relu(self.conv3(x))
x = torch.cat((x1, x2, x3), dim=1)
x = F.relu(self.conv_accum(x))
x = F.interpolate(x, input_h_w)
assert x.shape[2:] == input_h_w
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'dim': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 128
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = yindex // 4
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask & ymask, eviction_policy=
'evict_last')
tl.store(out_ptr0 + (y0 + 4 * x2 + 36 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 128
xnumel = 25
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = yindex // 4
tmp0 = tl.load(in_ptr0 + (x2 + 25 * y3), xmask & ymask, eviction_policy
='evict_last')
tl.store(out_ptr0 + (y0 + 4 * x2 + 100 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 80
xnumel = 49
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 4
y1 = yindex // 4
tmp0 = tl.load(in_ptr0 + (x2 + 49 * y3), xmask & ymask, eviction_policy
='evict_last')
tl.store(out_ptr0 + (y0 + 4 * x2 + 196 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused__adaptive_avg_pool2d_3(in_ptr0, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 2704
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex // 52 % 13
x1 = xindex // 4 % 13
x0 = xindex % 4
x3 = xindex // 676
x6 = xindex
tmp0 = 4 * x2 // 13
tmp1 = (16 + 4 * x2) // 13
tmp2 = tmp0 < tmp1
tmp3 = 4 * x1 // 13
tmp4 = (16 + 4 * x1) // 13
tmp5 = tmp3 < tmp4
tmp6 = tmp2 & tmp5
tmp7 = tl.load(in_ptr0 + (4 * (4 * x2 // 13) + 16 * x0 + 64 * x3 + 4 *
x1 // 13), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp8 = 1 + 4 * x1 // 13
tmp9 = tmp8 < tmp4
tmp10 = tmp2 & tmp9
tmp11 = tl.load(in_ptr0 + (1 + 4 * (4 * x2 // 13) + 16 * x0 + 64 * x3 +
4 * x1 // 13), tmp10 & xmask, eviction_policy='evict_last', other=0.0)
tmp12 = tmp11 + tmp7
tmp13 = 1 + 4 * x2 // 13
tmp14 = tmp13 < tmp1
tmp15 = tmp14 & tmp5
tmp16 = tl.load(in_ptr0 + (4 + 4 * (4 * x2 // 13) + 16 * x0 + 64 * x3 +
4 * x1 // 13), tmp15 & xmask, eviction_policy='evict_last', other=0.0)
tmp17 = tmp16 + tmp12
tmp18 = tmp14 & tmp9
tmp19 = tl.load(in_ptr0 + (5 + 4 * (4 * x2 // 13) + 16 * x0 + 64 * x3 +
4 * x1 // 13), tmp18 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tmp19 + tmp17
tmp21 = 1.0
tmp22 = tl.full(tmp21.shape, 0.0, tmp21.dtype)
tmp23 = tl.where(tmp6, tmp21, tmp22)
tmp24 = tl.where(tmp10, tmp21, tmp22)
tmp25 = tmp24 + tmp23
tmp26 = tl.where(tmp15, tmp21, tmp22)
tmp27 = tmp26 + tmp25
tmp28 = tl.where(tmp18, tmp21, tmp22)
tmp29 = tmp28 + tmp27
tmp30 = tmp20 / tmp29
tl.store(out_ptr0 + x6, tmp30, xmask)
@triton.jit
def triton_poi_fused_cat_4(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4,
in_ptr5, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 56784
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 84
x1 = xindex // 84
x2 = xindex
tmp0 = x0
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 32, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (32 * x1 + x0), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tl.load(in_ptr1 + x0, tmp4 & xmask, eviction_policy='evict_last',
other=0.0)
tmp7 = tmp5 + tmp6
tmp8 = tl.full([1], 0, tl.int32)
tmp9 = triton_helpers.maximum(tmp8, tmp7)
tmp10 = tl.full(tmp9.shape, 0.0, tmp9.dtype)
tmp11 = tl.where(tmp4, tmp9, tmp10)
tmp12 = tmp0 >= tmp3
tmp13 = tl.full([1], 64, tl.int64)
tmp14 = tmp0 < tmp13
tmp15 = tmp12 & tmp14
tmp16 = tl.load(in_ptr2 + (32 * x1 + (-32 + x0)), tmp15 & xmask,
eviction_policy='evict_last', other=0.0)
tmp17 = tl.load(in_ptr3 + (-32 + x0), tmp15 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp18 = tmp16 + tmp17
tmp19 = triton_helpers.maximum(tmp8, tmp18)
tmp20 = tl.full(tmp19.shape, 0.0, tmp19.dtype)
tmp21 = tl.where(tmp15, tmp19, tmp20)
tmp22 = tmp0 >= tmp13
tl.full([1], 84, tl.int64)
tmp25 = tl.load(in_ptr4 + (20 * x1 + (-64 + x0)), tmp22 & xmask,
eviction_policy='evict_last', other=0.0)
tmp26 = tl.load(in_ptr5 + (-64 + x0), tmp22 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp27 = tmp25 + tmp26
tmp28 = triton_helpers.maximum(tmp8, tmp27)
tmp29 = tl.full(tmp28.shape, 0.0, tmp28.dtype)
tmp30 = tl.where(tmp22, tmp28, tmp29)
tmp31 = tl.where(tmp15, tmp21, tmp30)
tmp32 = tl.where(tmp4, tmp11, tmp31)
tl.store(out_ptr0 + x2, tmp32, xmask)
@triton.jit
def triton_poi_fused__to_copy_add_arange_mul_5(out_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 3.25
tmp3 = tmp1 * tmp2
tmp4 = tmp3.to(tl.int32)
tl.store(out_ptr0 + x0, tmp4, xmask)
@triton.jit
def triton_poi_fused__unsafe_index_convolution_relu_6(in_ptr0, in_ptr1,
in_ptr2, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x0 = xindex % 4
x2 = xindex // 16
x4 = xindex
tmp0 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr2 + 0)
tmp11 = tl.broadcast_to(tmp10, [XBLOCK])
tmp1 = tl.full([XBLOCK], 13, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tmp6 = tmp5 + tmp1
tmp7 = tmp5 < 0
tmp8 = tl.where(tmp7, tmp6, tmp5)
tmp9 = tl.load(in_ptr1 + (tmp8 + 13 * tmp4 + 169 * x2), xmask,
eviction_policy='evict_last')
tmp12 = tmp9 + tmp11
tmp13 = tl.full([1], 0, tl.int32)
tmp14 = triton_helpers.maximum(tmp13, tmp12)
tl.store(out_ptr0 + x4, tmp14, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_7(in_ptr0, in_ptr1,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 676
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr1 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.full([1], 0, tl.int32)
tmp5 = triton_helpers.maximum(tmp4, tmp3)
tmp6 = 0.0
tmp7 = tmp5 <= tmp6
tl.store(out_ptr0 + x0, tmp7, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_8(in_ptr0, in_ptr1,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 13520
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 20
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + x2, tmp6, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_9(in_ptr0, in_ptr1,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 21632
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 32
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (32, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_3, (32,), (1,))
assert_size_stride(primals_4, (32, 4, 5, 5), (100, 25, 5, 1))
assert_size_stride(primals_5, (32,), (1,))
assert_size_stride(primals_6, (20, 4, 7, 7), (196, 49, 7, 1))
assert_size_stride(primals_7, (20,), (1,))
assert_size_stride(primals_8, (1, 84, 1, 1), (84, 1, 1, 1))
assert_size_stride(primals_9, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((32, 4, 3, 3), (36, 1, 12, 4), torch.float32)
get_raw_stream(0)
triton_poi_fused_0[grid(128, 9)](primals_2, buf0, 128, 9, XBLOCK=16,
YBLOCK=64, num_warps=4, num_stages=1)
del primals_2
buf1 = empty_strided_cuda((32, 4, 5, 5), (100, 1, 20, 4), torch.float32
)
triton_poi_fused_1[grid(128, 25)](primals_4, buf1, 128, 25, XBLOCK=
32, YBLOCK=32, num_warps=4, num_stages=1)
del primals_4
buf2 = empty_strided_cuda((20, 4, 7, 7), (196, 1, 28, 4), torch.float32
)
triton_poi_fused_2[grid(80, 49)](primals_6, buf2, 80, 49, XBLOCK=32,
YBLOCK=32, num_warps=4, num_stages=1)
del primals_6
buf3 = empty_strided_cuda((4, 4, 13, 13), (676, 1, 52, 4), torch.
float32)
triton_poi_fused__adaptive_avg_pool2d_3[grid(2704)](primals_1, buf3,
2704, XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
buf4 = extern_kernels.convolution(buf3, buf0, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 32, 13, 13), (5408, 1, 416, 32))
buf5 = extern_kernels.convolution(buf3, buf1, stride=(1, 1),
padding=(2, 2), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf5, (4, 32, 13, 13), (5408, 1, 416, 32))
buf6 = extern_kernels.convolution(buf3, buf2, stride=(1, 1),
padding=(3, 3), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf6, (4, 20, 13, 13), (3380, 1, 260, 20))
buf7 = empty_strided_cuda((4, 84, 13, 13), (14196, 1, 1092, 84),
torch.float32)
triton_poi_fused_cat_4[grid(56784)](buf4, primals_3, buf5,
primals_5, buf6, primals_7, buf7, 56784, XBLOCK=256, num_warps=
4, num_stages=1)
buf8 = extern_kernels.convolution(buf7, primals_8, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf8, (4, 1, 13, 13), (169, 1, 13, 1))
buf9 = empty_strided_cuda((4,), (1,), torch.int64)
triton_poi_fused__to_copy_add_arange_mul_5[grid(4)](buf9, 4, XBLOCK
=4, num_warps=1, num_stages=1)
buf10 = empty_strided_cuda((4, 1, 4, 4), (16, 16, 4, 1), torch.float32)
triton_poi_fused__unsafe_index_convolution_relu_6[grid(64)](buf9,
buf8, primals_9, buf10, 64, XBLOCK=64, num_warps=1, num_stages=1)
buf11 = empty_strided_cuda((4, 1, 13, 13), (169, 1, 13, 1), torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_7[grid(676)](buf8,
primals_9, buf11, 676, XBLOCK=256, num_warps=4, num_stages=1)
del buf8
del primals_9
buf12 = empty_strided_cuda((4, 20, 13, 13), (3380, 1, 260, 20),
torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_8[grid(13520)](
buf6, primals_7, buf12, 13520, XBLOCK=256, num_warps=4,
num_stages=1)
del buf6
del primals_7
buf13 = empty_strided_cuda((4, 32, 13, 13), (5408, 1, 416, 32),
torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_9[grid(21632)](
buf5, primals_5, buf13, 21632, XBLOCK=256, num_warps=4,
num_stages=1)
del buf5
del primals_5
buf14 = empty_strided_cuda((4, 32, 13, 13), (5408, 1, 416, 32),
torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_9[grid(21632)](
buf4, primals_3, buf14, 21632, XBLOCK=256, num_warps=4,
num_stages=1)
del buf4
del primals_3
return (buf10, buf0, buf1, buf2, primals_8, buf3, buf7, buf9, buf11,
buf12, buf13, buf14)
class CRNNew(torch.nn.Module):
def __init__(self, dim):
super(CRNNew, self).__init__()
self.h_w = 13, 13
self.downsample = torch.nn.AdaptiveAvgPool2d(self.h_w)
n_filters = [32, 32, 20]
self.conv1 = torch.nn.Conv2d(dim, n_filters[0], 3, padding=1)
self.conv2 = torch.nn.Conv2d(dim, n_filters[1], 5, padding=2)
self.conv3 = torch.nn.Conv2d(dim, n_filters[2], 7, padding=3)
self.conv_accum = torch.nn.Conv2d(sum(n_filters), 1, 1)
for m in self.modules():
if isinstance(m, torch.nn.Conv2d):
torch.nn.init.kaiming_normal_(m.weight, mode='fan_out',
nonlinearity='relu')
if m.bias is not None:
torch.nn.init.zeros_(m.bias)
def forward(self, input_0):
primals_2 = self.conv1.weight
primals_3 = self.conv1.bias
primals_4 = self.conv2.weight
primals_5 = self.conv2.bias
primals_6 = self.conv3.weight
primals_7 = self.conv3.bias
primals_8 = self.conv_accum.weight
primals_9 = self.conv_accum.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
| lulor/project_vg | CRN | false | 7,154 | [
"MIT"
] | 1 | 27b0c3b3038c5a666dde516a0a265ae8ddf2059f | https://github.com/lulor/project_vg/tree/27b0c3b3038c5a666dde516a0a265ae8ddf2059f | import torch
import torch.nn.functional as F
class Model(torch.nn.Module):
def __init__(self, dim):
super().__init__()
self.h_w = 13, 13
self.downsample = torch.nn.AdaptiveAvgPool2d(self.h_w)
n_filters = [32, 32, 20]
self.conv1 = torch.nn.Conv2d(dim, n_filters[0], 3, padding=1)
self.conv2 = torch.nn.Conv2d(dim, n_filters[1], 5, padding=2)
self.conv3 = torch.nn.Conv2d(dim, n_filters[2], 7, padding=3)
self.conv_accum = torch.nn.Conv2d(sum(n_filters), 1, 1)
for m in self.modules():
if isinstance(m, torch.nn.Conv2d):
torch.nn.init.kaiming_normal_(m.weight, mode='fan_out',
nonlinearity='relu')
if m.bias is not None:
torch.nn.init.zeros_(m.bias)
def forward(self, x):
input_h_w = x.shape[2:]
x = self.downsample(x)
x1 = F.relu(self.conv1(x))
x2 = F.relu(self.conv2(x))
x3 = F.relu(self.conv3(x))
x = torch.cat((x1, x2, x3), dim=1)
x = F.relu(self.conv_accum(x))
x = F.interpolate(x, input_h_w)
assert x.shape[2:] == input_h_w
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
DeepLiftRegressor | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/pc/cpcm6y2qksezrhcucxc47wirc7wtng24a4ikidoby3abodsmpda3.py
# Topologically Sorted Source Nodes: [conv2d, x], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d => convolution
# x => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_0 = async_compile.triton('triton_poi_fused_convolution_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1048576],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 691200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 3456) % 50
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/yc/cycnvm4kffm6fth3l55jncg7vgtrc54kp7amgxb3id3mjgpmg7ew.py
# Topologically Sorted Source Nodes: [conv2d_1, x_1, x_2], Original ATen: [aten.convolution, aten.relu, aten.mean, aten.threshold_backward]
# Source node to ATen node mapping:
# conv2d_1 => convolution_1
# x_1 => relu_1
# x_2 => mean
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_1,), kwargs = {})
# %mean : [num_users=2] = call_function[target=torch.ops.aten.mean.dim](args = (%view, [2]), kwargs = {})
# %le_1 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_1, 0), kwargs = {})
triton_red_fused_convolution_mean_relu_threshold_backward_1 = async_compile.triton('triton_red_fused_convolution_mean_relu_threshold_backward_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.reduction(
size_hints=[256, 4096],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*i1', 4: 'i32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_red_fused_convolution_mean_relu_threshold_backward_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_red_fused_convolution_mean_relu_threshold_backward_1(in_out_ptr0, in_ptr0, in_ptr1, out_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr, RBLOCK : tl.constexpr):
xnumel = 200
rnumel = 2816
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rbase = tl.arange(0, RBLOCK)[None, :]
x3 = xindex
x0 = xindex % 50
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
_tmp6 = tl.full([XBLOCK, RBLOCK], 0, tl.float32)
for roffset in range(0, rnumel, RBLOCK):
rindex = roffset + rbase
rmask = rindex < rnumel
r2 = rindex
tmp0 = tl.load(in_ptr0 + (r2 + (2816*x3)), rmask & xmask, eviction_policy='evict_first', other=0.0)
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1, 1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = tl.broadcast_to(tmp4, [XBLOCK, RBLOCK])
tmp7 = _tmp6 + tmp5
_tmp6 = tl.where(rmask & xmask, tmp7, _tmp6)
tmp8 = 0.0
tmp9 = tmp4 <= tmp8
tl.store(out_ptr0 + (r2 + (2816*x3)), tmp9, rmask & xmask)
tmp6 = tl.sum(_tmp6, 1)[:, None]
tmp10 = 2816.0
tmp11 = tmp6 / tmp10
tl.debug_barrier()
tl.store(in_out_ptr0 + (x3), tmp11, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/h2/ch2bj6a3qcae5ehxm5i5i5fvfnshjpu6uyzbg3mozketmq5np3ka.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# x_3 => relu_2
# Graph fragment:
# %add_tensor : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default, %primals_7), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor,), kwargs = {})
triton_poi_fused_relu_2 = async_compile.triton('triton_poi_fused_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 50
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (50, 4, 1, 11), (44, 11, 11, 1))
assert_size_stride(primals_2, (50, ), (1, ))
assert_size_stride(primals_3, (4, 4, 64, 64), (16384, 4096, 64, 1))
assert_size_stride(primals_4, (50, 50, 1, 11), (550, 11, 11, 1))
assert_size_stride(primals_5, (50, ), (1, ))
assert_size_stride(primals_6, (50, 50), (50, 1))
assert_size_stride(primals_7, (50, ), (1, ))
assert_size_stride(primals_8, (5, 50), (50, 1))
assert_size_stride(primals_9, (5, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 50, 64, 54), (172800, 3456, 54, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [conv2d, x], Original ATen: [aten.convolution, aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_relu_0.run(buf1, primals_2, 691200, grid=grid(691200), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 50, 64, 44), (140800, 2816, 44, 1))
buf3 = empty_strided_cuda((4, 50), (50, 1), torch.float32)
buf8 = empty_strided_cuda((4, 50, 64, 44), (140800, 2816, 44, 1), torch.bool)
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [conv2d_1, x_1, x_2], Original ATen: [aten.convolution, aten.relu, aten.mean, aten.threshold_backward]
triton_red_fused_convolution_mean_relu_threshold_backward_1.run(buf4, buf2, primals_5, buf8, 200, 2816, grid=grid(200), stream=stream0)
del buf2
del primals_5
buf5 = empty_strided_cuda((4, 50), (50, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf4, reinterpret_tensor(primals_6, (50, 50), (1, 50), 0), out=buf5)
buf6 = buf5; del buf5 # reuse
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.relu]
triton_poi_fused_relu_2.run(buf6, primals_7, 200, grid=grid(200), stream=stream0)
del primals_7
buf7 = empty_strided_cuda((4, 5), (5, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_9, buf6, reinterpret_tensor(primals_8, (50, 5), (1, 50), 0), alpha=1, beta=1, out=buf7)
del primals_9
return (buf7, primals_1, primals_3, primals_4, buf1, buf4, buf6, primals_8, primals_6, buf8, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((50, 4, 1, 11), (44, 11, 11, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((50, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 64, 64), (16384, 4096, 64, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((50, 50, 1, 11), (550, 11, 11, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((50, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((50, 50), (50, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((50, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((5, 50), (50, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((5, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class DeepLiftRegressor(nn.Module):
def __init__(self):
super(DeepLiftRegressor, self).__init__()
self.conv1 = nn.Conv2d(in_channels=4, out_channels=50, kernel_size=
(1, 11))
self.conv2 = nn.Conv2d(in_channels=50, out_channels=50, kernel_size
=(1, 11))
self.fc1 = nn.Linear(50, 50)
self.dropout1 = nn.Dropout2d(p=0.5)
self.fc2 = nn.Linear(50, 5)
def forward(self, input):
x = F.relu(self.conv1(input))
x = F.relu(self.conv2(x))
x = torch.mean(x.view(x.size(0), x.size(1), -1), dim=2)
x = F.relu(self.fc1(x))
x = self.dropout1(x)
x = self.fc2(x)
return x
def num_flat_features(self, x):
size = x.size()[1:]
num_features = 1
for s in size:
num_features *= s
return num_features
def get_inputs():
return [torch.rand([4, 4, 64, 64])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 691200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 3456 % 50
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_red_fused_convolution_mean_relu_threshold_backward_1(in_out_ptr0,
in_ptr0, in_ptr1, out_ptr0, xnumel, rnumel, XBLOCK: tl.constexpr,
RBLOCK: tl.constexpr):
xnumel = 200
rnumel = 2816
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rbase = tl.arange(0, RBLOCK)[None, :]
x3 = xindex
x0 = xindex % 50
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
_tmp6 = tl.full([XBLOCK, RBLOCK], 0, tl.float32)
for roffset in range(0, rnumel, RBLOCK):
rindex = roffset + rbase
rmask = rindex < rnumel
r2 = rindex
tmp0 = tl.load(in_ptr0 + (r2 + 2816 * x3), rmask & xmask,
eviction_policy='evict_first', other=0.0)
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1, 1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = tl.broadcast_to(tmp4, [XBLOCK, RBLOCK])
tmp7 = _tmp6 + tmp5
_tmp6 = tl.where(rmask & xmask, tmp7, _tmp6)
tmp8 = 0.0
tmp9 = tmp4 <= tmp8
tl.store(out_ptr0 + (r2 + 2816 * x3), tmp9, rmask & xmask)
tmp6 = tl.sum(_tmp6, 1)[:, None]
tmp10 = 2816.0
tmp11 = tmp6 / tmp10
tl.debug_barrier()
tl.store(in_out_ptr0 + x3, tmp11, xmask)
@triton.jit
def triton_poi_fused_relu_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 50
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (50, 4, 1, 11), (44, 11, 11, 1))
assert_size_stride(primals_2, (50,), (1,))
assert_size_stride(primals_3, (4, 4, 64, 64), (16384, 4096, 64, 1))
assert_size_stride(primals_4, (50, 50, 1, 11), (550, 11, 11, 1))
assert_size_stride(primals_5, (50,), (1,))
assert_size_stride(primals_6, (50, 50), (50, 1))
assert_size_stride(primals_7, (50,), (1,))
assert_size_stride(primals_8, (5, 50), (50, 1))
assert_size_stride(primals_9, (5,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,
1), padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 50, 64, 54), (172800, 3456, 54, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_relu_0[grid(691200)](buf1, primals_2,
691200, XBLOCK=512, num_warps=8, num_stages=1)
del primals_2
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 50, 64, 44), (140800, 2816, 44, 1))
buf3 = empty_strided_cuda((4, 50), (50, 1), torch.float32)
buf8 = empty_strided_cuda((4, 50, 64, 44), (140800, 2816, 44, 1),
torch.bool)
buf4 = buf3
del buf3
triton_red_fused_convolution_mean_relu_threshold_backward_1[grid(200)](
buf4, buf2, primals_5, buf8, 200, 2816, XBLOCK=1, RBLOCK=2048,
num_warps=16, num_stages=1)
del buf2
del primals_5
buf5 = empty_strided_cuda((4, 50), (50, 1), torch.float32)
extern_kernels.mm(buf4, reinterpret_tensor(primals_6, (50, 50), (1,
50), 0), out=buf5)
buf6 = buf5
del buf5
triton_poi_fused_relu_2[grid(200)](buf6, primals_7, 200, XBLOCK=128,
num_warps=4, num_stages=1)
del primals_7
buf7 = empty_strided_cuda((4, 5), (5, 1), torch.float32)
extern_kernels.addmm(primals_9, buf6, reinterpret_tensor(primals_8,
(50, 5), (1, 50), 0), alpha=1, beta=1, out=buf7)
del primals_9
return (buf7, primals_1, primals_3, primals_4, buf1, buf4, buf6,
primals_8, primals_6, buf8)
class DeepLiftRegressorNew(nn.Module):
def __init__(self):
super(DeepLiftRegressorNew, self).__init__()
self.conv1 = nn.Conv2d(in_channels=4, out_channels=50, kernel_size=
(1, 11))
self.conv2 = nn.Conv2d(in_channels=50, out_channels=50, kernel_size
=(1, 11))
self.fc1 = nn.Linear(50, 50)
self.dropout1 = nn.Dropout2d(p=0.5)
self.fc2 = nn.Linear(50, 5)
def num_flat_features(self, x):
size = x.size()[1:]
num_features = 1
for s in size:
num_features *= s
return num_features
def forward(self, input_0):
primals_1 = self.conv1.weight
primals_2 = self.conv1.bias
primals_4 = self.conv2.weight
primals_5 = self.conv2.bias
primals_6 = self.fc1.weight
primals_7 = self.fc1.bias
primals_8 = self.fc2.weight
primals_9 = self.fc2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
| lzamparo/SeqDemote | DeepLiftRegressor | false | 7,155 | [
"MIT"
] | 1 | 3eaf18e88c9dc6a3d1a69444ecdba9f9b5d9682a | https://github.com/lzamparo/SeqDemote/tree/3eaf18e88c9dc6a3d1a69444ecdba9f9b5d9682a | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self):
super().__init__()
self.conv1 = nn.Conv2d(in_channels=4, out_channels=50, kernel_size=
(1, 11))
self.conv2 = nn.Conv2d(in_channels=50, out_channels=50, kernel_size
=(1, 11))
self.fc1 = nn.Linear(50, 50)
self.dropout1 = nn.Dropout2d(p=0.5)
self.fc2 = nn.Linear(50, 5)
def forward(self, input):
x = F.relu(self.conv1(input))
x = F.relu(self.conv2(x))
x = torch.mean(x.view(x.size(0), x.size(1), -1), dim=2)
x = F.relu(self.fc1(x))
x = self.dropout1(x)
x = self.fc2(x)
return x
def num_flat_features(self, x):
size = x.size()[1:]
num_features = 1
for s in size:
num_features *= s
return num_features
def get_inputs():
return [torch.rand([4, 4, 64, 64])]
def get_init_inputs():
return []
|
MultiHeadAttention | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/rh/crhy6nilvaajphuuoyup37xl4ncuiyrcb3fnt5aboux6wyvcg7ie.py
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 16], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, in_ptr1, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (64*y1)), xmask & ymask)
tmp1 = tl.load(in_ptr1 + (y0), ymask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2 + (16*y3)), tmp2, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/oi/coiccv2naqo5guipwtokaoohzz54bpvgya5fumrfprestitumxks.py
# Topologically Sorted Source Nodes: [scores, neg, pow_1, mul, add, scores_1], Original ATen: [aten.div, aten.neg, aten.pow, aten.mul, aten.add, aten.reciprocal]
# Source node to ATen node mapping:
# add => add
# mul => mul
# neg => neg
# pow_1 => pow_1
# scores => div
# scores_1 => mul_1, reciprocal
# Graph fragment:
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%view_11, 1.0), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%div,), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Scalar](args = (2.718281828459045, %neg), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%pow_1, 20), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, 1), kwargs = {})
# %reciprocal : [num_users=1] = call_function[target=torch.ops.aten.reciprocal.default](args = (%add,), kwargs = {})
# %mul_1 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%reciprocal, 1), kwargs = {})
triton_poi_fused_add_div_mul_neg_pow_reciprocal_1 = async_compile.triton('triton_poi_fused_add_div_mul_neg_pow_reciprocal_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mul_neg_pow_reciprocal_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mul_neg_pow_reciprocal_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 4096
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), None)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp3 = -tmp2
tmp4 = 2.718281828459045
tmp5 = libdevice.pow(tmp4, tmp3)
tmp6 = 20.0
tmp7 = tmp5 * tmp6
tmp8 = tmp7 + tmp1
tmp9 = tl.full([1], 1, tl.int32)
tmp10 = tmp9 / tmp8
tmp11 = tmp10 * tmp1
tl.store(out_ptr0 + (x0), tmp11, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/mz/cmzlu2lip25blpsdqeby7ek5757op6xw3pdkxbdediou5szw32tx.py
# Topologically Sorted Source Nodes: [contiguous], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# contiguous => clone_3
# Graph fragment:
# %clone_3 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%permute_7,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_2 = async_compile.triton('triton_poi_fused_clone_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 16
y1 = (yindex // 16)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (16*x2) + (64*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (4, ), (1, ))
assert_size_stride(primals_9, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_10, (4, 4), (4, 1))
assert_size_stride(primals_11, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_4, (64, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf0)
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf1)
del primals_5
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_9, (64, 4), (4, 1), 0), reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf2)
del primals_7
buf3 = empty_strided_cuda((4, 4, 16, 1), (64, 16, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(buf1, primals_6, buf3, 16, 16, grid=grid(16, 16), stream=stream0)
del primals_6
buf4 = reinterpret_tensor(buf1, (4, 4, 1, 16), (64, 16, 16, 1), 0); del buf1 # reuse
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(buf0, primals_3, buf4, 16, 16, grid=grid(16, 16), stream=stream0)
del primals_3
buf5 = empty_strided_cuda((16, 16, 16), (256, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 16, 1), (16, 1, 0), 0), reinterpret_tensor(buf4, (16, 1, 16), (16, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 16, 16), (1024, 256, 16, 1), torch.float32)
# Topologically Sorted Source Nodes: [scores, neg, pow_1, mul, add, scores_1], Original ATen: [aten.div, aten.neg, aten.pow, aten.mul, aten.add, aten.reciprocal]
triton_poi_fused_add_div_mul_neg_pow_reciprocal_1.run(buf5, buf6, 4096, grid=grid(4096), stream=stream0)
buf7 = reinterpret_tensor(buf0, (4, 4, 16, 1), (64, 16, 1, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
triton_poi_fused_clone_0.run(buf2, primals_8, buf7, 16, 16, grid=grid(16, 16), stream=stream0)
del primals_8
buf8 = reinterpret_tensor(buf2, (16, 16, 1), (16, 1, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf6, (16, 16, 16), (256, 16, 1), 0), reinterpret_tensor(buf7, (16, 16, 1), (16, 1, 0), 0), out=buf8)
buf9 = empty_strided_cuda((4, 16, 4, 1), (64, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [contiguous], Original ATen: [aten.clone]
triton_poi_fused_clone_2.run(buf8, buf9, 64, 4, grid=grid(64, 4), stream=stream0)
buf10 = reinterpret_tensor(buf8, (64, 4), (4, 1), 0); del buf8 # reuse
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_11, reinterpret_tensor(buf9, (64, 4), (4, 1), 0), reinterpret_tensor(primals_10, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf10)
del primals_11
return (reinterpret_tensor(buf10, (4, 16, 4), (64, 4, 1), 0), buf6, reinterpret_tensor(primals_4, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (64, 4), (4, 1), 0), reinterpret_tensor(primals_9, (64, 4), (4, 1), 0), buf5, reinterpret_tensor(buf9, (64, 4), (4, 1), 0), primals_10, reinterpret_tensor(buf6, (16, 16, 16), (256, 1, 16), 0), reinterpret_tensor(buf7, (16, 1, 16), (16, 1, 1), 0), reinterpret_tensor(buf3, (16, 1, 16), (16, 1, 1), 0), reinterpret_tensor(buf4, (16, 16, 1), (16, 1, 16), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import numpy as np
import torch.nn as nn
def logistic(x, c=1, a=20, b=np.e):
return c / (1 + a * b ** -x)
def attention(q, k, v, d_k, mask=None, dropout=None):
scores = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(d_k)
if mask is not None:
mask = mask.unsqueeze(1)
scores = scores.masked_fill(mask == 0, -1000000000.0)
scores = logistic(scores)
output = torch.matmul(scores, v)
return output, scores
class MultiHeadAttention(nn.Module):
def __init__(self, heads, d_model, dropout=0.1, nheads=200,
share_params=True):
super().__init__()
self.d_model = d_model
self.d_k = d_model // heads
self.h = heads
if share_params is False:
self.q_linear = simple_projection_3d(d_model, d_model, nheads)
self.v_linear = simple_projection_3d(d_model, d_model, nheads)
self.k_linear = simple_projection_3d(d_model, d_model, nheads)
if share_params is True:
self.q_linear = nn.Linear(d_model, d_model)
self.v_linear = nn.Linear(d_model, d_model)
self.k_linear = nn.Linear(d_model, d_model)
self.dropout = nn.Dropout(dropout)
self.out = nn.Linear(d_model, d_model)
def forward(self, q, k, v, mask=None):
bs = q.size(0)
k = self.k_linear(k).view(bs, -1, self.h, self.d_k)
q = self.q_linear(q).view(bs, -1, self.h, self.d_k)
v = self.v_linear(v).view(bs, -1, self.h, self.d_k)
k = k.transpose(1, 2)
q = q.transpose(1, 2)
v = v.transpose(1, 2)
scores, w = attention(q, k, v, self.d_k, mask, self.dropout)
concat = scores.transpose(1, 2).contiguous().view(bs, -1, self.d_model)
output = self.out(concat)
return output, w
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [[], {'heads': 4, 'd_model': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import math
import numpy as np
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, in_ptr1, out_ptr0, ynumel, xnumel,
YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 64 * y1), xmask & ymask)
tmp1 = tl.load(in_ptr1 + y0, ymask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2 + 16 * y3), tmp2, xmask & ymask)
@triton.jit
def triton_poi_fused_add_div_mul_neg_pow_reciprocal_1(in_ptr0, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, None)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp3 = -tmp2
tmp4 = 2.718281828459045
tmp5 = libdevice.pow(tmp4, tmp3)
tmp6 = 20.0
tmp7 = tmp5 * tmp6
tmp8 = tmp7 + tmp1
tmp9 = tl.full([1], 1, tl.int32)
tmp10 = tmp9 / tmp8
tmp11 = tmp10 * tmp1
tl.store(out_ptr0 + x0, tmp11, None)
@triton.jit
def triton_poi_fused_clone_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 16
y1 = yindex // 16
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 16 * x2 + 64 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4,), (1,))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (4,), (1,))
assert_size_stride(primals_9, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_10, (4, 4), (4, 1))
assert_size_stride(primals_11, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_4, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf0)
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf1)
del primals_5
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_9, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf2)
del primals_7
buf3 = empty_strided_cuda((4, 4, 16, 1), (64, 16, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(16, 16)](buf1, primals_6, buf3, 16,
16, XBLOCK=16, YBLOCK=16, num_warps=4, num_stages=1)
del primals_6
buf4 = reinterpret_tensor(buf1, (4, 4, 1, 16), (64, 16, 16, 1), 0)
del buf1
triton_poi_fused_clone_0[grid(16, 16)](buf0, primals_3, buf4, 16,
16, XBLOCK=16, YBLOCK=16, num_warps=4, num_stages=1)
del primals_3
buf5 = empty_strided_cuda((16, 16, 16), (256, 16, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 16, 1), (16, 1, 0),
0), reinterpret_tensor(buf4, (16, 1, 16), (16, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 16, 16), (1024, 256, 16, 1), torch
.float32)
triton_poi_fused_add_div_mul_neg_pow_reciprocal_1[grid(4096)](buf5,
buf6, 4096, XBLOCK=128, num_warps=4, num_stages=1)
buf7 = reinterpret_tensor(buf0, (4, 4, 16, 1), (64, 16, 1, 1), 0)
del buf0
triton_poi_fused_clone_0[grid(16, 16)](buf2, primals_8, buf7, 16,
16, XBLOCK=16, YBLOCK=16, num_warps=4, num_stages=1)
del primals_8
buf8 = reinterpret_tensor(buf2, (16, 16, 1), (16, 1, 1), 0)
del buf2
extern_kernels.bmm(reinterpret_tensor(buf6, (16, 16, 16), (256, 16,
1), 0), reinterpret_tensor(buf7, (16, 16, 1), (16, 1, 0), 0),
out=buf8)
buf9 = empty_strided_cuda((4, 16, 4, 1), (64, 4, 1, 1), torch.float32)
triton_poi_fused_clone_2[grid(64, 4)](buf8, buf9, 64, 4, XBLOCK=4,
YBLOCK=32, num_warps=4, num_stages=1)
buf10 = reinterpret_tensor(buf8, (64, 4), (4, 1), 0)
del buf8
extern_kernels.addmm(primals_11, reinterpret_tensor(buf9, (64, 4),
(4, 1), 0), reinterpret_tensor(primals_10, (4, 4), (1, 4), 0),
alpha=1, beta=1, out=buf10)
del primals_11
return reinterpret_tensor(buf10, (4, 16, 4), (64, 4, 1), 0
), buf6, reinterpret_tensor(primals_4, (64, 4), (4, 1), 0
), reinterpret_tensor(primals_1, (64, 4), (4, 1), 0
), reinterpret_tensor(primals_9, (64, 4), (4, 1), 0
), buf5, reinterpret_tensor(buf9, (64, 4), (4, 1), 0
), primals_10, reinterpret_tensor(buf6, (16, 16, 16), (256, 1, 16), 0
), reinterpret_tensor(buf7, (16, 1, 16), (16, 1, 1), 0
), reinterpret_tensor(buf3, (16, 1, 16), (16, 1, 1), 0
), reinterpret_tensor(buf4, (16, 16, 1), (16, 1, 16), 0)
def logistic(x, c=1, a=20, b=np.e):
return c / (1 + a * b ** -x)
def attention(q, k, v, d_k, mask=None, dropout=None):
scores = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(d_k)
if mask is not None:
mask = mask.unsqueeze(1)
scores = scores.masked_fill(mask == 0, -1000000000.0)
scores = logistic(scores)
output = torch.matmul(scores, v)
return output, scores
class MultiHeadAttentionNew(nn.Module):
def __init__(self, heads, d_model, dropout=0.1, nheads=200,
share_params=True):
super().__init__()
self.d_model = d_model
self.d_k = d_model // heads
self.h = heads
if share_params is False:
self.q_linear = simple_projection_3d(d_model, d_model, nheads)
self.v_linear = simple_projection_3d(d_model, d_model, nheads)
self.k_linear = simple_projection_3d(d_model, d_model, nheads)
if share_params is True:
self.q_linear = nn.Linear(d_model, d_model)
self.v_linear = nn.Linear(d_model, d_model)
self.k_linear = nn.Linear(d_model, d_model)
self.dropout = nn.Dropout(dropout)
self.out = nn.Linear(d_model, d_model)
def forward(self, input_0, input_1, input_2):
primals_2 = self.q_linear.weight
primals_3 = self.q_linear.bias
primals_5 = self.v_linear.weight
primals_6 = self.v_linear.bias
primals_7 = self.k_linear.weight
primals_8 = self.k_linear.bias
primals_10 = self.out.weight
primals_11 = self.out.bias
primals_1 = input_0
primals_4 = input_1
primals_9 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11])
return output[0], output[1]
| lysecret2/explainability-simulation | MultiHeadAttention | false | 7,156 | [
"MIT"
] | 1 | e558f6f527ac2ff66f00fcb37aeeaf404c32ff66 | https://github.com/lysecret2/explainability-simulation/tree/e558f6f527ac2ff66f00fcb37aeeaf404c32ff66 | import math
import torch
import numpy as np
import torch.nn as nn
def logistic(x, c=1, a=20, b=np.e):
return c / (1 + a * b ** -x)
def attention(q, k, v, d_k, mask=None, dropout=None):
scores = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(d_k)
if mask is not None:
mask = mask.unsqueeze(1)
scores = scores.masked_fill(mask == 0, -1000000000.0)
scores = logistic(scores)
output = torch.matmul(scores, v)
return output, scores
class Model(nn.Module):
def __init__(self, heads, d_model, dropout=0.1, nheads=200,
share_params=True):
super().__init__()
self.d_model = d_model
self.d_k = d_model // heads
self.h = heads
if share_params is False:
self.q_linear = simple_projection_3d(d_model, d_model, nheads)
self.v_linear = simple_projection_3d(d_model, d_model, nheads)
self.k_linear = simple_projection_3d(d_model, d_model, nheads)
if share_params is True:
self.q_linear = nn.Linear(d_model, d_model)
self.v_linear = nn.Linear(d_model, d_model)
self.k_linear = nn.Linear(d_model, d_model)
self.dropout = nn.Dropout(dropout)
self.out = nn.Linear(d_model, d_model)
def forward(self, q, k, v, mask=None):
bs = q.size(0)
k = self.k_linear(k).view(bs, -1, self.h, self.d_k)
q = self.q_linear(q).view(bs, -1, self.h, self.d_k)
v = self.v_linear(v).view(bs, -1, self.h, self.d_k)
k = k.transpose(1, 2)
q = q.transpose(1, 2)
v = v.transpose(1, 2)
scores, w = attention(q, k, v, self.d_k, mask, self.dropout)
concat = scores.transpose(1, 2).contiguous().view(bs, -1, self.d_model)
output = self.out(concat)
return output, w
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
L2Norm | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/w2/cw2qyt2fur3m5wcjxq34xum55a7h7iifg56c5zbipask4jcl4kpf.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.div]
# Source node to ATen node mapping:
# x => div
# Graph fragment:
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg0_1, %unsqueeze), kwargs = {})
triton_poi_fused_div_0 = async_compile.triton('triton_poi_fused_div_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = 1e-10
tmp13 = tmp11 + tmp12
tmp14 = libdevice.sqrt(tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + (x2), tmp15, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_div_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class L2Norm(nn.Module):
"""l2-normalization as layer. """
def __init__(self, *, eps: float=1e-10) ->None:
super().__init__()
self.eps = eps
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
norm = torch.sqrt(torch.sum(x * x, dim=-1) + self.eps)
x = x / norm.unsqueeze(-1)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_div_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = 1e-10
tmp13 = tmp11 + tmp12
tmp14 = libdevice.sqrt(tmp13)
tmp15 = tmp0 / tmp14
tl.store(out_ptr0 + x2, tmp15, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_div_0[grid(256)](arg0_1, buf0, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del arg0_1
return buf0,
class L2NormNew(nn.Module):
"""l2-normalization as layer. """
def __init__(self, *, eps: float=1e-10) ->None:
super().__init__()
self.eps = eps
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| manyids2/mkd_pytorch | L2Norm | false | 7,157 | [
"MIT"
] | 1 | fb97c4285f93f38371b2aac904a133f970be247e | https://github.com/manyids2/mkd_pytorch/tree/fb97c4285f93f38371b2aac904a133f970be247e | import torch
import torch.nn as nn
class Model(nn.Module):
"""l2-normalization as layer. """
def __init__(self, *, eps: float=1e-10) ->None:
super().__init__()
self.eps = eps
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
norm = torch.sqrt(torch.sum(x * x, dim=-1) + self.eps)
x = x / norm.unsqueeze(-1)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
MLP_VAE | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/3q/c3qwr2d2rrpjzvnddomnmdy6cwva4hjlvrn2y5epemk4ak3k2m6c.py
# Topologically Sorted Source Nodes: [h1], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# h1 => relu
# Graph fragment:
# %add_tensor_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default_1, %primals_3), kwargs = {})
# %relu : [num_users=3] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor_1,), kwargs = {})
triton_poi_fused_relu_0 = async_compile.triton('triton_poi_fused_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 400
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/3a/c3altqui4bw2inrmbr26ko7caryay6zlxm2w75x7ol5kkzotz753.py
# Topologically Sorted Source Nodes: [mul, std, mul_1, z], Original ATen: [aten.mul, aten.exp, aten.add]
# Source node to ATen node mapping:
# mul => mul
# mul_1 => mul_1
# std => exp
# z => add
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%addmm_2, 0.5), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%mul,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%randn, %exp), kwargs = {})
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%addmm_1, %mul_1), kwargs = {})
triton_poi_fused_add_exp_mul_1 = async_compile.triton('triton_poi_fused_add_exp_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_exp_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_exp_mul_1(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask)
tmp2 = tl.load(in_ptr2 + (x0), xmask)
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp1 * tmp5
tmp7 = tmp0 + tmp6
tl.store(out_ptr0 + (x0), tmp7, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11 = args
args.clear()
assert_size_stride(primals_1, (4, 1024), (1024, 1))
assert_size_stride(primals_2, (400, 1024), (1024, 1))
assert_size_stride(primals_3, (400, ), (1, ))
assert_size_stride(primals_4, (4, 400), (400, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, 400), (400, 1))
assert_size_stride(primals_7, (4, ), (1, ))
assert_size_stride(primals_8, (400, 4), (4, 1))
assert_size_stride(primals_9, (400, ), (1, ))
assert_size_stride(primals_10, (1024, 400), (400, 1))
assert_size_stride(primals_11, (1024, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(primals_1, reinterpret_tensor(primals_2, (1024, 400), (1, 1024), 0), out=buf0)
del primals_2
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [h1], Original ATen: [aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_0.run(buf1, primals_3, 1600, grid=grid(1600), stream=stream0)
del primals_3
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mu], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, buf1, reinterpret_tensor(primals_4, (400, 4), (1, 400), 0), alpha=1, beta=1, out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [logvar], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_7, buf1, reinterpret_tensor(primals_6, (400, 4), (1, 400), 0), alpha=1, beta=1, out=buf3)
del primals_7
# Topologically Sorted Source Nodes: [eps], Original ATen: [aten.randn_like]
buf4 = torch.ops.aten.randn.default([4, 4], dtype=torch.float32, device=device(type='cuda', index=0), pin_memory=False)
buf5 = buf4
del buf4
buf6 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, std, mul_1, z], Original ATen: [aten.mul, aten.exp, aten.add]
triton_poi_fused_add_exp_mul_1.run(buf2, buf5, buf3, buf6, 16, grid=grid(16), stream=stream0)
buf7 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf6, reinterpret_tensor(primals_8, (4, 400), (1, 4), 0), out=buf7)
buf8 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [h3], Original ATen: [aten.relu]
triton_poi_fused_relu_0.run(buf8, primals_9, 1600, grid=grid(1600), stream=stream0)
del primals_9
buf9 = empty_strided_cuda((4, 1024), (1024, 1), torch.float32)
# Topologically Sorted Source Nodes: [reconstruction_x], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_11, buf8, reinterpret_tensor(primals_10, (400, 1024), (1, 400), 0), alpha=1, beta=1, out=buf9)
del primals_11
return (buf9, buf2, buf3, primals_1, buf1, buf3, buf5, buf6, buf8, primals_10, primals_8, primals_6, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 1024), (1024, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((400, 1024), (1024, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((400, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 400), (400, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 400), (400, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((400, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((400, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((1024, 400), (400, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((1024, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
class MLP_VAE(nn.Module):
def __init__(self, ZDIMS):
super().__init__()
self.z_dims = ZDIMS
self.fc1 = nn.Linear(1024, 400)
self.relu = nn.ReLU()
self.fc21 = nn.Linear(400, ZDIMS)
self.fc22 = nn.Linear(400, ZDIMS)
self.fc3 = nn.Linear(ZDIMS, 400)
self.fc4 = nn.Linear(400, 1024)
def encoder(self, x):
"""
Input vector x --> fully connected 1 --> RELU --> fully connected 21, fully connected 22
Parameters
----------
x: [batch size, 784], batch size number of digits of 28x28 pixels each
Returns
-------
(mu, logvar): ZDIMS mean units one for each latent dimension, ZDIMS variance units one for each
latent dimension
"""
batch_size = x.shape[0]
x = x.view(batch_size, -1)
h1 = self.relu(self.fc1(x))
mu = self.fc21(h1)
logvar = self.fc22(h1)
return mu, logvar
def reparameterize(self, mu, logvar):
std = torch.exp(0.5 * logvar)
eps = torch.randn_like(std)
return mu + eps * std
def decoder(self, z):
h3 = self.relu(self.fc3(z))
return self.fc4(h3)
def forward(self, x):
mu, logvar = self.encoder(x.view(-1, 1024))
z = self.reparameterize(mu, logvar)
reconstruction_x = self.decoder(z)
return reconstruction_x, mu, logvar
def sample(self, n):
with torch.no_grad():
z = torch.randn(n, self.z_dims)
z = z
samples = self.decoder(z)
samples = torch.clamp(samples, 0, 1)
return samples.cpu().numpy()
def get_inputs():
return [torch.rand([4, 1024])]
def get_init_inputs():
return [[], {'ZDIMS': 4}]
| import torch
from torch import device
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 1600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 400
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_add_exp_mul_1(in_ptr0, in_ptr1, in_ptr2, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask)
tmp2 = tl.load(in_ptr2 + x0, xmask)
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp1 * tmp5
tmp7 = tmp0 + tmp6
tl.store(out_ptr0 + x0, tmp7, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11) = args
args.clear()
assert_size_stride(primals_1, (4, 1024), (1024, 1))
assert_size_stride(primals_2, (400, 1024), (1024, 1))
assert_size_stride(primals_3, (400,), (1,))
assert_size_stride(primals_4, (4, 400), (400, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4, 400), (400, 1))
assert_size_stride(primals_7, (4,), (1,))
assert_size_stride(primals_8, (400, 4), (4, 1))
assert_size_stride(primals_9, (400,), (1,))
assert_size_stride(primals_10, (1024, 400), (400, 1))
assert_size_stride(primals_11, (1024,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
extern_kernels.mm(primals_1, reinterpret_tensor(primals_2, (1024,
400), (1, 1024), 0), out=buf0)
del primals_2
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_relu_0[grid(1600)](buf1, primals_3, 1600, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_3
buf2 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, buf1, reinterpret_tensor(primals_4,
(400, 4), (1, 400), 0), alpha=1, beta=1, out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_7, buf1, reinterpret_tensor(primals_6,
(400, 4), (1, 400), 0), alpha=1, beta=1, out=buf3)
del primals_7
buf4 = torch.ops.aten.randn.default([4, 4], dtype=torch.float32,
device=device(type='cuda', index=0), pin_memory=False)
buf5 = buf4
del buf4
buf6 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
triton_poi_fused_add_exp_mul_1[grid(16)](buf2, buf5, buf3, buf6, 16,
XBLOCK=16, num_warps=1, num_stages=1)
buf7 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
extern_kernels.mm(buf6, reinterpret_tensor(primals_8, (4, 400), (1,
4), 0), out=buf7)
buf8 = buf7
del buf7
triton_poi_fused_relu_0[grid(1600)](buf8, primals_9, 1600, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_9
buf9 = empty_strided_cuda((4, 1024), (1024, 1), torch.float32)
extern_kernels.addmm(primals_11, buf8, reinterpret_tensor(
primals_10, (400, 1024), (1, 400), 0), alpha=1, beta=1, out=buf9)
del primals_11
return (buf9, buf2, buf3, primals_1, buf1, buf3, buf5, buf6, buf8,
primals_10, primals_8, primals_6, primals_4)
class MLP_VAENew(nn.Module):
def __init__(self, ZDIMS):
super().__init__()
self.z_dims = ZDIMS
self.fc1 = nn.Linear(1024, 400)
self.relu = nn.ReLU()
self.fc21 = nn.Linear(400, ZDIMS)
self.fc22 = nn.Linear(400, ZDIMS)
self.fc3 = nn.Linear(ZDIMS, 400)
self.fc4 = nn.Linear(400, 1024)
def encoder(self, x):
"""
Input vector x --> fully connected 1 --> RELU --> fully connected 21, fully connected 22
Parameters
----------
x: [batch size, 784], batch size number of digits of 28x28 pixels each
Returns
-------
(mu, logvar): ZDIMS mean units one for each latent dimension, ZDIMS variance units one for each
latent dimension
"""
batch_size = x.shape[0]
x = x.view(batch_size, -1)
h1 = self.relu(self.fc1(x))
mu = self.fc21(h1)
logvar = self.fc22(h1)
return mu, logvar
def reparameterize(self, mu, logvar):
std = torch.exp(0.5 * logvar)
eps = torch.randn_like(std)
return mu + eps * std
def decoder(self, z):
h3 = self.relu(self.fc3(z))
return self.fc4(h3)
def sample(self, n):
with torch.no_grad():
z = torch.randn(n, self.z_dims)
z = z
samples = self.decoder(z)
samples = torch.clamp(samples, 0, 1)
return samples.cpu().numpy()
def forward(self, input_0):
primals_2 = self.fc1.weight
primals_3 = self.fc1.bias
primals_4 = self.fc21.weight
primals_5 = self.fc21.bias
primals_6 = self.fc22.weight
primals_7 = self.fc22.bias
primals_8 = self.fc3.weight
primals_9 = self.fc3.bias
primals_10 = self.fc4.weight
primals_11 = self.fc4.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11])
return output[0], output[1], output[2]
| manuelladron/artistic_style_robotic_painting | MLP_VAE | false | 7,158 | [
"MIT"
] | 1 | 3769fc470bb4f69d2ea77d2713e4eb9bf0eaa4e9 | https://github.com/manuelladron/artistic_style_robotic_painting/tree/3769fc470bb4f69d2ea77d2713e4eb9bf0eaa4e9 | import torch
from torch import nn
class Model(nn.Module):
def __init__(self, ZDIMS):
super().__init__()
self.z_dims = ZDIMS
self.fc1 = nn.Linear(1024, 400)
self.relu = nn.ReLU()
self.fc21 = nn.Linear(400, ZDIMS)
self.fc22 = nn.Linear(400, ZDIMS)
self.fc3 = nn.Linear(ZDIMS, 400)
self.fc4 = nn.Linear(400, 1024)
def encoder(self, x):
"""
Input vector x --> fully connected 1 --> RELU --> fully connected 21, fully connected 22
Parameters
----------
x: [batch size, 784], batch size number of digits of 28x28 pixels each
Returns
-------
(mu, logvar): ZDIMS mean units one for each latent dimension, ZDIMS variance units one for each
latent dimension
"""
batch_size = x.shape[0]
x = x.view(batch_size, -1)
h1 = self.relu(self.fc1(x))
mu = self.fc21(h1)
logvar = self.fc22(h1)
return mu, logvar
def reparameterize(self, mu, logvar):
std = torch.exp(0.5 * logvar)
eps = torch.randn_like(std)
return mu + eps * std
def decoder(self, z):
h3 = self.relu(self.fc3(z))
return self.fc4(h3)
def forward(self, x):
mu, logvar = self.encoder(x.view(-1, 1024))
z = self.reparameterize(mu, logvar)
reconstruction_x = self.decoder(z)
return reconstruction_x, mu, logvar
def sample(self, n):
with torch.no_grad():
z = torch.randn(n, self.z_dims)
z = z
samples = self.decoder(z)
samples = torch.clamp(samples, 0, 1)
return samples.cpu().numpy()
def get_inputs():
return [torch.rand([4, 1024])]
def get_init_inputs():
return [4]
|
SoftDiceLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/gd/cgdtd7ki7lurypoeyfwjebdfquygdeupjef4ltfbbbdk5u7owcpl.py
# Topologically Sorted Source Nodes: [intersection, sum_1, sum_2, sum_3], Original ATen: [aten.mul, aten.sum]
# Source node to ATen node mapping:
# intersection => mul
# sum_1 => sum_1
# sum_2 => sum_2
# sum_3 => sum_3
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view, %view_1), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [1]), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%view, [1]), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%view_1, [1]), kwargs = {})
triton_per_fused_mul_sum_0 = async_compile.triton('triton_per_fused_mul_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[4, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mul_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 3, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mul_sum_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 4
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (64*x0)), xmask, other=0.0)
tmp2 = tl.load(in_ptr1 + (r1 + (64*x0)), xmask, other=0.0)
tmp1 = tl.sigmoid(tmp0)
tmp3 = tmp1 * tmp2
tmp4 = tl.broadcast_to(tmp3, [XBLOCK, RBLOCK])
tmp6 = tl.where(xmask, tmp4, 0)
tmp7 = tl.sum(tmp6, 1)[:, None]
tmp8 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp10 = tl.where(xmask, tmp8, 0)
tmp11 = tl.sum(tmp10, 1)[:, None]
tmp12 = tl.broadcast_to(tmp2, [XBLOCK, RBLOCK])
tmp14 = tl.where(xmask, tmp12, 0)
tmp15 = tl.sum(tmp14, 1)[:, None]
tl.store(out_ptr0 + (x0), tmp7, xmask)
tl.store(out_ptr1 + (x0), tmp11, xmask)
tl.store(out_ptr2 + (x0), tmp15, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/vq/cvqiixp4wmb73ig2cla6idbqq7i6vd5n3qmdluadrv32f52pdgw3.py
# Topologically Sorted Source Nodes: [add, mul_1, add_1, add_2, score, sum_4, truediv_1, score_1], Original ATen: [aten.add, aten.mul, aten.div, aten.sum, aten.rsub]
# Source node to ATen node mapping:
# add => add
# add_1 => add_1
# add_2 => add_2
# mul_1 => mul_1
# score => div
# score_1 => sub
# sum_4 => sum_4
# truediv_1 => div_1
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sum_1, 1), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add, 2.0), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sum_2, %sum_3), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_1, 1), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul_1, %add_2), kwargs = {})
# %sum_4 : [num_users=1] = call_function[target=torch.ops.aten.sum.default](args = (%div,), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sum_4, 4), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %div_1), kwargs = {})
triton_per_fused_add_div_mul_rsub_sum_1 = async_compile.triton('triton_per_fused_add_div_mul_rsub_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 4],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {4: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=(4,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_div_mul_rsub_sum_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_div_mul_rsub_sum_1(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 4
RBLOCK: tl.constexpr = 4
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp5 = tl.load(in_ptr1 + (r0), None)
tmp6 = tl.load(in_ptr2 + (r0), None)
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp3 = 2.0
tmp4 = tmp2 * tmp3
tmp7 = tmp5 + tmp6
tmp8 = tmp7 + tmp1
tmp9 = tmp4 / tmp8
tmp10 = tl.broadcast_to(tmp9, [XBLOCK, RBLOCK])
tmp12 = tl.sum(tmp10, 1)[:, None]
tmp13 = 0.25
tmp14 = tmp12 * tmp13
tmp15 = tmp1 - tmp14
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp15, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, ), (1, ), torch.float32)
buf1 = empty_strided_cuda((4, ), (1, ), torch.float32)
buf2 = empty_strided_cuda((4, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [intersection, sum_1, sum_2, sum_3], Original ATen: [aten.mul, aten.sum]
stream0 = get_raw_stream(0)
triton_per_fused_mul_sum_0.run(arg0_1, arg1_1, buf0, buf1, buf2, 4, 64, grid=grid(4), stream=stream0)
del arg0_1
del arg1_1
buf3 = empty_strided_cuda((), (), torch.float32)
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [add, mul_1, add_1, add_2, score, sum_4, truediv_1, score_1], Original ATen: [aten.add, aten.mul, aten.div, aten.sum, aten.rsub]
triton_per_fused_add_div_mul_rsub_sum_1.run(buf4, buf0, buf1, buf2, 1, 4, grid=grid(1), stream=stream0)
del buf0
del buf1
del buf2
return (buf4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class SoftDiceLoss(nn.Module):
def __init__(self):
super().__init__()
def forward(self, logits, labels):
probs = torch.sigmoid(logits)
num = labels.size(0)
m1 = probs.view(num, -1)
m2 = labels.view(num, -1)
intersection = m1 * m2
score = 2.0 * (intersection.sum(1) + 1) / (m1.sum(1) + m2.sum(1) + 1)
score = 1 - score.sum() / num
return score
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_mul_sum_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
out_ptr2, xnumel, rnumel, XBLOCK: tl.constexpr):
xnumel = 4
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 64 * x0), xmask, other=0.0)
tmp2 = tl.load(in_ptr1 + (r1 + 64 * x0), xmask, other=0.0)
tmp1 = tl.sigmoid(tmp0)
tmp3 = tmp1 * tmp2
tmp4 = tl.broadcast_to(tmp3, [XBLOCK, RBLOCK])
tmp6 = tl.where(xmask, tmp4, 0)
tmp7 = tl.sum(tmp6, 1)[:, None]
tmp8 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp10 = tl.where(xmask, tmp8, 0)
tmp11 = tl.sum(tmp10, 1)[:, None]
tmp12 = tl.broadcast_to(tmp2, [XBLOCK, RBLOCK])
tmp14 = tl.where(xmask, tmp12, 0)
tmp15 = tl.sum(tmp14, 1)[:, None]
tl.store(out_ptr0 + x0, tmp7, xmask)
tl.store(out_ptr1 + x0, tmp11, xmask)
tl.store(out_ptr2 + x0, tmp15, xmask)
@triton.jit
def triton_per_fused_add_div_mul_rsub_sum_1(in_out_ptr0, in_ptr0, in_ptr1,
in_ptr2, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 4
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp5 = tl.load(in_ptr1 + r0, None)
tmp6 = tl.load(in_ptr2 + r0, None)
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp3 = 2.0
tmp4 = tmp2 * tmp3
tmp7 = tmp5 + tmp6
tmp8 = tmp7 + tmp1
tmp9 = tmp4 / tmp8
tmp10 = tl.broadcast_to(tmp9, [XBLOCK, RBLOCK])
tmp12 = tl.sum(tmp10, 1)[:, None]
tmp13 = 0.25
tmp14 = tmp12 * tmp13
tmp15 = tmp1 - tmp14
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp15, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4,), (1,), torch.float32)
buf1 = empty_strided_cuda((4,), (1,), torch.float32)
buf2 = empty_strided_cuda((4,), (1,), torch.float32)
get_raw_stream(0)
triton_per_fused_mul_sum_0[grid(4)](arg0_1, arg1_1, buf0, buf1,
buf2, 4, 64, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
buf3 = empty_strided_cuda((), (), torch.float32)
buf4 = buf3
del buf3
triton_per_fused_add_div_mul_rsub_sum_1[grid(1)](buf4, buf0, buf1,
buf2, 1, 4, XBLOCK=1, num_warps=2, num_stages=1)
del buf0
del buf1
del buf2
return buf4,
class SoftDiceLossNew(nn.Module):
def __init__(self):
super().__init__()
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| marcomatteo/steel-segmentation-nbdev | SoftDiceLoss | false | 7,159 | [
"Apache-2.0"
] | 1 | dde19b0b3bf7657ab575e691bca1751592aecc67 | https://github.com/marcomatteo/steel-segmentation-nbdev/tree/dde19b0b3bf7657ab575e691bca1751592aecc67 | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self):
super().__init__()
def forward(self, logits, labels):
probs = torch.sigmoid(logits)
num = labels.size(0)
m1 = probs.view(num, -1)
m2 = labels.view(num, -1)
intersection = m1 * m2
score = 2.0 * (intersection.sum(1) + 1) / (m1.sum(1) + m2.sum(1) + 1)
score = 1 - score.sum() / num
return score
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
ResizeTransform | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/2h/c2hzrtqhbvxaedsmk5yf4w3blae4viyram4eduvj75lltgf3jdhn.py
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.arange, aten._to_copy, aten.mul, aten.clamp, aten._unsafe_index, aten.sub, aten.add]
# Source node to ATen node mapping:
# x => _unsafe_index, _unsafe_index_1, add_1, clamp_max_1, clamp_min, clamp_min_1, convert_element_type, convert_element_type_1, iota, mul, mul_1, sub, sub_1
# x_1 => mul_2
# Graph fragment:
# %iota : [num_users=1] = call_function[target=torch.ops.prims.iota.default](args = (1,), kwargs = {start: 0, step: 1, dtype: torch.int64, device: cuda:0, requires_grad: False})
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%iota, torch.float32), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%convert_element_type, 0), kwargs = {})
# %clamp_min : [num_users=2] = call_function[target=torch.ops.aten.clamp_min.default](args = (%mul, 0.0), kwargs = {})
# %convert_element_type_1 : [num_users=3] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%clamp_min, torch.int64), kwargs = {})
# %_unsafe_index_1 : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%arg0_1, [None, None, %clamp_max]), kwargs = {})
# %_unsafe_index : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%arg0_1, [None, None, %convert_element_type_1]), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%_unsafe_index_1, %_unsafe_index), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%clamp_min, %convert_element_type_1), kwargs = {})
# %clamp_min_1 : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub, 0.0), kwargs = {})
# %clamp_max_1 : [num_users=1] = call_function[target=torch.ops.aten.clamp_max.default](args = (%clamp_min_1, 1.0), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %clamp_max_1), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%_unsafe_index, %mul_1), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_1, 0.25), kwargs = {})
triton_poi_fused__to_copy__unsafe_index_add_arange_clamp_mul_sub_0 = async_compile.triton('triton_poi_fused__to_copy__unsafe_index_add_arange_clamp_mul_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy__unsafe_index_add_arange_clamp_mul_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy__unsafe_index_add_arange_clamp_mul_sub_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 - tmp0
tmp3 = 0.0
tmp4 = tmp2 * tmp3
tmp5 = tmp0 + tmp4
tmp6 = 0.25
tmp7 = tmp5 * tmp6
tl.store(out_ptr0 + (x0), tmp7, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4), (16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.arange, aten._to_copy, aten.mul, aten.clamp, aten._unsafe_index, aten.sub, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused__to_copy__unsafe_index_add_arange_clamp_mul_sub_0.run(arg0_1, buf0, 16, grid=grid(16), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as nnf
class ResizeTransform(nn.Module):
"""
Resize a transform, which involves resizing the vector field *and* rescaling it.
"""
def __init__(self, vel_resize, ndims):
super().__init__()
self.factor = 1.0 / vel_resize
self.mode = 'linear'
if ndims == 2:
self.mode = 'bi' + self.mode
elif ndims == 3:
self.mode = 'tri' + self.mode
def forward(self, x):
if self.factor < 1:
x = nnf.interpolate(x, align_corners=True, scale_factor=self.
factor, mode=self.mode)
x = self.factor * x
elif self.factor > 1:
x = self.factor * x
x = nnf.interpolate(x, align_corners=True, scale_factor=self.
factor, mode=self.mode)
return x
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'vel_resize': 4, 'ndims': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused__to_copy__unsafe_index_add_arange_clamp_mul_sub_0(in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp1 - tmp0
tmp3 = 0.0
tmp4 = tmp2 * tmp3
tmp5 = tmp0 + tmp4
tmp6 = 0.25
tmp7 = tmp5 * tmp6
tl.store(out_ptr0 + x0, tmp7, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4), (16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1), (4, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__to_copy__unsafe_index_add_arange_clamp_mul_sub_0[grid
(16)](arg0_1, buf0, 16, XBLOCK=16, num_warps=1, num_stages=1)
del arg0_1
return buf0,
class ResizeTransformNew(nn.Module):
"""
Resize a transform, which involves resizing the vector field *and* rescaling it.
"""
def __init__(self, vel_resize, ndims):
super().__init__()
self.factor = 1.0 / vel_resize
self.mode = 'linear'
if ndims == 2:
self.mode = 'bi' + self.mode
elif ndims == 3:
self.mode = 'tri' + self.mode
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mariakesa/ZebraFishRegistrationPipeline | ResizeTransform | false | 7,160 | [
"MIT"
] | 1 | 4955044eb69dc04c579f59ccb24e02e4451aebcc | https://github.com/mariakesa/ZebraFishRegistrationPipeline/tree/4955044eb69dc04c579f59ccb24e02e4451aebcc | import torch
import torch.nn as nn
import torch.nn.functional as nnf
class Model(nn.Module):
"""
Resize a transform, which involves resizing the vector field *and* rescaling it.
"""
def __init__(self, vel_resize, ndims):
super().__init__()
self.factor = 1.0 / vel_resize
self.mode = 'linear'
if ndims == 2:
self.mode = 'bi' + self.mode
elif ndims == 3:
self.mode = 'tri' + self.mode
def forward(self, x):
if self.factor < 1:
x = nnf.interpolate(x, align_corners=True, scale_factor=self.
factor, mode=self.mode)
x = self.factor * x
elif self.factor > 1:
x = self.factor * x
x = nnf.interpolate(x, align_corners=True, scale_factor=self.
factor, mode=self.mode)
return x
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
BahdanauAttention | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/uq/cuqtcap4fkr6ayb745ve57ljczzazno5ktair7zz4d5qsiqe3rvi.py
# Topologically Sorted Source Nodes: [add, tanh], Original ATen: [aten.add, aten.tanh]
# Source node to ATen node mapping:
# add => add
# tanh => tanh
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_1, %view_3), kwargs = {})
# %tanh : [num_users=2] = call_function[target=torch.ops.aten.tanh.default](args = (%add,), kwargs = {})
triton_poi_fused_add_tanh_0 = async_compile.triton('triton_poi_fused_add_tanh_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_tanh_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_tanh_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask)
tmp2 = tmp0 + tmp1
tmp3 = libdevice.tanh(tmp2)
tl.store(in_out_ptr0 + (x0), tmp3, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ts/ctscnzvbagjv4t25zui245b3recij5udu7nvujnr5rixcyo7elc6.py
# Topologically Sorted Source Nodes: [attention_weight], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attention_weight => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%squeeze, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%squeeze, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_1 = async_compile.triton('triton_poi_fused__softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/k6/ck6fz3qsfeqgn5jtm4ugikmu7cwvvlq3jpttijbb5kdniicwtyz6.py
# Topologically Sorted Source Nodes: [attention_weight], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attention_weight => div, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %div : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_2 = async_compile.triton('triton_poi_fused__softmax_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_5, (1, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_4, (16, 4), (4, 1), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf1)
del primals_3
buf2 = reinterpret_tensor(buf0, (4, 4, 4), (16, 4, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [add, tanh], Original ATen: [aten.add, aten.tanh]
stream0 = get_raw_stream(0)
triton_poi_fused_add_tanh_0.run(buf2, buf1, 64, grid=grid(64), stream=stream0)
del buf1
buf3 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [score], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf2, (16, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 1), (1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [attention_weight], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf3, buf4, 16, grid=grid(16), stream=stream0)
buf5 = reinterpret_tensor(buf3, (4, 4), (4, 1), 0); del buf3 # reuse
# Topologically Sorted Source Nodes: [attention_weight], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf4, buf5, 16, grid=grid(16), stream=stream0)
buf6 = reinterpret_tensor(buf4, (4, 1, 4), (4, 4, 1), 0); del buf4 # reuse
# Topologically Sorted Source Nodes: [bmm], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf5, (4, 1, 4), (4, 4, 1), 0), primals_4, out=buf6)
return (buf5, reinterpret_tensor(buf6, (4, 4), (4, 1), 0), primals_4, reinterpret_tensor(primals_2, (16, 4), (4, 1), 0), buf2, buf5, primals_5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((1, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import Tensor
import torch.nn as nn
from typing import Tuple
import torch.nn.functional as F
class BahdanauAttention(nn.Module):
def __init__(self, dec_dim: 'int', enc_dim: 'int', num_hiddens: 'int'):
super().__init__()
self.W1 = nn.Linear(dec_dim, num_hiddens, bias=False)
self.W2 = nn.Linear(enc_dim, num_hiddens, bias=False)
self.v = nn.Linear(num_hiddens, 1, False)
def forward(self, query: 'Tensor', value: 'Tensor') ->Tuple[Tensor, Tensor
]:
"""
Args:
value (Tensor(batch size, seq_len, encoder hidden dimension): the hidden_state of tokens in encoder
query (Tensor(batch size, 1, decoder hidden dimension)): the hidden state of decoder at time step t
Returns:
attention_weight (Tensor)
context_vector (Tensor)
"""
score = self.v(torch.tanh(self.W1(query) + self.W2(value)))
attention_weight = F.softmax(score.squeeze(-1), dim=1)
context_vector = torch.bmm(attention_weight.unsqueeze(1), value
).squeeze(1)
return attention_weight, context_vector
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'dec_dim': 4, 'enc_dim': 4, 'num_hiddens': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_tanh_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask)
tmp2 = tmp0 + tmp1
tmp3 = libdevice.tanh(tmp2)
tl.store(in_out_ptr0 + x0, tmp3, xmask)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_5, (1, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_2, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_4, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), out=buf1)
del primals_3
buf2 = reinterpret_tensor(buf0, (4, 4, 4), (16, 4, 1), 0)
del buf0
get_raw_stream(0)
triton_poi_fused_add_tanh_0[grid(64)](buf2, buf1, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf1
buf3 = empty_strided_cuda((16, 1), (1, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf2, (16, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 1), (1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
triton_poi_fused__softmax_1[grid(16)](buf3, buf4, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf5 = reinterpret_tensor(buf3, (4, 4), (4, 1), 0)
del buf3
triton_poi_fused__softmax_2[grid(16)](buf4, buf5, 16, XBLOCK=16,
num_warps=1, num_stages=1)
buf6 = reinterpret_tensor(buf4, (4, 1, 4), (4, 4, 1), 0)
del buf4
extern_kernels.bmm(reinterpret_tensor(buf5, (4, 1, 4), (4, 4, 1), 0
), primals_4, out=buf6)
return buf5, reinterpret_tensor(buf6, (4, 4), (4, 1), 0
), primals_4, reinterpret_tensor(primals_2, (16, 4), (4, 1), 0
), buf2, buf5, primals_5
class BahdanauAttentionNew(nn.Module):
def __init__(self, dec_dim: 'int', enc_dim: 'int', num_hiddens: 'int'):
super().__init__()
self.W1 = nn.Linear(dec_dim, num_hiddens, bias=False)
self.W2 = nn.Linear(enc_dim, num_hiddens, bias=False)
self.v = nn.Linear(num_hiddens, 1, False)
def forward(self, input_0, input_1):
primals_1 = self.W1.weight
primals_3 = self.W2.weight
primals_5 = self.v.weight
primals_2 = input_0
primals_4 = input_1
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0], output[1]
| manhtrantienhn/Sentiment-with-pretrain-model | BahdanauAttention | false | 7,161 | [
"MIT"
] | 1 | bbbbaa94cf481afcfe704cbcb27b602308f43de5 | https://github.com/manhtrantienhn/Sentiment-with-pretrain-model/tree/bbbbaa94cf481afcfe704cbcb27b602308f43de5 | import torch
from torch import Tensor
import torch.nn as nn
from typing import Tuple
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self, dec_dim: 'int', enc_dim: 'int', num_hiddens: 'int'):
super().__init__()
self.W1 = nn.Linear(dec_dim, num_hiddens, bias=False)
self.W2 = nn.Linear(enc_dim, num_hiddens, bias=False)
self.v = nn.Linear(num_hiddens, 1, False)
def forward(self, query: 'Tensor', value: 'Tensor') ->Tuple[Tensor, Tensor
]:
"""
Args:
value (Tensor(batch size, seq_len, encoder hidden dimension): the hidden_state of tokens in encoder
query (Tensor(batch size, 1, decoder hidden dimension)): the hidden state of decoder at time step t
Returns:
attention_weight (Tensor)
context_vector (Tensor)
"""
score = self.v(torch.tanh(self.W1(query) + self.W2(value)))
attention_weight = F.softmax(score.squeeze(-1), dim=1)
context_vector = torch.bmm(attention_weight.unsqueeze(1), value
).squeeze(1)
return attention_weight, context_vector
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4])]
def get_init_inputs():
return [4, 4, 4]
|
CRF | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/hf/chfy6golfkp5hyoqxyfqo3cuiy7r25sg3yr3dcq6fpt7nsbdq6zt.py
# Topologically Sorted Source Nodes: [crf_scores], Original ATen: [aten.add]
# Source node to ATen node mapping:
# crf_scores => add
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%expand, %unsqueeze_2), kwargs = {})
triton_poi_fused_add_0 = async_compile.triton('triton_poi_fused_add_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = (xindex // 4)
x4 = xindex % 16
x5 = xindex
tmp0 = tl.load(in_ptr0 + (x3), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x4), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x5), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [crf_scores], Original ATen: [aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_0.run(primals_1, primals_2, buf0, 256, grid=grid(256), stream=stream0)
del primals_1
del primals_2
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.utils.data.dataloader
import torch.nn
class CRF(torch.nn.Module):
"""
Conditional Random Field Implementation according to sgrvinod (https://github.com/sgrvinod).
Classifier which predicts single tag / class / label for given word based on not just the word,
but also on previous seen annotations.
"""
def __init__(self, tag_dictionary, tagset_size: 'int',
init_from_state_dict: 'bool'):
"""
:param tag_dictionary: tag dictionary in order to find ID for start and stop tags
:param tagset_size: number of tag from tag dictionary
:param init_from_state_dict: whether we load pretrained model from state dict
"""
super(CRF, self).__init__()
self.tagset_size = tagset_size
self.transitions = torch.nn.Parameter(torch.randn(tagset_size,
tagset_size))
if not init_from_state_dict:
self.transitions.detach()[tag_dictionary.get_idx_for_item(
START_TAG), :] = -10000
self.transitions.detach()[:, tag_dictionary.get_idx_for_item(
STOP_TAG)] = -10000
self
def forward(self, features: 'torch.Tensor') ->torch.Tensor:
"""
Forward propagation of Conditional Random Field.
:param features: output from RNN / Linear layer in shape (batch size, seq len, hidden size)
:return: CRF scores (emission scores for each token + transitions prob from previous state) in
shape (batch_size, seq len, tagset size, tagset size)
"""
batch_size, seq_len = features.size()[:2]
emission_scores = features
emission_scores = emission_scores.unsqueeze(-1).expand(batch_size,
seq_len, self.tagset_size, self.tagset_size)
crf_scores = emission_scores + self.transitions.unsqueeze(0).unsqueeze(
0)
return crf_scores
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'tag_dictionary': 4, 'tagset_size': 4,
'init_from_state_dict': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.utils.data.dataloader
import torch.nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex // 4
x4 = xindex % 16
x5 = xindex
tmp0 = tl.load(in_ptr0 + x3, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + x4, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x5, tmp2, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_0[grid(256)](primals_1, primals_2, buf0, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
del primals_2
return buf0,
class CRFNew(torch.nn.Module):
"""
Conditional Random Field Implementation according to sgrvinod (https://github.com/sgrvinod).
Classifier which predicts single tag / class / label for given word based on not just the word,
but also on previous seen annotations.
"""
def __init__(self, tag_dictionary, tagset_size: 'int',
init_from_state_dict: 'bool'):
"""
:param tag_dictionary: tag dictionary in order to find ID for start and stop tags
:param tagset_size: number of tag from tag dictionary
:param init_from_state_dict: whether we load pretrained model from state dict
"""
super(CRFNew, self).__init__()
self.tagset_size = tagset_size
self.transitions = torch.nn.Parameter(torch.randn(tagset_size,
tagset_size))
if not init_from_state_dict:
self.transitions.detach()[tag_dictionary.get_idx_for_item(
START_TAG), :] = -10000
self.transitions.detach()[:, tag_dictionary.get_idx_for_item(
STOP_TAG)] = -10000
self
def forward(self, input_0):
primals_2 = self.transitions
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0]
| marleneDebatin/flair | CRF | false | 7,162 | [
"MIT"
] | 1 | 4d17509f358158f66d43e85db1b6990523b0b095 | https://github.com/marleneDebatin/flair/tree/4d17509f358158f66d43e85db1b6990523b0b095 | import torch
import torch.utils.data.dataloader
import torch.nn
class Model(torch.nn.Module):
"""
Conditional Random Field Implementation according to sgrvinod (https://github.com/sgrvinod).
Classifier which predicts single tag / class / label for given word based on not just the word,
but also on previous seen annotations.
"""
def __init__(self, tag_dictionary, tagset_size: 'int',
init_from_state_dict: 'bool'):
"""
:param tag_dictionary: tag dictionary in order to find ID for start and stop tags
:param tagset_size: number of tag from tag dictionary
:param init_from_state_dict: whether we load pretrained model from state dict
"""
super().__init__()
self.tagset_size = tagset_size
self.transitions = torch.nn.Parameter(torch.randn(tagset_size,
tagset_size))
if not init_from_state_dict:
self.transitions.detach()[tag_dictionary.get_idx_for_item(
START_TAG), :] = -10000
self.transitions.detach()[:, tag_dictionary.get_idx_for_item(
STOP_TAG)] = -10000
self
def forward(self, features: 'torch.Tensor') ->torch.Tensor:
"""
Forward propagation of Conditional Random Field.
:param features: output from RNN / Linear layer in shape (batch size, seq len, hidden size)
:return: CRF scores (emission scores for each token + transitions prob from previous state) in
shape (batch_size, seq len, tagset size, tagset size)
"""
batch_size, seq_len = features.size()[:2]
emission_scores = features
emission_scores = emission_scores.unsqueeze(-1).expand(batch_size,
seq_len, self.tagset_size, self.tagset_size)
crf_scores = emission_scores + self.transitions.unsqueeze(0).unsqueeze(
0)
return crf_scores
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'tag_dictionary': 4, 'tagset_size': 4,
'init_from_state_dict': 4}]
|
FocalLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/34/c34k3cchubxdqxplz72ypcl6x5y5m2we27einuhhp6e2qepedw3p.py
# Topologically Sorted Source Nodes: [bce_loss, neg, pt, sub, pow_1, mul, focal_loss, mean], Original ATen: [aten.binary_cross_entropy_with_logits, aten.neg, aten.exp, aten.rsub, aten.pow, aten.mul, aten.mean]
# Source node to ATen node mapping:
# bce_loss => abs_1, exp, full_default, log1p, minimum, mul, neg, sub, sub_1, sub_2
# focal_loss => mul_2
# mean => mean
# mul => mul_1
# neg => neg_1
# pow_1 => pow_1
# pt => exp_1
# sub => sub_3
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %arg0_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %arg1_1), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %minimum : [num_users=1] = call_function[target=torch.ops.aten.minimum.default](args = (%full_default, %arg1_1), kwargs = {})
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%arg1_1,), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%abs_1,), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%neg,), kwargs = {})
# %log1p : [num_users=1] = call_function[target=torch.ops.aten.log1p.default](args = (%exp,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%minimum, %log1p), kwargs = {})
# %sub_2 : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul, %sub_1), kwargs = {})
# %neg_1 : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%sub_2,), kwargs = {})
# %exp_1 : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%neg_1,), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %exp_1), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_3, 2), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%pow_1, 0.5), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_1, %sub_2), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%mul_2,), kwargs = {})
triton_per_fused_binary_cross_entropy_with_logits_exp_mean_mul_neg_pow_rsub_0 = async_compile.triton('triton_per_fused_binary_cross_entropy_with_logits_exp_mean_mul_neg_pow_rsub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_binary_cross_entropy_with_logits_exp_mean_mul_neg_pow_rsub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_binary_cross_entropy_with_logits_exp_mean_mul_neg_pow_rsub_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp3 = tl.load(in_ptr1 + (r0), None)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp2 * tmp3
tmp5 = 0.0
tmp6 = triton_helpers.minimum(tmp5, tmp3)
tmp7 = tl_math.abs(tmp3)
tmp8 = -tmp7
tmp9 = tl_math.exp(tmp8)
tmp10 = libdevice.log1p(tmp9)
tmp11 = tmp6 - tmp10
tmp12 = tmp4 - tmp11
tmp13 = -tmp12
tmp14 = tl_math.exp(tmp13)
tmp15 = tmp1 - tmp14
tmp16 = tmp15 * tmp15
tmp17 = 0.5
tmp18 = tmp16 * tmp17
tmp19 = tmp18 * tmp12
tmp20 = tl.broadcast_to(tmp19, [RBLOCK])
tmp22 = triton_helpers.promote_to_tensor(tl.sum(tmp20, 0))
tmp23 = 256.0
tmp24 = tmp22 / tmp23
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp24, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [bce_loss, neg, pt, sub, pow_1, mul, focal_loss, mean], Original ATen: [aten.binary_cross_entropy_with_logits, aten.neg, aten.exp, aten.rsub, aten.pow, aten.mul, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_binary_cross_entropy_with_logits_exp_mean_mul_neg_pow_rsub_0.run(buf1, arg0_1, arg1_1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn.functional as F
from torch import nn as nn
class FocalLoss(nn.Module):
"""Focal loss function for imbalanced dataset.
Args:
alpha (float): weighing factor between 0 and 1. Alpha may be set by inverse
class frequency
gamma (float): modulating factor reduces the loss contribution from easy
examples and extends the range in which an example receives
low loss. Usually between 0 - 5.
"""
def __init__(self, alpha=0.5, gamma=2):
super().__init__()
self.alpha = alpha
self.gamma = gamma
def forward(self, logits, y):
bce_loss = F.binary_cross_entropy_with_logits(logits, y, reduction=
'none')
pt = torch.exp(-bce_loss)
focal_loss = self.alpha * (1 - pt) ** self.gamma * bce_loss
return focal_loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch import nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_binary_cross_entropy_with_logits_exp_mean_mul_neg_pow_rsub_0(
in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp3 = tl.load(in_ptr1 + r0, None)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp2 * tmp3
tmp5 = 0.0
tmp6 = triton_helpers.minimum(tmp5, tmp3)
tmp7 = tl_math.abs(tmp3)
tmp8 = -tmp7
tmp9 = tl_math.exp(tmp8)
tmp10 = libdevice.log1p(tmp9)
tmp11 = tmp6 - tmp10
tmp12 = tmp4 - tmp11
tmp13 = -tmp12
tmp14 = tl_math.exp(tmp13)
tmp15 = tmp1 - tmp14
tmp16 = tmp15 * tmp15
tmp17 = 0.5
tmp18 = tmp16 * tmp17
tmp19 = tmp18 * tmp12
tmp20 = tl.broadcast_to(tmp19, [RBLOCK])
tmp22 = triton_helpers.promote_to_tensor(tl.sum(tmp20, 0))
tmp23 = 256.0
tmp24 = tmp22 / tmp23
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp24, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_per_fused_binary_cross_entropy_with_logits_exp_mean_mul_neg_pow_rsub_0[
grid(1)](buf1, arg0_1, arg1_1, 1, 256, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf1,
class FocalLossNew(nn.Module):
"""Focal loss function for imbalanced dataset.
Args:
alpha (float): weighing factor between 0 and 1. Alpha may be set by inverse
class frequency
gamma (float): modulating factor reduces the loss contribution from easy
examples and extends the range in which an example receives
low loss. Usually between 0 - 5.
"""
def __init__(self, alpha=0.5, gamma=2):
super().__init__()
self.alpha = alpha
self.gamma = gamma
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| marshuang80/pe-slice-finder | FocalLoss | false | 7,163 | [
"Apache-2.0"
] | 1 | 2426a55c404e8eb694110351d604d6bdd613e5ae | https://github.com/marshuang80/pe-slice-finder/tree/2426a55c404e8eb694110351d604d6bdd613e5ae | import torch
import torch.nn.functional as F
from torch import nn as nn
class Model(nn.Module):
"""Focal loss function for imbalanced dataset.
Args:
alpha (float): weighing factor between 0 and 1. Alpha may be set by inverse
class frequency
gamma (float): modulating factor reduces the loss contribution from easy
examples and extends the range in which an example receives
low loss. Usually between 0 - 5.
"""
def __init__(self, alpha=0.5, gamma=2):
super().__init__()
self.alpha = alpha
self.gamma = gamma
def forward(self, logits, y):
bce_loss = F.binary_cross_entropy_with_logits(logits, y, reduction=
'none')
pt = torch.exp(-bce_loss)
focal_loss = self.alpha * (1 - pt) ** self.gamma * bce_loss
return focal_loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
TFBCELoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/xo/cxojlva4uuyq2scwxej6apmzdandla3hahj5qvral7m5yzo4lp4y.py
# Topologically Sorted Source Nodes: [relu_logits, mul, term1, abs_1, neg_abs_logits, exp, term2, loss, sum_1], Original ATen: [aten.relu, aten.mul, aten.sub, aten.abs, aten.neg, aten.exp, aten.log1p, aten.add, aten.sum]
# Source node to ATen node mapping:
# abs_1 => abs_1
# exp => exp
# loss => add
# mul => mul
# neg_abs_logits => neg
# relu_logits => relu
# sum_1 => sum_1
# term1 => sub
# term2 => log1p
# Graph fragment:
# %relu : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%arg0_1,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, %arg1_1), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%relu, %mul), kwargs = {})
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%arg0_1,), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%abs_1,), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%neg,), kwargs = {})
# %log1p : [num_users=1] = call_function[target=torch.ops.aten.log1p.default](args = (%exp,), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sub, %log1p), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%add, [-1]), kwargs = {})
triton_poi_fused_abs_add_exp_log1p_mul_neg_relu_sub_sum_0 = async_compile.triton('triton_poi_fused_abs_add_exp_log1p_mul_neg_relu_sub_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_abs_add_exp_log1p_mul_neg_relu_sub_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_abs_add_exp_log1p_mul_neg_relu_sub_sum_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp13 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp24 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp33 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp35 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp1 = tl.full([1], 0, tl.int32)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = tmp0 * tmp3
tmp5 = tmp2 - tmp4
tmp6 = tl_math.abs(tmp0)
tmp7 = -tmp6
tmp8 = tl_math.exp(tmp7)
tmp9 = libdevice.log1p(tmp8)
tmp10 = tmp5 + tmp9
tmp12 = triton_helpers.maximum(tmp1, tmp11)
tmp14 = tmp11 * tmp13
tmp15 = tmp12 - tmp14
tmp16 = tl_math.abs(tmp11)
tmp17 = -tmp16
tmp18 = tl_math.exp(tmp17)
tmp19 = libdevice.log1p(tmp18)
tmp20 = tmp15 + tmp19
tmp21 = tmp10 + tmp20
tmp23 = triton_helpers.maximum(tmp1, tmp22)
tmp25 = tmp22 * tmp24
tmp26 = tmp23 - tmp25
tmp27 = tl_math.abs(tmp22)
tmp28 = -tmp27
tmp29 = tl_math.exp(tmp28)
tmp30 = libdevice.log1p(tmp29)
tmp31 = tmp26 + tmp30
tmp32 = tmp21 + tmp31
tmp34 = triton_helpers.maximum(tmp1, tmp33)
tmp36 = tmp33 * tmp35
tmp37 = tmp34 - tmp36
tmp38 = tl_math.abs(tmp33)
tmp39 = -tmp38
tmp40 = tl_math.exp(tmp39)
tmp41 = libdevice.log1p(tmp40)
tmp42 = tmp37 + tmp41
tmp43 = tmp32 + tmp42
tl.store(out_ptr0 + (x0), tmp43, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/kx/ckxdoomfej5mhehr32jl2nyeqf6tanuvuodvsv227nlcu244ph7y.py
# Topologically Sorted Source Nodes: [loss_1], Original ATen: [aten.mean]
# Source node to ATen node mapping:
# loss_1 => mean
# Graph fragment:
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%sum_1, [-1]), kwargs = {})
triton_poi_fused_mean_1 = async_compile.triton('triton_poi_fused_mean_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mean_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mean_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tl.store(out_ptr0 + (x0), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [relu_logits, mul, term1, abs_1, neg_abs_logits, exp, term2, loss, sum_1], Original ATen: [aten.relu, aten.mul, aten.sub, aten.abs, aten.neg, aten.exp, aten.log1p, aten.add, aten.sum]
stream0 = get_raw_stream(0)
triton_poi_fused_abs_add_exp_log1p_mul_neg_relu_sub_sum_0.run(arg0_1, arg1_1, buf0, 64, grid=grid(64), stream=stream0)
del arg0_1
del arg1_1
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [loss_1], Original ATen: [aten.mean]
triton_poi_fused_mean_1.run(buf0, buf1, 16, grid=grid(16), stream=stream0)
del buf0
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class TFBCELoss(nn.Module):
def __init__(self, pos_weight):
super().__init__()
self.pos_weight = pos_weight
def forward(self, logits, targets):
relu_logits = F.relu(logits)
neg_abs_logits = -torch.abs(logits)
term1 = relu_logits - logits * targets
term2 = torch.log1p(torch.exp(neg_abs_logits))
loss = term1 + term2
loss = loss.sum(dim=-1).mean(dim=-1)
return loss
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'pos_weight': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_abs_add_exp_log1p_mul_neg_relu_sub_sum_0(in_ptr0,
in_ptr1, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp13 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp22 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp24 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp33 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp35 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp1 = tl.full([1], 0, tl.int32)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = tmp0 * tmp3
tmp5 = tmp2 - tmp4
tmp6 = tl_math.abs(tmp0)
tmp7 = -tmp6
tmp8 = tl_math.exp(tmp7)
tmp9 = libdevice.log1p(tmp8)
tmp10 = tmp5 + tmp9
tmp12 = triton_helpers.maximum(tmp1, tmp11)
tmp14 = tmp11 * tmp13
tmp15 = tmp12 - tmp14
tmp16 = tl_math.abs(tmp11)
tmp17 = -tmp16
tmp18 = tl_math.exp(tmp17)
tmp19 = libdevice.log1p(tmp18)
tmp20 = tmp15 + tmp19
tmp21 = tmp10 + tmp20
tmp23 = triton_helpers.maximum(tmp1, tmp22)
tmp25 = tmp22 * tmp24
tmp26 = tmp23 - tmp25
tmp27 = tl_math.abs(tmp22)
tmp28 = -tmp27
tmp29 = tl_math.exp(tmp28)
tmp30 = libdevice.log1p(tmp29)
tmp31 = tmp26 + tmp30
tmp32 = tmp21 + tmp31
tmp34 = triton_helpers.maximum(tmp1, tmp33)
tmp36 = tmp33 * tmp35
tmp37 = tmp34 - tmp36
tmp38 = tl_math.abs(tmp33)
tmp39 = -tmp38
tmp40 = tl_math.exp(tmp39)
tmp41 = libdevice.log1p(tmp40)
tmp42 = tmp37 + tmp41
tmp43 = tmp32 + tmp42
tl.store(out_ptr0 + x0, tmp43, xmask)
@triton.jit
def triton_poi_fused_mean_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tl.store(out_ptr0 + x0, tmp8, xmask)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_abs_add_exp_log1p_mul_neg_relu_sub_sum_0[grid(64)](
arg0_1, arg1_1, buf0, 64, XBLOCK=64, num_warps=1, num_stages=1)
del arg0_1
del arg1_1
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
triton_poi_fused_mean_1[grid(16)](buf0, buf1, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del buf0
return buf1,
class TFBCELossNew(nn.Module):
def __init__(self, pos_weight):
super().__init__()
self.pos_weight = pos_weight
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| marload/DAFT | TFBCELoss | false | 7,164 | [
"Apache-2.0"
] | 1 | 22ebe1cc1d1ca8d4b1f7557bf5833983c63ba330 | https://github.com/marload/DAFT/tree/22ebe1cc1d1ca8d4b1f7557bf5833983c63ba330 | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self, pos_weight):
super().__init__()
self.pos_weight = pos_weight
def forward(self, logits, targets):
relu_logits = F.relu(logits)
neg_abs_logits = -torch.abs(logits)
term1 = relu_logits - logits * targets
term2 = torch.log1p(torch.exp(neg_abs_logits))
loss = term1 + term2
loss = loss.sum(dim=-1).mean(dim=-1)
return loss
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
PyramidDown | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/gy/cgyfx47penjbxrnlqdpur6wznrc2npiddss2rmhvsk53kjjd4wdb.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%convolution, %convolution_1, %convolution_2, %convolution_3], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x0 = xindex % 4
x2 = (xindex // 16)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (4*x2)), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + (4*x2)), tmp9 & xmask, eviction_policy='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x0 + (4*x2)), tmp14 & xmask, eviction_policy='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tmp17 = tl.full([1], 4, tl.int64)
tmp18 = tmp0 < tmp17
tmp19 = tl.load(in_ptr3 + (x0 + (4*x2)), tmp16 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + (x3), tmp22, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (1, 1, 5, 5), (25, 25, 5, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 16, 4, 1), 0), arg1_1, stride=(2, 2), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 1, 2, 2), (4, 4, 2, 1))
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 16, 4, 1), 16), arg1_1, stride=(2, 2), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 1, 2, 2), (4, 4, 2, 1))
# Topologically Sorted Source Nodes: [conv2d_2], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 16, 4, 1), 32), arg1_1, stride=(2, 2), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 2, 2), (4, 4, 2, 1))
# Topologically Sorted Source Nodes: [conv2d_3], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 16, 4, 1), 48), arg1_1, stride=(2, 2), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 1, 2, 2), (4, 4, 2, 1))
del arg0_1
del arg1_1
buf4 = empty_strided_cuda((4, 4, 2, 2), (16, 4, 2, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(buf0, buf1, buf2, buf3, buf4, 64, grid=grid(64), stream=stream0)
del buf0
del buf1
del buf2
del buf3
return (buf4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((1, 1, 5, 5), (25, 25, 5, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
from torch.nn import functional as F
class PyramidDown(nn.Module):
def __init__(self) ->None:
super(PyramidDown, self).__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(x[:, i:i + 1, :, :], self.filter,
padding=2, stride=2))
return torch.cat(results, dim=1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x0 = xindex % 4
x2 = xindex // 16
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 4 * x2), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + 4 * x2), tmp9 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x0 + 4 * x2), tmp14 & xmask, eviction_policy
='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tl.full([1], 4, tl.int64)
tmp19 = tl.load(in_ptr3 + (x0 + 4 * x2), tmp16 & xmask, eviction_policy
='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + x3, tmp22, xmask)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (1, 1, 5, 5), (25, 25, 5, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 16, 4, 1), 0), arg1_1, stride=(2, 2), padding=(2, 2
), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf0, (4, 1, 2, 2), (4, 4, 2, 1))
buf1 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 16, 4, 1), 16), arg1_1, stride=(2, 2), padding=(2,
2), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf1, (4, 1, 2, 2), (4, 4, 2, 1))
buf2 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 16, 4, 1), 32), arg1_1, stride=(2, 2), padding=(2,
2), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 2, 2), (4, 4, 2, 1))
buf3 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 16, 4, 1), 48), arg1_1, stride=(2, 2), padding=(2,
2), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf3, (4, 1, 2, 2), (4, 4, 2, 1))
del arg0_1
del arg1_1
buf4 = empty_strided_cuda((4, 4, 2, 2), (16, 4, 2, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(64)](buf0, buf1, buf2, buf3, buf4, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del buf0
del buf1
del buf2
del buf3
return buf4,
class PyramidDownNew(nn.Module):
def __init__(self) ->None:
super(PyramidDownNew, self).__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, input_0):
arg1_1 = self.filter
arg0_1 = input_0
output = call([arg0_1, arg1_1])
return output[0]
| masanorihirano/pytorch_extra_mhirano | PyramidDown | false | 7,165 | [
"MIT"
] | 1 | d19e07445567c069793b7ca1a22a846d7cbce58d | https://github.com/masanorihirano/pytorch_extra_mhirano/tree/d19e07445567c069793b7ca1a22a846d7cbce58d | import torch
import torch.nn as nn
from torch.nn import functional as F
class Model(nn.Module):
def __init__(self) ->None:
super().__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(x[:, i:i + 1, :, :], self.filter,
padding=2, stride=2))
return torch.cat(results, dim=1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
Net | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/v3/cv3hbcc77t43qvhdwsbud75pm4bbv4e7bc4hsezz7oxunbuwitzv.py
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv2d => convolution
# Graph fragment:
# %convolution : [num_users=2] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16384],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16000
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 400) % 10
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/j7/cj7ibqexk6ivanurr6dwb3nxx6m2qysl2tiw3v7k3mwp5sdwo7rq.py
# Topologically Sorted Source Nodes: [max_pool2d, x], Original ATen: [aten.max_pool2d_with_indices, aten.relu]
# Source node to ATen node mapping:
# max_pool2d => _low_memory_max_pool2d_with_offsets, getitem_1
# x => relu
# Graph fragment:
# %_low_memory_max_pool2d_with_offsets : [num_users=2] = call_function[target=torch.ops.prims._low_memory_max_pool2d_with_offsets.default](args = (%convolution, [2, 2], [2, 2], [0, 0], [1, 1], False), kwargs = {})
# %getitem_1 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%getitem,), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_relu_1 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i8', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_relu_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_relu_1(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 4000
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 10
x1 = (xindex // 10)
x2 = xindex
tmp0 = tl.load(in_ptr0 + ((2*x0) + (40*x1)), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (2*x0) + (40*x1)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (20 + (2*x0) + (40*x1)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (21 + (2*x0) + (40*x1)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 > tmp0
tmp3 = tl.full([1], 1, tl.int8)
tmp4 = tl.full([1], 0, tl.int8)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp6 = triton_helpers.maximum(tmp1, tmp0)
tmp8 = tmp7 > tmp6
tmp9 = tl.full([1], 2, tl.int8)
tmp10 = tl.where(tmp8, tmp9, tmp5)
tmp11 = triton_helpers.maximum(tmp7, tmp6)
tmp13 = tmp12 > tmp11
tmp14 = tl.full([1], 3, tl.int8)
tmp15 = tl.where(tmp13, tmp14, tmp10)
tmp16 = triton_helpers.maximum(tmp12, tmp11)
tmp17 = tl.full([1], 0, tl.int32)
tmp18 = triton_helpers.maximum(tmp17, tmp16)
tl.store(out_ptr0 + (x2), tmp15, xmask)
tl.store(out_ptr1 + (x2), tmp18, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/vc/cvcco7d3skvxpcw6soqmc2gai56s47m5gx5maj6poaozwjp2jilz.py
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv2d_1 => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=2] = call_function[target=torch.ops.aten.convolution.default](args = (%relu, %primals_4, %primals_5, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_2 = async_compile.triton('triton_poi_fused_convolution_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 2880
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 36) % 20
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/uh/cuhjbrwzmkx3ucrmix635v4jijiphbyjajem22lkdhipzywvav5e.py
# Topologically Sorted Source Nodes: [max_pool2d_1, x_1], Original ATen: [aten.max_pool2d_with_indices, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# max_pool2d_1 => _low_memory_max_pool2d_with_offsets_1, getitem_3
# x_1 => relu_1
# Graph fragment:
# %_low_memory_max_pool2d_with_offsets_1 : [num_users=2] = call_function[target=torch.ops.prims._low_memory_max_pool2d_with_offsets.default](args = (%convolution_1, [2, 2], [2, 2], [0, 0], [1, 1], False), kwargs = {})
# %getitem_3 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_1, 1), kwargs = {})
# %relu_1 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%getitem_2,), kwargs = {})
# %le_1 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_1, 0), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_relu_threshold_backward_3 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_relu_threshold_backward_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*i8', 2: '*fp32', 3: '*i1', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_relu_threshold_backward_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_relu_threshold_backward_3(in_ptr0, out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 720
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 3
x1 = (xindex // 3)
x2 = xindex
tmp0 = tl.load(in_ptr0 + ((2*x0) + (12*x1)), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (2*x0) + (12*x1)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (6 + (2*x0) + (12*x1)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (7 + (2*x0) + (12*x1)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 > tmp0
tmp3 = tl.full([1], 1, tl.int8)
tmp4 = tl.full([1], 0, tl.int8)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp6 = triton_helpers.maximum(tmp1, tmp0)
tmp8 = tmp7 > tmp6
tmp9 = tl.full([1], 2, tl.int8)
tmp10 = tl.where(tmp8, tmp9, tmp5)
tmp11 = triton_helpers.maximum(tmp7, tmp6)
tmp13 = tmp12 > tmp11
tmp14 = tl.full([1], 3, tl.int8)
tmp15 = tl.where(tmp13, tmp14, tmp10)
tmp16 = triton_helpers.maximum(tmp12, tmp11)
tmp17 = tl.full([1], 0, tl.int32)
tmp18 = triton_helpers.maximum(tmp17, tmp16)
tmp19 = 0.0
tmp20 = tmp18 <= tmp19
tl.store(out_ptr0 + (x2), tmp15, xmask)
tl.store(out_ptr1 + (x2), tmp18, xmask)
tl.store(out_ptr2 + (x2), tmp20, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/xi/cxiouqdy3aqmwg2p7remcu57ny7vnw4f263cpqtavvo5sya77u64.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# x_3 => relu_2
# Graph fragment:
# %add_tensor : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default, %primals_7), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor,), kwargs = {})
triton_poi_fused_relu_4 = async_compile.triton('triton_poi_fused_relu_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_4', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_4(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 50
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/u6/cu6h2do6fxp5xndssweqjlj424anye3onmy3qdp667s6ssb2acmm.py
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# log_softmax => amax, exp, log, sub, sub_1, sum_1
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%addmm_1, [1], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%addmm_1, %amax), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %log), kwargs = {})
triton_per_fused__log_softmax_5 = async_compile.triton('triton_per_fused__log_softmax_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[4, 8],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused__log_softmax_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 2, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused__log_softmax_5(in_ptr0, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 4
rnumel = 8
RBLOCK: tl.constexpr = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (8*x0)), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, float("-inf"))
tmp4 = triton_helpers.max2(tmp3, 1)[:, None]
tmp5 = tmp0 - tmp4
tmp6 = tl_math.exp(tmp5)
tmp7 = tl.broadcast_to(tmp6, [XBLOCK, RBLOCK])
tmp9 = tl.where(xmask, tmp7, 0)
tmp10 = tl.sum(tmp9, 1)[:, None]
tmp11 = tl_math.log(tmp10)
tmp12 = tmp5 - tmp11
tl.store(out_ptr2 + (r1 + (8*x0)), tmp12, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9 = args
args.clear()
assert_size_stride(primals_1, (10, 3, 5, 5), (75, 25, 5, 1))
assert_size_stride(primals_2, (10, ), (1, ))
assert_size_stride(primals_3, (4, 3, 24, 24), (1728, 576, 24, 1))
assert_size_stride(primals_4, (20, 10, 5, 5), (250, 25, 5, 1))
assert_size_stride(primals_5, (20, ), (1, ))
assert_size_stride(primals_6, (50, 180), (180, 1))
assert_size_stride(primals_7, (50, ), (1, ))
assert_size_stride(primals_8, (8, 50), (50, 1))
assert_size_stride(primals_9, (8, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 10, 20, 20), (4000, 400, 20, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(buf1, primals_2, 16000, grid=grid(16000), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((4, 10, 10, 10), (1000, 100, 10, 1), torch.int8)
buf3 = empty_strided_cuda((4, 10, 10, 10), (1000, 100, 10, 1), torch.float32)
# Topologically Sorted Source Nodes: [max_pool2d, x], Original ATen: [aten.max_pool2d_with_indices, aten.relu]
triton_poi_fused_max_pool2d_with_indices_relu_1.run(buf1, buf2, buf3, 4000, grid=grid(4000), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(buf3, primals_4, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 20, 6, 6), (720, 36, 6, 1))
buf5 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
triton_poi_fused_convolution_2.run(buf5, primals_5, 2880, grid=grid(2880), stream=stream0)
del primals_5
buf6 = empty_strided_cuda((4, 20, 3, 3), (180, 9, 3, 1), torch.int8)
buf7 = empty_strided_cuda((4, 20, 3, 3), (180, 9, 3, 1), torch.float32)
buf14 = empty_strided_cuda((4, 20, 3, 3), (180, 9, 3, 1), torch.bool)
# Topologically Sorted Source Nodes: [max_pool2d_1, x_1], Original ATen: [aten.max_pool2d_with_indices, aten.relu, aten.threshold_backward]
triton_poi_fused_max_pool2d_with_indices_relu_threshold_backward_3.run(buf5, buf6, buf7, buf14, 720, grid=grid(720), stream=stream0)
buf8 = empty_strided_cuda((4, 50), (50, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf7, (4, 180), (180, 1), 0), reinterpret_tensor(primals_6, (180, 50), (1, 180), 0), out=buf8)
buf9 = buf8; del buf8 # reuse
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.relu]
triton_poi_fused_relu_4.run(buf9, primals_7, 200, grid=grid(200), stream=stream0)
del primals_7
buf10 = empty_strided_cuda((4, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_9, buf9, reinterpret_tensor(primals_8, (50, 8), (1, 50), 0), alpha=1, beta=1, out=buf10)
del primals_9
buf13 = empty_strided_cuda((4, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [log_softmax], Original ATen: [aten._log_softmax]
triton_per_fused__log_softmax_5.run(buf10, buf13, 4, 8, grid=grid(4), stream=stream0)
del buf10
return (buf13, primals_1, primals_3, primals_4, buf1, buf2, buf3, buf5, buf6, reinterpret_tensor(buf7, (4, 180), (180, 1), 0), buf9, buf13, primals_8, primals_6, buf14, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((10, 3, 5, 5), (75, 25, 5, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((10, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 3, 24, 24), (1728, 576, 24, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((20, 10, 5, 5), (250, 25, 5, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((20, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((50, 180), (180, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((50, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((8, 50), (50, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((8, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(3, 10, kernel_size=5)
self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
self.conv2_drop = nn.Dropout2d()
self.fc1 = nn.Linear(180, 50)
self.fc2 = nn.Linear(50, 8)
def forward(self, x):
x = F.relu(F.max_pool2d(self.conv1(x), 2))
x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))
x = x.view(-1, 180)
x = F.relu(self.fc1(x))
x = F.dropout(x, training=self.training)
x = self.fc2(x)
return F.log_softmax(x, dim=1)
def get_inputs():
return [torch.rand([4, 3, 24, 24])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 16000
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 400 % 10
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_relu_1(in_ptr0, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 4000
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 10
x1 = xindex // 10
x2 = xindex
tmp0 = tl.load(in_ptr0 + (2 * x0 + 40 * x1), xmask, eviction_policy=
'evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 2 * x0 + 40 * x1), xmask, eviction_policy
='evict_last')
tmp7 = tl.load(in_ptr0 + (20 + 2 * x0 + 40 * x1), xmask,
eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (21 + 2 * x0 + 40 * x1), xmask,
eviction_policy='evict_last')
tmp2 = tmp1 > tmp0
tmp3 = tl.full([1], 1, tl.int8)
tmp4 = tl.full([1], 0, tl.int8)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp6 = triton_helpers.maximum(tmp1, tmp0)
tmp8 = tmp7 > tmp6
tmp9 = tl.full([1], 2, tl.int8)
tmp10 = tl.where(tmp8, tmp9, tmp5)
tmp11 = triton_helpers.maximum(tmp7, tmp6)
tmp13 = tmp12 > tmp11
tmp14 = tl.full([1], 3, tl.int8)
tmp15 = tl.where(tmp13, tmp14, tmp10)
tmp16 = triton_helpers.maximum(tmp12, tmp11)
tmp17 = tl.full([1], 0, tl.int32)
tmp18 = triton_helpers.maximum(tmp17, tmp16)
tl.store(out_ptr0 + x2, tmp15, xmask)
tl.store(out_ptr1 + x2, tmp18, xmask)
@triton.jit
def triton_poi_fused_convolution_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 2880
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 36 % 20
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_relu_threshold_backward_3(in_ptr0,
out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK: tl.constexpr):
xnumel = 720
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 3
x1 = xindex // 3
x2 = xindex
tmp0 = tl.load(in_ptr0 + (2 * x0 + 12 * x1), xmask, eviction_policy=
'evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 2 * x0 + 12 * x1), xmask, eviction_policy
='evict_last')
tmp7 = tl.load(in_ptr0 + (6 + 2 * x0 + 12 * x1), xmask, eviction_policy
='evict_last')
tmp12 = tl.load(in_ptr0 + (7 + 2 * x0 + 12 * x1), xmask,
eviction_policy='evict_last')
tmp2 = tmp1 > tmp0
tmp3 = tl.full([1], 1, tl.int8)
tmp4 = tl.full([1], 0, tl.int8)
tmp5 = tl.where(tmp2, tmp3, tmp4)
tmp6 = triton_helpers.maximum(tmp1, tmp0)
tmp8 = tmp7 > tmp6
tmp9 = tl.full([1], 2, tl.int8)
tmp10 = tl.where(tmp8, tmp9, tmp5)
tmp11 = triton_helpers.maximum(tmp7, tmp6)
tmp13 = tmp12 > tmp11
tmp14 = tl.full([1], 3, tl.int8)
tmp15 = tl.where(tmp13, tmp14, tmp10)
tmp16 = triton_helpers.maximum(tmp12, tmp11)
tmp17 = tl.full([1], 0, tl.int32)
tmp18 = triton_helpers.maximum(tmp17, tmp16)
tmp19 = 0.0
tmp20 = tmp18 <= tmp19
tl.store(out_ptr0 + x2, tmp15, xmask)
tl.store(out_ptr1 + x2, tmp18, xmask)
tl.store(out_ptr2 + x2, tmp20, xmask)
@triton.jit
def triton_poi_fused_relu_4(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 50
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_per_fused__log_softmax_5(in_ptr0, out_ptr2, xnumel, rnumel,
XBLOCK: tl.constexpr):
xnumel = 4
RBLOCK: tl.constexpr = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 8 * x0), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, float('-inf'))
tmp4 = triton_helpers.max2(tmp3, 1)[:, None]
tmp5 = tmp0 - tmp4
tmp6 = tl_math.exp(tmp5)
tmp7 = tl.broadcast_to(tmp6, [XBLOCK, RBLOCK])
tmp9 = tl.where(xmask, tmp7, 0)
tmp10 = tl.sum(tmp9, 1)[:, None]
tmp11 = tl_math.log(tmp10)
tmp12 = tmp5 - tmp11
tl.store(out_ptr2 + (r1 + 8 * x0), tmp12, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9) = args
args.clear()
assert_size_stride(primals_1, (10, 3, 5, 5), (75, 25, 5, 1))
assert_size_stride(primals_2, (10,), (1,))
assert_size_stride(primals_3, (4, 3, 24, 24), (1728, 576, 24, 1))
assert_size_stride(primals_4, (20, 10, 5, 5), (250, 25, 5, 1))
assert_size_stride(primals_5, (20,), (1,))
assert_size_stride(primals_6, (50, 180), (180, 1))
assert_size_stride(primals_7, (50,), (1,))
assert_size_stride(primals_8, (8, 50), (50, 1))
assert_size_stride(primals_9, (8,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,
1), padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 10, 20, 20), (4000, 400, 20, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(16000)](buf1, primals_2, 16000,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((4, 10, 10, 10), (1000, 100, 10, 1),
torch.int8)
buf3 = empty_strided_cuda((4, 10, 10, 10), (1000, 100, 10, 1),
torch.float32)
triton_poi_fused_max_pool2d_with_indices_relu_1[grid(4000)](buf1,
buf2, buf3, 4000, XBLOCK=128, num_warps=4, num_stages=1)
buf4 = extern_kernels.convolution(buf3, primals_4, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 20, 6, 6), (720, 36, 6, 1))
buf5 = buf4
del buf4
triton_poi_fused_convolution_2[grid(2880)](buf5, primals_5, 2880,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_5
buf6 = empty_strided_cuda((4, 20, 3, 3), (180, 9, 3, 1), torch.int8)
buf7 = empty_strided_cuda((4, 20, 3, 3), (180, 9, 3, 1), torch.float32)
buf14 = empty_strided_cuda((4, 20, 3, 3), (180, 9, 3, 1), torch.bool)
triton_poi_fused_max_pool2d_with_indices_relu_threshold_backward_3[grid
(720)](buf5, buf6, buf7, buf14, 720, XBLOCK=256, num_warps=4,
num_stages=1)
buf8 = empty_strided_cuda((4, 50), (50, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf7, (4, 180), (180, 1), 0),
reinterpret_tensor(primals_6, (180, 50), (1, 180), 0), out=buf8)
buf9 = buf8
del buf8
triton_poi_fused_relu_4[grid(200)](buf9, primals_7, 200, XBLOCK=256,
num_warps=4, num_stages=1)
del primals_7
buf10 = empty_strided_cuda((4, 8), (8, 1), torch.float32)
extern_kernels.addmm(primals_9, buf9, reinterpret_tensor(primals_8,
(50, 8), (1, 50), 0), alpha=1, beta=1, out=buf10)
del primals_9
buf13 = empty_strided_cuda((4, 8), (8, 1), torch.float32)
triton_per_fused__log_softmax_5[grid(4)](buf10, buf13, 4, 8, XBLOCK
=1, num_warps=2, num_stages=1)
del buf10
return (buf13, primals_1, primals_3, primals_4, buf1, buf2, buf3, buf5,
buf6, reinterpret_tensor(buf7, (4, 180), (180, 1), 0), buf9, buf13,
primals_8, primals_6, buf14)
class NetNew(nn.Module):
def __init__(self):
super(NetNew, self).__init__()
self.conv1 = nn.Conv2d(3, 10, kernel_size=5)
self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
self.conv2_drop = nn.Dropout2d()
self.fc1 = nn.Linear(180, 50)
self.fc2 = nn.Linear(50, 8)
def forward(self, input_0):
primals_1 = self.conv1.weight
primals_2 = self.conv1.bias
primals_4 = self.conv2.weight
primals_5 = self.conv2.bias
primals_6 = self.fc1.weight
primals_7 = self.fc1.bias
primals_8 = self.fc2.weight
primals_9 = self.fc2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9])
return output[0]
| lzhbrian/FashionAI-1 | Net | false | 7,166 | [
"MIT"
] | 1 | 1fede16044c8a4516ba4dd6766add44d47245f6b | https://github.com/lzhbrian/FashionAI-1/tree/1fede16044c8a4516ba4dd6766add44d47245f6b | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self):
super().__init__()
self.conv1 = nn.Conv2d(3, 10, kernel_size=5)
self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
self.conv2_drop = nn.Dropout2d()
self.fc1 = nn.Linear(180, 50)
self.fc2 = nn.Linear(50, 8)
def forward(self, x):
x = F.relu(F.max_pool2d(self.conv1(x), 2))
x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))
x = x.view(-1, 180)
x = F.relu(self.fc1(x))
x = F.dropout(x, training=self.training)
x = self.fc2(x)
return F.log_softmax(x, dim=1)
def get_inputs():
return [torch.rand([4, 3, 24, 24])]
def get_init_inputs():
return []
|
DotProductAttention | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/r6/cr6neze6yovkog6kjrk5k2db63h47ozkojywfys6karxe7dlumrz.py
# Topologically Sorted Source Nodes: [attn], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attn => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%bmm, [2], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%bmm, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_0 = async_compile.triton('triton_poi_fused__softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/kj/ckjtlefzavjukjsytvkak6ek26zmzexpcbnlwelx4k5kascjxlf3.py
# Topologically Sorted Source Nodes: [attn], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attn => div, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [2], True), kwargs = {})
# %div : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_1 = async_compile.triton('triton_poi_fused__softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_5, (4, 4, 4), (16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [q], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_3, reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [k], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_3, reinterpret_tensor(primals_4, (16, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf1)
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [v], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_3, reinterpret_tensor(primals_5, (16, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del primals_2
del primals_3
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [a], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf1, (4, 4, 4), (16, 1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [attn], Original ATen: [aten._softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__softmax_0.run(buf3, buf4, 64, grid=grid(64), stream=stream0)
buf5 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [attn], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf4, buf5, 64, grid=grid(64), stream=stream0)
buf6 = buf4; del buf4 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(buf5, reinterpret_tensor(buf2, (4, 4, 4), (16, 4, 1), 0), out=buf6)
return (buf6, buf5, reinterpret_tensor(primals_1, (16, 4), (4, 1), 0), reinterpret_tensor(primals_4, (16, 4), (4, 1), 0), reinterpret_tensor(primals_5, (16, 4), (4, 1), 0), buf5, reinterpret_tensor(buf2, (4, 4, 4), (16, 1, 4), 0), reinterpret_tensor(buf0, (4, 4, 4), (16, 1, 4), 0), reinterpret_tensor(buf1, (4, 4, 4), (16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import warnings
from typing import Optional
from typing import Tuple
import torch.nn as nn
class DotProductAttention(nn.Module):
"""DotProductAttention.
.. math::
\\mathrm{DotProductAttention}(Q, K, V) &=& \\mathrm{softmax}(qk^T) v
q &=& QW_1 + b_1
k &=& KW_2 + b_2
v &=& VW_3 + b_3
Args:
qdim: dimension of the model, i.e., dimension of Q
hidden_dim: dimension of hidden layer, i.e., dimension of q, k, v. Default: 512
output_dim: dimension of output layer, i.e., dimension of output. Default: None
dropout: a Dropout layer on attn_output_weights. Default: 0.0.
transform: q = Q, k = K, v = V if it is False. Default: True
bias: add bias as module parameter. Default: True.
same_embd: W1 = W2 = W3, b1 = b2 = b3 if it is True. Default: True
add_bias_kv: add bias to the key and value sequences at dim=0.
kdim: total number of features in key. Default: None.
vdim: total number of features in key. Default: None.
Note: if kdim and vdim are None, they will be set to embed_dim such that
query, key, and value have the same number of features.
Examples::
>>> attn = DotProductAttention(query_dim)
>>> attn_output, attn_output_weights = attn(query, key, value)
"""
def __init__(self, qdim: 'int', output_dim: 'Optional[int]'=None,
dropout: 'float'=0.0, transform: 'bool'=True, bias: 'bool'=True,
same_embd: 'bool'=True, add_bias_kv: 'Optional[bool]'=None, kdim:
'Optional[int]'=None, vdim: 'Optional[int]'=None, batch_first:
'bool'=True, scaled: 'bool'=False) ->None:
super(DotProductAttention, self).__init__()
self.qdim: 'int' = qdim
self.transform: 'bool' = transform
self.bias: 'bool' = bias
self.same_embd: 'bool' = same_embd
self.kdim: 'int' = kdim if kdim is not None else self.qdim
self.vdim: 'int' = vdim if vdim is not None else self.qdim
self.output_dim: 'int' = (output_dim if output_dim is not None else
self.vdim)
if self.same_embd and (self.qdim != self.kdim or self.qdim != self.vdim
):
raise AssertionError(
'qdim, kdim, vdim should be the same dimensions if same_embd is True'
)
self.add_bias_kv: 'bool' = (add_bias_kv if add_bias_kv is not None else
self.bias)
if self.same_embd and self.bias != self.add_bias_kv:
raise AssertionError(
'bias and add_bias_kv should be the same if same_embd is True')
self.batch_first: 'bool' = batch_first
self.scaled: 'bool' = scaled
self.fc_q: 'nn.Module' = nn.Linear(self.qdim, self.output_dim, bias
=bias)
self.fc_k: 'nn.Module'
self.fc_v: 'nn.Module'
if self.same_embd:
self.fc_k = self.fc_q
self.fc_v = self.fc_k
else:
self.fc_k = nn.Linear(self.kdim, self.output_dim, bias=self.
add_bias_kv)
self.fc_v = nn.Linear(self.vdim, self.output_dim, bias=self.
add_bias_kv)
self.dropout: 'nn.Module' = nn.Dropout(p=dropout)
self.softmax: 'nn.Module' = nn.Softmax(dim=2)
def forward(self, query: 'torch.Tensor', key: 'torch.Tensor', value:
'torch.Tensor', key_padding_mask: 'Optional[torch.Tensor]'=None,
attn_mask: 'Optional[torch.Tensor]'=None) ->Tuple[torch.Tensor,
torch.Tensor]:
if key_padding_mask is not None:
warnings.warn(
"'key_padding_mask' in 'DotProductAttention' is currently an experimental version.When you use this, please check if this is working correctly or not very carefully."
)
if not self.batch_first:
query = torch.transpose(query, 0, 1)
key = torch.transpose(key, 0, 1)
value = torch.transpose(value, 0, 1)
bsz, _tgt_len, _ = query.size()
q = self.fc_q(query)
q = self.dropout(q)
k = self.fc_k(key)
k = self.dropout(k)
v = self.fc_v(value)
v = self.dropout(v)
if k.size() != v.size():
raise AssertionError(
'The sizes of key and value should be the same.')
src_len = k.size(1)
if key_padding_mask is not None:
if key_padding_mask.size(0) != bsz:
raise AssertionError(
'The first dimension of kay padding mask size must be the same as batch size'
)
if key_padding_mask.size(1) != src_len:
raise AssertionError(
'The second dimension of key padding mask size must be the same as source length'
)
a = torch.bmm(q, torch.transpose(k, 1, 2))
if self.scaled:
a /= math.sqrt(self.output_dim)
if attn_mask is not None:
a += attn_mask
if key_padding_mask is not None:
a = a.masked_fill(key_padding_mask.unsqueeze(1), float('-inf'))
attn = self.softmax(a)
output = torch.bmm(attn, v)
if not self.batch_first:
output = torch.transpose(output, 0, 1)
return output, attn
def generate_square_subsequent_mask(self, sz: 'int') ->torch.Tensor:
"""Generate a square mask for the sequence. The masked positions are filled with float('-inf').
Unmasked positions are filled with float(0.0).
"""
mask = (torch.triu(torch.ones(sz, sz)) == 1).transpose(0, 1)
mask = mask.float().masked_fill(mask == 0, float('-inf')).masked_fill(
mask == 1, float(0.0))
return mask
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4]), torch.rand([4, 4, 4])
]
def get_init_inputs():
return [[], {'qdim': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
from typing import Optional
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_5, (4, 4, 4), (16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_3, reinterpret_tensor(primals_1, (16,
4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
buf1 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_3, reinterpret_tensor(primals_4, (16,
4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf1)
buf2 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_3, reinterpret_tensor(primals_5, (16,
4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf2)
del primals_2
del primals_3
buf3 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf0, (4, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf1, (4, 4, 4), (16, 1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__softmax_0[grid(64)](buf3, buf4, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf5 = buf3
del buf3
triton_poi_fused__softmax_1[grid(64)](buf4, buf5, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf6 = buf4
del buf4
extern_kernels.bmm(buf5, reinterpret_tensor(buf2, (4, 4, 4), (16, 4,
1), 0), out=buf6)
return buf6, buf5, reinterpret_tensor(primals_1, (16, 4), (4, 1), 0
), reinterpret_tensor(primals_4, (16, 4), (4, 1), 0
), reinterpret_tensor(primals_5, (16, 4), (4, 1), 0
), buf5, reinterpret_tensor(buf2, (4, 4, 4), (16, 1, 4), 0
), reinterpret_tensor(buf0, (4, 4, 4), (16, 1, 4), 0
), reinterpret_tensor(buf1, (4, 4, 4), (16, 4, 1), 0)
class DotProductAttentionNew(nn.Module):
"""DotProductAttention.
.. math::
\\mathrm{DotProductAttention}(Q, K, V) &=& \\mathrm{softmax}(qk^T) v
q &=& QW_1 + b_1
k &=& KW_2 + b_2
v &=& VW_3 + b_3
Args:
qdim: dimension of the model, i.e., dimension of Q
hidden_dim: dimension of hidden layer, i.e., dimension of q, k, v. Default: 512
output_dim: dimension of output layer, i.e., dimension of output. Default: None
dropout: a Dropout layer on attn_output_weights. Default: 0.0.
transform: q = Q, k = K, v = V if it is False. Default: True
bias: add bias as module parameter. Default: True.
same_embd: W1 = W2 = W3, b1 = b2 = b3 if it is True. Default: True
add_bias_kv: add bias to the key and value sequences at dim=0.
kdim: total number of features in key. Default: None.
vdim: total number of features in key. Default: None.
Note: if kdim and vdim are None, they will be set to embed_dim such that
query, key, and value have the same number of features.
Examples::
>>> attn = DotProductAttention(query_dim)
>>> attn_output, attn_output_weights = attn(query, key, value)
"""
def __init__(self, qdim: 'int', output_dim: 'Optional[int]'=None,
dropout: 'float'=0.0, transform: 'bool'=True, bias: 'bool'=True,
same_embd: 'bool'=True, add_bias_kv: 'Optional[bool]'=None, kdim:
'Optional[int]'=None, vdim: 'Optional[int]'=None, batch_first:
'bool'=True, scaled: 'bool'=False) ->None:
super(DotProductAttentionNew, self).__init__()
self.qdim: 'int' = qdim
self.transform: 'bool' = transform
self.bias: 'bool' = bias
self.same_embd: 'bool' = same_embd
self.kdim: 'int' = kdim if kdim is not None else self.qdim
self.vdim: 'int' = vdim if vdim is not None else self.qdim
self.output_dim: 'int' = (output_dim if output_dim is not None else
self.vdim)
if self.same_embd and (self.qdim != self.kdim or self.qdim != self.vdim
):
raise AssertionError(
'qdim, kdim, vdim should be the same dimensions if same_embd is True'
)
self.add_bias_kv: 'bool' = (add_bias_kv if add_bias_kv is not None else
self.bias)
if self.same_embd and self.bias != self.add_bias_kv:
raise AssertionError(
'bias and add_bias_kv should be the same if same_embd is True')
self.batch_first: 'bool' = batch_first
self.scaled: 'bool' = scaled
self.fc_q: 'nn.Module' = nn.Linear(self.qdim, self.output_dim, bias
=bias)
self.fc_k: 'nn.Module'
self.fc_v: 'nn.Module'
if self.same_embd:
self.fc_k = self.fc_q
self.fc_v = self.fc_k
else:
self.fc_k = nn.Linear(self.kdim, self.output_dim, bias=self.
add_bias_kv)
self.fc_v = nn.Linear(self.vdim, self.output_dim, bias=self.
add_bias_kv)
self.dropout: 'nn.Module' = nn.Dropout(p=dropout)
self.softmax: 'nn.Module' = nn.Softmax(dim=2)
def generate_square_subsequent_mask(self, sz: 'int') ->torch.Tensor:
"""Generate a square mask for the sequence. The masked positions are filled with float('-inf').
Unmasked positions are filled with float(0.0).
"""
mask = (torch.triu(torch.ones(sz, sz)) == 1).transpose(0, 1)
mask = mask.float().masked_fill(mask == 0, float('-inf')).masked_fill(
mask == 1, float(0.0))
return mask
def forward(self, input_0, input_1, input_2):
primals_2 = self.fc_q.weight
primals_3 = self.fc_q.bias
primals_1 = input_0
primals_4 = input_1
primals_5 = input_2
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0], output[1]
| masanorihirano/pytorch_extra_mhirano | DotProductAttention | false | 7,167 | [
"MIT"
] | 1 | d19e07445567c069793b7ca1a22a846d7cbce58d | https://github.com/masanorihirano/pytorch_extra_mhirano/tree/d19e07445567c069793b7ca1a22a846d7cbce58d | import math
import torch
import warnings
from typing import Optional
from typing import Tuple
import torch.nn as nn
class Model(nn.Module):
"""DotProductAttention.
.. math::
\\mathrm{DotProductAttention}(Q, K, V) &=& \\mathrm{softmax}(qk^T) v
q &=& QW_1 + b_1
k &=& KW_2 + b_2
v &=& VW_3 + b_3
Args:
qdim: dimension of the model, i.e., dimension of Q
hidden_dim: dimension of hidden layer, i.e., dimension of q, k, v. Default: 512
output_dim: dimension of output layer, i.e., dimension of output. Default: None
dropout: a Dropout layer on attn_output_weights. Default: 0.0.
transform: q = Q, k = K, v = V if it is False. Default: True
bias: add bias as module parameter. Default: True.
same_embd: W1 = W2 = W3, b1 = b2 = b3 if it is True. Default: True
add_bias_kv: add bias to the key and value sequences at dim=0.
kdim: total number of features in key. Default: None.
vdim: total number of features in key. Default: None.
Note: if kdim and vdim are None, they will be set to embed_dim such that
query, key, and value have the same number of features.
Examples::
>>> attn = DotProductAttention(query_dim)
>>> attn_output, attn_output_weights = attn(query, key, value)
"""
def __init__(self, qdim: 'int', output_dim: 'Optional[int]'=None,
dropout: 'float'=0.0, transform: 'bool'=True, bias: 'bool'=True,
same_embd: 'bool'=True, add_bias_kv: 'Optional[bool]'=None, kdim:
'Optional[int]'=None, vdim: 'Optional[int]'=None, batch_first:
'bool'=True, scaled: 'bool'=False) ->None:
super().__init__()
self.qdim: 'int' = qdim
self.transform: 'bool' = transform
self.bias: 'bool' = bias
self.same_embd: 'bool' = same_embd
self.kdim: 'int' = kdim if kdim is not None else self.qdim
self.vdim: 'int' = vdim if vdim is not None else self.qdim
self.output_dim: 'int' = (output_dim if output_dim is not None else
self.vdim)
if self.same_embd and (self.qdim != self.kdim or self.qdim != self.vdim
):
raise AssertionError(
'qdim, kdim, vdim should be the same dimensions if same_embd is True'
)
self.add_bias_kv: 'bool' = (add_bias_kv if add_bias_kv is not None else
self.bias)
if self.same_embd and self.bias != self.add_bias_kv:
raise AssertionError(
'bias and add_bias_kv should be the same if same_embd is True')
self.batch_first: 'bool' = batch_first
self.scaled: 'bool' = scaled
self.fc_q: 'nn.Module' = nn.Linear(self.qdim, self.output_dim, bias
=bias)
self.fc_k: 'nn.Module'
self.fc_v: 'nn.Module'
if self.same_embd:
self.fc_k = self.fc_q
self.fc_v = self.fc_k
else:
self.fc_k = nn.Linear(self.kdim, self.output_dim, bias=self.
add_bias_kv)
self.fc_v = nn.Linear(self.vdim, self.output_dim, bias=self.
add_bias_kv)
self.dropout: 'nn.Module' = nn.Dropout(p=dropout)
self.softmax: 'nn.Module' = nn.Softmax(dim=2)
def forward(self, query: 'torch.Tensor', key: 'torch.Tensor', value:
'torch.Tensor', key_padding_mask: 'Optional[torch.Tensor]'=None,
attn_mask: 'Optional[torch.Tensor]'=None) ->Tuple[torch.Tensor,
torch.Tensor]:
if key_padding_mask is not None:
warnings.warn(
"'key_padding_mask' in 'DotProductAttention' is currently an experimental version.When you use this, please check if this is working correctly or not very carefully."
)
if not self.batch_first:
query = torch.transpose(query, 0, 1)
key = torch.transpose(key, 0, 1)
value = torch.transpose(value, 0, 1)
bsz, _tgt_len, _ = query.size()
# ... truncated (>4000 chars) for memory efficiency |
KLDivLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/je/cjecll24hlvz7r62a3tl4rp6ndcpgavf4ti4czumksi7lopih6y6.py
# Topologically Sorted Source Nodes: [kl_div, log_input], Original ATen: [aten.xlogy, aten.log, aten.mul, aten.sub, aten.mean]
# Source node to ATen node mapping:
# kl_div => eq, full_default, full_default_1, isnan, log_1, mean, mul, mul_1, sub, where, where_1
# log_input => log
# Graph fragment:
# %isnan : [num_users=1] = call_function[target=torch.ops.aten.isnan.default](args = (%arg1_1,), kwargs = {})
# %full_default_1 : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], nan), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %eq : [num_users=1] = call_function[target=torch.ops.aten.eq.Scalar](args = (%arg1_1, 0), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %log_1 : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%arg1_1,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg1_1, %log_1), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%eq, %full_default, %mul_1), kwargs = {})
# %where_1 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%isnan, %full_default_1, %where), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%arg0_1,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg1_1, %log), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%where_1, %mul), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sub,), kwargs = {})
triton_per_fused_log_mean_mul_sub_xlogy_0 = async_compile.triton('triton_per_fused_log_mean_mul_sub_xlogy_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_log_mean_mul_sub_xlogy_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_log_mean_mul_sub_xlogy_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp9 = tl.load(in_ptr1 + (r0), None)
tmp1 = libdevice.isnan(tmp0).to(tl.int1)
tmp2 = 0.0
tmp3 = tmp0 == tmp2
tmp4 = tl_math.log(tmp0)
tmp5 = tmp0 * tmp4
tmp6 = tl.where(tmp3, tmp2, tmp5)
tmp7 = float("nan")
tmp8 = tl.where(tmp1, tmp7, tmp6)
tmp10 = tl_math.log(tmp9)
tmp11 = tmp0 * tmp10
tmp12 = tmp8 - tmp11
tmp13 = tl.broadcast_to(tmp12, [RBLOCK])
tmp15 = triton_helpers.promote_to_tensor(tl.sum(tmp13, 0))
tmp16 = 256.0
tmp17 = tmp15 / tmp16
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp17, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [kl_div, log_input], Original ATen: [aten.xlogy, aten.log, aten.mul, aten.sub, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_log_mean_mul_sub_xlogy_0.run(buf1, arg1_1, arg0_1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from typing import Optional
from torch.nn import functional as F
from torch.nn.modules.loss import _Loss
class KLDivLoss(_Loss):
def __init__(self, size_average: 'Optional[bool]'=None, reduce:
'Optional[bool]'=None, reduction: 'str'='mean') ->None:
super(KLDivLoss, self).__init__(size_average, reduce, reduction)
def forward(self, inputs: 'torch.Tensor', targets: 'torch.Tensor'
) ->torch.Tensor:
log_input = torch.log(inputs)
return F.kl_div(log_input, targets, reduction=self.reduction)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from typing import Optional
from torch.nn.modules.loss import _Loss
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_log_mean_mul_sub_xlogy_0(in_out_ptr0, in_ptr0, in_ptr1,
xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp9 = tl.load(in_ptr1 + r0, None)
tmp1 = libdevice.isnan(tmp0).to(tl.int1)
tmp2 = 0.0
tmp3 = tmp0 == tmp2
tmp4 = tl_math.log(tmp0)
tmp5 = tmp0 * tmp4
tmp6 = tl.where(tmp3, tmp2, tmp5)
tmp7 = float('nan')
tmp8 = tl.where(tmp1, tmp7, tmp6)
tmp10 = tl_math.log(tmp9)
tmp11 = tmp0 * tmp10
tmp12 = tmp8 - tmp11
tmp13 = tl.broadcast_to(tmp12, [RBLOCK])
tmp15 = triton_helpers.promote_to_tensor(tl.sum(tmp13, 0))
tmp16 = 256.0
tmp17 = tmp15 / tmp16
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp17, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_per_fused_log_mean_mul_sub_xlogy_0[grid(1)](buf1, arg1_1,
arg0_1, 1, 256, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf1,
class KLDivLossNew(_Loss):
def __init__(self, size_average: 'Optional[bool]'=None, reduce:
'Optional[bool]'=None, reduction: 'str'='mean') ->None:
super(KLDivLossNew, self).__init__(size_average, reduce, reduction)
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| masanorihirano/pytorch_extra_mhirano | KLDivLoss | false | 7,168 | [
"MIT"
] | 1 | d19e07445567c069793b7ca1a22a846d7cbce58d | https://github.com/masanorihirano/pytorch_extra_mhirano/tree/d19e07445567c069793b7ca1a22a846d7cbce58d | import torch
from typing import Optional
from torch.nn import functional as F
from torch.nn.modules.loss import _Loss
class Model(_Loss):
def __init__(self, size_average: 'Optional[bool]'=None, reduce:
'Optional[bool]'=None, reduction: 'str'='mean') ->None:
super().__init__(size_average, reduce, reduction)
def forward(self, inputs: 'torch.Tensor', targets: 'torch.Tensor'
) ->torch.Tensor:
log_input = torch.log(inputs)
return F.kl_div(log_input, targets, reduction=self.reduction)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
VGGBase | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/f7/cf7tayhctr3m6ezk7xezotpdlc5h4drokdkbz4vy2pfkbdxnmn4q.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_0 = async_compile.triton('triton_poi_fused_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 192
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 3
y1 = (yindex // 3)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (3*x2) + (27*y1)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/5b/c5brnjme4e4oybuabwsko4vuljormwjqoawce7jgxo5fbkhzx55r.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_1 = async_compile.triton('triton_poi_fused_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4096], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 12
xnumel = 4096
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
x2 = xindex
y3 = yindex
y0 = yindex % 3
y1 = (yindex // 3)
tmp0 = tl.load(in_ptr0 + (x2 + (4096*y3)), ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (3*x2) + (12288*y1)), tmp0, ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/xq/cxq75w43anllid5ys7ss3yyizuoeph3vvaqlvm5lo434hrywtyle.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_2 = async_compile.triton('triton_poi_fused_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 4096
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 64
y1 = (yindex // 64)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (64*x2) + (576*y1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/nw/cnwm6ljuusoqjcwr2jdx6p2ue7ldghxjdr3oe62stiuqhsboiczy.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_3 = async_compile.triton('triton_poi_fused_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8192, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_3(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 8192
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 64
y1 = (yindex // 64)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (64*x2) + (576*y1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/32/c32xiwptfqtyhbnde262mvq5tzywzo6zquurttkv7sztqnze6yni.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_4 = async_compile.triton('triton_poi_fused_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16384, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_4(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16384
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 128
y1 = (yindex // 128)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (128*x2) + (1152*y1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/jj/cjjz4tpbucpuc3faa2ky32crfwhb5fbnssd6o2yfkgdcjg2acfmo.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_5 = async_compile.triton('triton_poi_fused_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32768, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_5(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 32768
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 128
y1 = (yindex // 128)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (128*x2) + (1152*y1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/tg/ctgdsxjd3rciejxtjvi3y2w5fmmggh5lm3mivuygvkdzeb3zulmc.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_6 = async_compile.triton('triton_poi_fused_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[65536, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_6(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 65536
xnumel = 9
yoffset = (tl.program_id(1) + tl.program_id(2) * tl.num_programs(1)) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 256
y1 = (yindex // 256)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (256*x2) + (2304*y1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/e7/ce7jqsdrj5poslb2hpufqd2wdux5xiab5n2auqal3ztzvkzrmnzl.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_7 = async_compile.triton('triton_poi_fused_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[131072, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_7(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 131072
xnumel = 9
yoffset = (tl.program_id(1) + tl.program_id(2) * tl.num_programs(1)) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 256
y1 = (yindex // 256)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (256*x2) + (2304*y1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ks/ckso6iiq5yfqfxmx7ilr6ufrmz6mlkiy75pexzhyf3ierq4pu3zl.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_8 = async_compile.triton('triton_poi_fused_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[262144, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_8', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_8(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 262144
xnumel = 9
yoffset = (tl.program_id(1) + tl.program_id(2) * tl.num_programs(1)) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 512
y1 = (yindex // 512)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (512*x2) + (4608*y1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/cq/ccq66rrhrzjmgxnrmkqjfjou7btyc5dncveqmqkrdoivqkmduchd.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_9 = async_compile.triton('triton_poi_fused_9', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[524288, 16], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_9', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_9(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 524288
xnumel = 9
yoffset = (tl.program_id(1) + tl.program_id(2) * tl.num_programs(1)) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 512
y1 = (yindex // 512)
tmp0 = tl.load(in_ptr0 + (x2 + (9*y3)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (512*x2) + (4608*y1)), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/y7/cy74ayecev2pcofz3fyu6lc473nqeaato7assx62kzcpdkdyzi7o.py
# Topologically Sorted Source Nodes: [conv2d, out], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d => convolution
# out => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_10 = async_compile.triton('triton_poi_fused_convolution_relu_10', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1048576],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_10', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_10(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1048576
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/mv/cmvofpunraye55pqf22y3ewvph2z6nefokvusriez7hf4qcucdfo.py
# Topologically Sorted Source Nodes: [out_2], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# out_2 => getitem, getitem_1
# Graph fragment:
# %getitem : [num_users=2] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 0), kwargs = {})
# %getitem_1 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_11 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_11', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[262144],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i8', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_11', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_11(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 262144
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 64
x1 = (xindex // 64) % 32
x2 = (xindex // 2048)
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (128*x1) + (8192*x2)), None)
tmp1 = tl.load(in_ptr0 + (64 + x0 + (128*x1) + (8192*x2)), None)
tmp3 = tl.load(in_ptr0 + (4096 + x0 + (128*x1) + (8192*x2)), None)
tmp5 = tl.load(in_ptr0 + (4160 + x0 + (128*x1) + (8192*x2)), None)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + (x3), tmp6, None)
tl.store(out_ptr1 + (x3), tmp16, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/n3/cn34mbt2rtob3eeqb7butchvtwaa2lxs5ritiirymjwyzcwqeits.py
# Topologically Sorted Source Nodes: [conv2d_2, out_3], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_2 => convolution_2
# out_3 => relu_2
# Graph fragment:
# %convolution_2 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%getitem, %primals_6, %primals_7, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_2,), kwargs = {})
triton_poi_fused_convolution_relu_12 = async_compile.triton('triton_poi_fused_convolution_relu_12', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[524288],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_12', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_12(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 524288
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 128
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/il/cilq2hip74d6rz7ttvmpmzknbqn3td7uoov3rzjb5ny3apynoqme.py
# Topologically Sorted Source Nodes: [out_5], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# out_5 => getitem_2, getitem_3
# Graph fragment:
# %getitem_2 : [num_users=2] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_1, 0), kwargs = {})
# %getitem_3 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_1, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_13 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_13', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[131072],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i8', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_13', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_13(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 131072
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 128
x1 = (xindex // 128) % 16
x2 = (xindex // 2048)
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (256*x1) + (8192*x2)), None)
tmp1 = tl.load(in_ptr0 + (128 + x0 + (256*x1) + (8192*x2)), None)
tmp3 = tl.load(in_ptr0 + (4096 + x0 + (256*x1) + (8192*x2)), None)
tmp5 = tl.load(in_ptr0 + (4224 + x0 + (256*x1) + (8192*x2)), None)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + (x3), tmp6, None)
tl.store(out_ptr1 + (x3), tmp16, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/r4/cr4cxr5slxie5num5fkjya5y6p2mpesokrymomcbss4ipccdadwk.py
# Topologically Sorted Source Nodes: [conv2d_4, out_6], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_4 => convolution_4
# out_6 => relu_4
# Graph fragment:
# %convolution_4 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%getitem_2, %primals_10, %primals_11, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_4 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_4,), kwargs = {})
triton_poi_fused_convolution_relu_14 = async_compile.triton('triton_poi_fused_convolution_relu_14', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[262144],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_14', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_14(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 262144
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 256
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/n3/cn35qanq7ew2y4riv4ein355sody4dyznrtk6o5akgf2oqgx5ok7.py
# Topologically Sorted Source Nodes: [out_9], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# out_9 => getitem_4, getitem_5
# Graph fragment:
# %getitem_4 : [num_users=2] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_2, 0), kwargs = {})
# %getitem_5 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_2, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_15 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_15', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[65536],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i8', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_15', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_15(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 65536
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 256
x1 = (xindex // 256) % 8
x2 = (xindex // 2048)
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (512*x1) + (8192*x2)), None)
tmp1 = tl.load(in_ptr0 + (256 + x0 + (512*x1) + (8192*x2)), None)
tmp3 = tl.load(in_ptr0 + (4096 + x0 + (512*x1) + (8192*x2)), None)
tmp5 = tl.load(in_ptr0 + (4352 + x0 + (512*x1) + (8192*x2)), None)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + (x3), tmp6, None)
tl.store(out_ptr1 + (x3), tmp16, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/63/c63ymadmqa5pewt6lz2e5vbnqla654yqubhkwemi5viikn2tjwlb.py
# Topologically Sorted Source Nodes: [conv2d_7, out_10], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_7 => convolution_7
# out_10 => relu_7
# Graph fragment:
# %convolution_7 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%getitem_4, %primals_16, %primals_17, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_7 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_7,), kwargs = {})
triton_poi_fused_convolution_relu_16 = async_compile.triton('triton_poi_fused_convolution_relu_16', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[131072],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_16', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_16(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 131072
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 512
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/zu/czunwyy22bkt66zyeary3r6wtcheigfh75hfciirz6pkqyjbo5yl.py
# Topologically Sorted Source Nodes: [conv2d_9, out_12], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_9 => convolution_9
# out_12 => relu_9
# Graph fragment:
# %convolution_9 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_8, %primals_20, %primals_21, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_9 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_9,), kwargs = {})
triton_poi_fused_convolution_relu_17 = async_compile.triton('triton_poi_fused_convolution_relu_17', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048, 64], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_17', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_17(in_ptr0, in_ptr1, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 2048
xnumel = 64
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 512
y1 = (yindex // 512)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (512*x2) + (32768*y1)), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (y0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1, 1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(out_ptr0 + (x2 + (64*y3)), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/bz/cbzo2gj6jhtht3ai6xpbsoye3rtape6hpo2rq4zzug767jhtvlrx.py
# Topologically Sorted Source Nodes: [out_13], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# out_13 => getitem_6, getitem_7
# Graph fragment:
# %getitem_6 : [num_users=2] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_3, 0), kwargs = {})
# %getitem_7 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_3, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_18 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_18', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048, 16], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i8', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_18', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_18(in_ptr0, out_ptr0, out_ptr1, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 2048
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex % 4
x3 = (xindex // 4)
y4 = yindex
x5 = xindex
y0 = yindex % 512
y1 = (yindex // 512)
tmp0 = tl.load(in_ptr0 + ((2*x2) + (16*x3) + (64*y4)), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (2*x2) + (16*x3) + (64*y4)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (8 + (2*x2) + (16*x3) + (64*y4)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (9 + (2*x2) + (16*x3) + (64*y4)), xmask, eviction_policy='evict_last')
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1, 1], 1, tl.int8)
tmp9 = tl.full([1, 1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1, 1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1, 1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + (y0 + (512*x5) + (8192*y1)), tmp6, xmask)
tl.store(out_ptr1 + (y0 + (512*x5) + (8192*y1)), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/pq/cpqwtybzwrjxjgxnzovhuhgkbi64boj6znsrze46xhxgut5r5rks.py
# Topologically Sorted Source Nodes: [conv2d_10, out_14], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_10 => convolution_10
# out_14 => relu_10
# Graph fragment:
# %convolution_10 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%getitem_6, %primals_22, %primals_23, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_10 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_10,), kwargs = {})
triton_poi_fused_convolution_relu_19 = async_compile.triton('triton_poi_fused_convolution_relu_19', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32768],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_19', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_19(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32768
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 512
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/g6/cg64wx5bddwxgg5xvvugg3wdo2tuwcmeybxsisjz2myhpd3oii5q.py
# Topologically Sorted Source Nodes: [out_17], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# out_17 => getitem_8, getitem_9
# Graph fragment:
# %getitem_8 : [num_users=2] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_4, 0), kwargs = {})
# %getitem_9 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_4, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_20 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_20', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32768],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i8', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_20', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 9, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_20(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 32768
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = (xindex // 2048) % 4
x1 = (xindex // 512) % 4
x6 = xindex
tmp0 = (-1) + x2
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = (-1) + x1
tmp7 = tmp6 >= tmp1
tmp8 = tmp6 < tmp3
tmp9 = tmp7 & tmp8
tmp10 = tmp5 & tmp9
tmp11 = tl.load(in_ptr0 + ((-2560) + x6), tmp10, other=float("-inf"))
tmp12 = x1
tmp13 = tmp12 >= tmp1
tmp14 = tmp12 < tmp3
tmp15 = tmp13 & tmp14
tmp16 = tmp5 & tmp15
tmp17 = tl.load(in_ptr0 + ((-2048) + x6), tmp16, other=float("-inf"))
tmp18 = triton_helpers.maximum(tmp17, tmp11)
tmp19 = 1 + x1
tmp20 = tmp19 >= tmp1
tmp21 = tmp19 < tmp3
tmp22 = tmp20 & tmp21
tmp23 = tmp5 & tmp22
tmp24 = tl.load(in_ptr0 + ((-1536) + x6), tmp23, other=float("-inf"))
tmp25 = triton_helpers.maximum(tmp24, tmp18)
tmp26 = x2
tmp27 = tmp26 >= tmp1
tmp28 = tmp26 < tmp3
tmp29 = tmp27 & tmp28
tmp30 = tmp29 & tmp9
tmp31 = tl.load(in_ptr0 + ((-512) + x6), tmp30, other=float("-inf"))
tmp32 = triton_helpers.maximum(tmp31, tmp25)
tmp33 = tmp29 & tmp15
tmp34 = tl.load(in_ptr0 + (x6), tmp33, other=float("-inf"))
tmp35 = triton_helpers.maximum(tmp34, tmp32)
tmp36 = tmp29 & tmp22
tmp37 = tl.load(in_ptr0 + (512 + x6), tmp36, other=float("-inf"))
tmp38 = triton_helpers.maximum(tmp37, tmp35)
tmp39 = 1 + x2
tmp40 = tmp39 >= tmp1
tmp41 = tmp39 < tmp3
tmp42 = tmp40 & tmp41
tmp43 = tmp42 & tmp9
tmp44 = tl.load(in_ptr0 + (1536 + x6), tmp43, other=float("-inf"))
tmp45 = triton_helpers.maximum(tmp44, tmp38)
tmp46 = tmp42 & tmp15
tmp47 = tl.load(in_ptr0 + (2048 + x6), tmp46, other=float("-inf"))
tmp48 = triton_helpers.maximum(tmp47, tmp45)
tmp49 = tmp42 & tmp22
tmp50 = tl.load(in_ptr0 + (2560 + x6), tmp49, other=float("-inf"))
tmp51 = triton_helpers.maximum(tmp50, tmp48)
tmp52 = tmp17 > tmp11
tmp53 = tl.full([1], 1, tl.int8)
tmp54 = tl.full([1], 0, tl.int8)
tmp55 = tl.where(tmp52, tmp53, tmp54)
tmp56 = tmp24 > tmp18
tmp57 = tl.full([1], 2, tl.int8)
tmp58 = tl.where(tmp56, tmp57, tmp55)
tmp59 = tmp31 > tmp25
tmp60 = tl.full([1], 3, tl.int8)
tmp61 = tl.where(tmp59, tmp60, tmp58)
tmp62 = tmp34 > tmp32
tmp63 = tl.full([1], 4, tl.int8)
tmp64 = tl.where(tmp62, tmp63, tmp61)
tmp65 = tmp37 > tmp35
tmp66 = tl.full([1], 5, tl.int8)
tmp67 = tl.where(tmp65, tmp66, tmp64)
tmp68 = tmp44 > tmp38
tmp69 = tl.full([1], 6, tl.int8)
tmp70 = tl.where(tmp68, tmp69, tmp67)
tmp71 = tmp47 > tmp45
tmp72 = tl.full([1], 7, tl.int8)
tmp73 = tl.where(tmp71, tmp72, tmp70)
tmp74 = tmp50 > tmp48
tmp75 = tl.full([1], 8, tl.int8)
tmp76 = tl.where(tmp74, tmp75, tmp73)
tl.store(out_ptr0 + (x6), tmp51, None)
tl.store(out_ptr1 + (x6), tmp76, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/j4/cj4skfvetxhoc7uzi7rl2fedifxp4uvrfozvckid3ugnt2vuch3n.py
# Topologically Sorted Source Nodes: [conv2d_13, out_18], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# conv2d_13 => convolution_13
# out_18 => relu_13
# Graph fragment:
# %convolution_13 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%getitem_8, %primals_28, %primals_29, [1, 1], [6, 6], [6, 6], False, [0, 0], 1), kwargs = {})
# %relu_13 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_13,), kwargs = {})
triton_poi_fused_convolution_relu_21 = async_compile.triton('triton_poi_fused_convolution_relu_21', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[65536],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_21', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_21(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 65536
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 1024
tmp0 = tl.load(in_out_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/tr/ctrhf6y6tp7beclzz7ocdp4ysczz3oyym47rdpqgsowyowvnsrd6.py
# Topologically Sorted Source Nodes: [conv2d_14, conv7_feats], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# conv2d_14 => convolution_14
# conv7_feats => relu_14
# Graph fragment:
# %convolution_14 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_13, %primals_30, %primals_31, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_14 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_14,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_14, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_22 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_22', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096, 16], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*i1', 4: 'i32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_22', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_22(in_ptr0, in_ptr1, out_ptr0, out_ptr1, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 4096
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 1024
y1 = (yindex // 1024)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (1024*x2) + (16384*y1)), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (y0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1, 1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x2 + (16*y3)), tmp4, xmask)
tl.store(out_ptr1 + (y0 + (1024*x2) + (16384*y1)), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14, primals_15, primals_16, primals_17, primals_18, primals_19, primals_20, primals_21, primals_22, primals_23, primals_24, primals_25, primals_26, primals_27, primals_28, primals_29, primals_30, primals_31 = args
args.clear()
assert_size_stride(primals_1, (64, 3, 3, 3), (27, 9, 3, 1))
assert_size_stride(primals_2, (64, ), (1, ))
assert_size_stride(primals_3, (4, 3, 64, 64), (12288, 4096, 64, 1))
assert_size_stride(primals_4, (64, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_5, (64, ), (1, ))
assert_size_stride(primals_6, (128, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_7, (128, ), (1, ))
assert_size_stride(primals_8, (128, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_9, (128, ), (1, ))
assert_size_stride(primals_10, (256, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_11, (256, ), (1, ))
assert_size_stride(primals_12, (256, 256, 3, 3), (2304, 9, 3, 1))
assert_size_stride(primals_13, (256, ), (1, ))
assert_size_stride(primals_14, (256, 256, 3, 3), (2304, 9, 3, 1))
assert_size_stride(primals_15, (256, ), (1, ))
assert_size_stride(primals_16, (512, 256, 3, 3), (2304, 9, 3, 1))
assert_size_stride(primals_17, (512, ), (1, ))
assert_size_stride(primals_18, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_19, (512, ), (1, ))
assert_size_stride(primals_20, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_21, (512, ), (1, ))
assert_size_stride(primals_22, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_23, (512, ), (1, ))
assert_size_stride(primals_24, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_25, (512, ), (1, ))
assert_size_stride(primals_26, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_27, (512, ), (1, ))
assert_size_stride(primals_28, (1024, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_29, (1024, ), (1, ))
assert_size_stride(primals_30, (1024, 1024, 1, 1), (1024, 1, 1, 1))
assert_size_stride(primals_31, (1024, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 3, 3, 3), (27, 1, 9, 3), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
stream0 = get_raw_stream(0)
triton_poi_fused_0.run(primals_1, buf0, 192, 9, grid=grid(192, 9), stream=stream0)
del primals_1
buf1 = empty_strided_cuda((4, 3, 64, 64), (12288, 1, 192, 3), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_1.run(primals_3, buf1, 12, 4096, grid=grid(12, 4096), stream=stream0)
del primals_3
buf2 = empty_strided_cuda((64, 64, 3, 3), (576, 1, 192, 64), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_2.run(primals_4, buf2, 4096, 9, grid=grid(4096, 9), stream=stream0)
del primals_4
buf3 = empty_strided_cuda((128, 64, 3, 3), (576, 1, 192, 64), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_3.run(primals_6, buf3, 8192, 9, grid=grid(8192, 9), stream=stream0)
del primals_6
buf4 = empty_strided_cuda((128, 128, 3, 3), (1152, 1, 384, 128), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_4.run(primals_8, buf4, 16384, 9, grid=grid(16384, 9), stream=stream0)
del primals_8
buf5 = empty_strided_cuda((256, 128, 3, 3), (1152, 1, 384, 128), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_5.run(primals_10, buf5, 32768, 9, grid=grid(32768, 9), stream=stream0)
del primals_10
buf6 = empty_strided_cuda((256, 256, 3, 3), (2304, 1, 768, 256), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_6.run(primals_12, buf6, 65536, 9, grid=grid(65536, 9), stream=stream0)
del primals_12
buf7 = empty_strided_cuda((256, 256, 3, 3), (2304, 1, 768, 256), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_6.run(primals_14, buf7, 65536, 9, grid=grid(65536, 9), stream=stream0)
del primals_14
buf8 = empty_strided_cuda((512, 256, 3, 3), (2304, 1, 768, 256), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_7.run(primals_16, buf8, 131072, 9, grid=grid(131072, 9), stream=stream0)
del primals_16
buf9 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_8.run(primals_18, buf9, 262144, 9, grid=grid(262144, 9), stream=stream0)
del primals_18
buf10 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_8.run(primals_20, buf10, 262144, 9, grid=grid(262144, 9), stream=stream0)
del primals_20
buf11 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_8.run(primals_22, buf11, 262144, 9, grid=grid(262144, 9), stream=stream0)
del primals_22
buf12 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_8.run(primals_24, buf12, 262144, 9, grid=grid(262144, 9), stream=stream0)
del primals_24
buf13 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_8.run(primals_26, buf13, 262144, 9, grid=grid(262144, 9), stream=stream0)
del primals_26
buf14 = empty_strided_cuda((1024, 512, 3, 3), (4608, 1, 1536, 512), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
triton_poi_fused_9.run(primals_28, buf14, 524288, 9, grid=grid(524288, 9), stream=stream0)
del primals_28
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf15 = extern_kernels.convolution(buf1, buf0, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf15, (4, 64, 64, 64), (262144, 1, 4096, 64))
buf16 = buf15; del buf15 # reuse
# Topologically Sorted Source Nodes: [conv2d, out], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_10.run(buf16, primals_2, 1048576, grid=grid(1048576), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf17 = extern_kernels.convolution(buf16, buf2, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf17, (4, 64, 64, 64), (262144, 1, 4096, 64))
buf18 = buf17; del buf17 # reuse
# Topologically Sorted Source Nodes: [conv2d_1, out_1], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_10.run(buf18, primals_5, 1048576, grid=grid(1048576), stream=stream0)
del primals_5
buf19 = empty_strided_cuda((4, 64, 32, 32), (65536, 1, 2048, 64), torch.float32)
buf20 = empty_strided_cuda((4, 64, 32, 32), (65536, 1, 2048, 64), torch.int8)
# Topologically Sorted Source Nodes: [out_2], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_11.run(buf18, buf19, buf20, 262144, grid=grid(262144), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_2], Original ATen: [aten.convolution]
buf21 = extern_kernels.convolution(buf19, buf3, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf21, (4, 128, 32, 32), (131072, 1, 4096, 128))
buf22 = buf21; del buf21 # reuse
# Topologically Sorted Source Nodes: [conv2d_2, out_3], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_12.run(buf22, primals_7, 524288, grid=grid(524288), stream=stream0)
del primals_7
# Topologically Sorted Source Nodes: [conv2d_3], Original ATen: [aten.convolution]
buf23 = extern_kernels.convolution(buf22, buf4, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf23, (4, 128, 32, 32), (131072, 1, 4096, 128))
buf24 = buf23; del buf23 # reuse
# Topologically Sorted Source Nodes: [conv2d_3, out_4], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_12.run(buf24, primals_9, 524288, grid=grid(524288), stream=stream0)
del primals_9
buf25 = empty_strided_cuda((4, 128, 16, 16), (32768, 1, 2048, 128), torch.float32)
buf26 = empty_strided_cuda((4, 128, 16, 16), (32768, 1, 2048, 128), torch.int8)
# Topologically Sorted Source Nodes: [out_5], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_13.run(buf24, buf25, buf26, 131072, grid=grid(131072), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_4], Original ATen: [aten.convolution]
buf27 = extern_kernels.convolution(buf25, buf5, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf27, (4, 256, 16, 16), (65536, 1, 4096, 256))
buf28 = buf27; del buf27 # reuse
# Topologically Sorted Source Nodes: [conv2d_4, out_6], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_14.run(buf28, primals_11, 262144, grid=grid(262144), stream=stream0)
del primals_11
# Topologically Sorted Source Nodes: [conv2d_5], Original ATen: [aten.convolution]
buf29 = extern_kernels.convolution(buf28, buf6, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf29, (4, 256, 16, 16), (65536, 1, 4096, 256))
buf30 = buf29; del buf29 # reuse
# Topologically Sorted Source Nodes: [conv2d_5, out_7], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_14.run(buf30, primals_13, 262144, grid=grid(262144), stream=stream0)
del primals_13
# Topologically Sorted Source Nodes: [conv2d_6], Original ATen: [aten.convolution]
buf31 = extern_kernels.convolution(buf30, buf7, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf31, (4, 256, 16, 16), (65536, 1, 4096, 256))
buf32 = buf31; del buf31 # reuse
# Topologically Sorted Source Nodes: [conv2d_6, out_8], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_14.run(buf32, primals_15, 262144, grid=grid(262144), stream=stream0)
del primals_15
buf33 = empty_strided_cuda((4, 256, 8, 8), (16384, 1, 2048, 256), torch.float32)
buf34 = empty_strided_cuda((4, 256, 8, 8), (16384, 1, 2048, 256), torch.int8)
# Topologically Sorted Source Nodes: [out_9], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_15.run(buf32, buf33, buf34, 65536, grid=grid(65536), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_7], Original ATen: [aten.convolution]
buf35 = extern_kernels.convolution(buf33, buf8, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf35, (4, 512, 8, 8), (32768, 1, 4096, 512))
buf36 = buf35; del buf35 # reuse
# Topologically Sorted Source Nodes: [conv2d_7, out_10], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_16.run(buf36, primals_17, 131072, grid=grid(131072), stream=stream0)
del primals_17
# Topologically Sorted Source Nodes: [conv2d_8], Original ATen: [aten.convolution]
buf37 = extern_kernels.convolution(buf36, buf9, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf37, (4, 512, 8, 8), (32768, 1, 4096, 512))
buf38 = buf37; del buf37 # reuse
# Topologically Sorted Source Nodes: [conv2d_8, out_11], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_16.run(buf38, primals_19, 131072, grid=grid(131072), stream=stream0)
del primals_19
# Topologically Sorted Source Nodes: [conv2d_9], Original ATen: [aten.convolution]
buf39 = extern_kernels.convolution(buf38, buf10, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf39, (4, 512, 8, 8), (32768, 1, 4096, 512))
buf40 = empty_strided_cuda((4, 512, 8, 8), (32768, 64, 8, 1), torch.float32)
# Topologically Sorted Source Nodes: [conv2d_9, out_12], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_17.run(buf39, primals_21, buf40, 2048, 64, grid=grid(2048, 64), stream=stream0)
del buf39
del primals_21
buf41 = empty_strided_cuda((4, 512, 4, 4), (8192, 1, 2048, 512), torch.float32)
buf42 = empty_strided_cuda((4, 512, 4, 4), (8192, 1, 2048, 512), torch.int8)
# Topologically Sorted Source Nodes: [out_13], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_18.run(buf40, buf41, buf42, 2048, 16, grid=grid(2048, 16), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_10], Original ATen: [aten.convolution]
buf43 = extern_kernels.convolution(buf41, buf11, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf43, (4, 512, 4, 4), (8192, 1, 2048, 512))
buf44 = buf43; del buf43 # reuse
# Topologically Sorted Source Nodes: [conv2d_10, out_14], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_19.run(buf44, primals_23, 32768, grid=grid(32768), stream=stream0)
del primals_23
# Topologically Sorted Source Nodes: [conv2d_11], Original ATen: [aten.convolution]
buf45 = extern_kernels.convolution(buf44, buf12, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf45, (4, 512, 4, 4), (8192, 1, 2048, 512))
buf46 = buf45; del buf45 # reuse
# Topologically Sorted Source Nodes: [conv2d_11, out_15], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_19.run(buf46, primals_25, 32768, grid=grid(32768), stream=stream0)
del primals_25
# Topologically Sorted Source Nodes: [conv2d_12], Original ATen: [aten.convolution]
buf47 = extern_kernels.convolution(buf46, buf13, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf47, (4, 512, 4, 4), (8192, 1, 2048, 512))
buf48 = buf47; del buf47 # reuse
# Topologically Sorted Source Nodes: [conv2d_12, out_16], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_19.run(buf48, primals_27, 32768, grid=grid(32768), stream=stream0)
del primals_27
buf49 = empty_strided_cuda((4, 512, 4, 4), (8192, 1, 2048, 512), torch.float32)
buf50 = empty_strided_cuda((4, 512, 4, 4), (8192, 1, 2048, 512), torch.int8)
# Topologically Sorted Source Nodes: [out_17], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_20.run(buf48, buf49, buf50, 32768, grid=grid(32768), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_13], Original ATen: [aten.convolution]
buf51 = extern_kernels.convolution(buf49, buf14, stride=(1, 1), padding=(6, 6), dilation=(6, 6), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf51, (4, 1024, 4, 4), (16384, 1, 4096, 1024))
buf52 = buf51; del buf51 # reuse
# Topologically Sorted Source Nodes: [conv2d_13, out_18], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_21.run(buf52, primals_29, 65536, grid=grid(65536), stream=stream0)
del primals_29
# Topologically Sorted Source Nodes: [conv2d_14], Original ATen: [aten.convolution]
buf53 = extern_kernels.convolution(buf52, primals_30, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf53, (4, 1024, 4, 4), (16384, 1, 4096, 1024))
buf54 = empty_strided_cuda((4, 1024, 4, 4), (16384, 16, 4, 1), torch.float32)
buf55 = empty_strided_cuda((4, 1024, 4, 4), (16384, 1, 4096, 1024), torch.bool)
# Topologically Sorted Source Nodes: [conv2d_14, conv7_feats], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_22.run(buf53, primals_31, buf54, buf55, 4096, 16, grid=grid(4096, 16), stream=stream0)
del buf53
del primals_31
return (buf40, buf54, buf0, buf1, buf2, buf3, buf4, buf5, buf6, buf7, buf8, buf9, buf10, buf11, buf12, buf13, buf14, primals_30, buf16, buf18, buf19, buf20, buf22, buf24, buf25, buf26, buf28, buf30, buf32, buf33, buf34, buf36, buf38, buf40, buf41, buf42, buf44, buf46, buf48, buf49, buf50, buf52, buf55, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((64, 3, 3, 3), (27, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 3, 64, 64), (12288, 4096, 64, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((64, 64, 3, 3), (576, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((128, 64, 3, 3), (576, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((128, 128, 3, 3), (1152, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((256, 128, 3, 3), (1152, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((256, 256, 3, 3), (2304, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_14 = rand_strided((256, 256, 3, 3), (2304, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_15 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_16 = rand_strided((512, 256, 3, 3), (2304, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_17 = rand_strided((512, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_18 = rand_strided((512, 512, 3, 3), (4608, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_19 = rand_strided((512, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_20 = rand_strided((512, 512, 3, 3), (4608, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_21 = rand_strided((512, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_22 = rand_strided((512, 512, 3, 3), (4608, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_23 = rand_strided((512, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_24 = rand_strided((512, 512, 3, 3), (4608, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_25 = rand_strided((512, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_26 = rand_strided((512, 512, 3, 3), (4608, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_27 = rand_strided((512, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_28 = rand_strided((1024, 512, 3, 3), (4608, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_29 = rand_strided((1024, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_30 = rand_strided((1024, 1024, 1, 1), (1024, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_31 = rand_strided((1024, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14, primals_15, primals_16, primals_17, primals_18, primals_19, primals_20, primals_21, primals_22, primals_23, primals_24, primals_25, primals_26, primals_27, primals_28, primals_29, primals_30, primals_31])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torchvision
from torch import nn
import torch.nn.functional as F
from itertools import product as product
import torch.optim
import torch.utils.data
def decimate(tensor, m):
"""
Decimate a tensor by a factor 'm', i.e. downsample by keeping every 'm'th value.
This is used when we convert FC layers to equivalent Convolutional layers, BUT of a smaller size.
:param tensor: tensor to be decimated
:param m: list of decimation factors for each dimension of the tensor; None if not to be decimated along a dimension
:return: decimated tensor
"""
assert tensor.dim() == len(m)
for d in range(tensor.dim()):
if m[d] is not None:
tensor = tensor.index_select(dim=d, index=torch.arange(start=0,
end=tensor.size(d), step=m[d]).long())
return tensor
class VGGBase(nn.Module):
"""
VGG base convolutions to produce lower-level feature maps.
"""
def __init__(self):
super(VGGBase, self).__init__()
self.conv1_1 = nn.Conv2d(3, 64, kernel_size=3, padding=1)
self.conv1_2 = nn.Conv2d(64, 64, kernel_size=3, padding=1)
self.pool1 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv2_1 = nn.Conv2d(64, 128, kernel_size=3, padding=1)
self.conv2_2 = nn.Conv2d(128, 128, kernel_size=3, padding=1)
self.pool2 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv3_1 = nn.Conv2d(128, 256, kernel_size=3, padding=1)
self.conv3_2 = nn.Conv2d(256, 256, kernel_size=3, padding=1)
self.conv3_3 = nn.Conv2d(256, 256, kernel_size=3, padding=1)
self.pool3 = nn.MaxPool2d(kernel_size=2, stride=2, ceil_mode=True)
self.conv4_1 = nn.Conv2d(256, 512, kernel_size=3, padding=1)
self.conv4_2 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv4_3 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.pool4 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv5_1 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv5_2 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv5_3 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.pool5 = nn.MaxPool2d(kernel_size=3, stride=1, padding=1)
self.conv6 = nn.Conv2d(512, 1024, kernel_size=3, padding=6, dilation=6)
self.conv7 = nn.Conv2d(1024, 1024, kernel_size=1)
self.load_pretrained_layers()
def forward(self, image):
"""
Forward propagation.
:param image: images, a tensor of dimensions (N, 3, 300, 300)
:return: lower-level feature maps conv4_3 and conv7
"""
out = F.relu(self.conv1_1(image))
out = F.relu(self.conv1_2(out))
out = self.pool1(out)
out = F.relu(self.conv2_1(out))
out = F.relu(self.conv2_2(out))
out = self.pool2(out)
out = F.relu(self.conv3_1(out))
out = F.relu(self.conv3_2(out))
out = F.relu(self.conv3_3(out))
out = self.pool3(out)
out = F.relu(self.conv4_1(out))
out = F.relu(self.conv4_2(out))
out = F.relu(self.conv4_3(out))
conv4_3_feats = out
out = self.pool4(out)
out = F.relu(self.conv5_1(out))
out = F.relu(self.conv5_2(out))
out = F.relu(self.conv5_3(out))
out = self.pool5(out)
out = F.relu(self.conv6(out))
conv7_feats = F.relu(self.conv7(out))
return conv4_3_feats, conv7_feats
def load_pretrained_layers(self):
"""
As in the paper, we use a VGG-16 pretrained on the ImageNet task as the base network.
There's one available in PyTorch, see https://pytorch.org/docs/stable/torchvision/models.html#torchvision.models.vgg16
We copy these parameters into our network. It's straightforward for conv1 to conv5.
However, the original VGG-16 does not contain the conv6 and con7 layers.
Therefore, we convert fc6 and fc7 into convolutional layers, and subsample by decimation. See 'decimate' in utils.py.
"""
state_dict = self.state_dict()
param_names = list(state_dict.keys())
pretrained_state_dict = torchvision.models.vgg16(pretrained=True
).state_dict()
pretrained_param_names = list(pretrained_state_dict.keys())
for i, param in enumerate(param_names[:-4]):
state_dict[param] = pretrained_state_dict[pretrained_param_names[i]
]
conv_fc6_weight = pretrained_state_dict['classifier.0.weight'].view(
4096, 512, 7, 7)
conv_fc6_bias = pretrained_state_dict['classifier.0.bias']
state_dict['conv6.weight'] = decimate(conv_fc6_weight, m=[4, None,
3, 3])
state_dict['conv6.bias'] = decimate(conv_fc6_bias, m=[4])
conv_fc7_weight = pretrained_state_dict['classifier.3.weight'].view(
4096, 4096, 1, 1)
conv_fc7_bias = pretrained_state_dict['classifier.3.bias']
state_dict['conv7.weight'] = decimate(conv_fc7_weight, m=[4, 4,
None, None])
state_dict['conv7.bias'] = decimate(conv_fc7_bias, m=[4])
self.load_state_dict(state_dict)
None
def get_inputs():
return [torch.rand([4, 3, 64, 64])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torchvision
from torch import nn
from itertools import product as product
import torch.optim
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 192
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 3
y1 = yindex // 3
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask & ymask, eviction_policy=
'evict_last')
tl.store(out_ptr0 + (y0 + 3 * x2 + 27 * y1), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 12
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
x2 = xindex
y3 = yindex
y0 = yindex % 3
y1 = yindex // 3
tmp0 = tl.load(in_ptr0 + (x2 + 4096 * y3), ymask, eviction_policy=
'evict_last')
tl.store(out_ptr0 + (y0 + 3 * x2 + 12288 * y1), tmp0, ymask)
@triton.jit
def triton_poi_fused_2(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 64
y1 = yindex // 64
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + (y0 + 64 * x2 + 576 * y1), tmp0, xmask)
@triton.jit
def triton_poi_fused_3(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 64
y1 = yindex // 64
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + (y0 + 64 * x2 + 576 * y1), tmp0, xmask)
@triton.jit
def triton_poi_fused_4(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 128
y1 = yindex // 128
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + (y0 + 128 * x2 + 1152 * y1), tmp0, xmask)
@triton.jit
def triton_poi_fused_5(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
xnumel = 9
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 128
y1 = yindex // 128
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + (y0 + 128 * x2 + 1152 * y1), tmp0, xmask)
@triton.jit
def triton_poi_fused_6(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
xnumel = 9
yoffset = (tl.program_id(1) + tl.program_id(2) * tl.num_programs(1)
) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 256
y1 = yindex // 256
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + (y0 + 256 * x2 + 2304 * y1), tmp0, xmask)
@triton.jit
def triton_poi_fused_7(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
xnumel = 9
yoffset = (tl.program_id(1) + tl.program_id(2) * tl.num_programs(1)
) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 256
y1 = yindex // 256
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + (y0 + 256 * x2 + 2304 * y1), tmp0, xmask)
@triton.jit
def triton_poi_fused_8(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
xnumel = 9
yoffset = (tl.program_id(1) + tl.program_id(2) * tl.num_programs(1)
) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 512
y1 = yindex // 512
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + (y0 + 512 * x2 + 4608 * y1), tmp0, xmask)
@triton.jit
def triton_poi_fused_9(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
xnumel = 9
yoffset = (tl.program_id(1) + tl.program_id(2) * tl.num_programs(1)
) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y3 = yindex
y0 = yindex % 512
y1 = yindex // 512
tmp0 = tl.load(in_ptr0 + (x2 + 9 * y3), xmask, eviction_policy='evict_last'
)
tl.store(out_ptr0 + (y0 + 512 * x2 + 4608 * y1), tmp0, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_10(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 64
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_11(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 64
x1 = xindex // 64 % 32
x2 = xindex // 2048
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 128 * x1 + 8192 * x2), None)
tmp1 = tl.load(in_ptr0 + (64 + x0 + 128 * x1 + 8192 * x2), None)
tmp3 = tl.load(in_ptr0 + (4096 + x0 + 128 * x1 + 8192 * x2), None)
tmp5 = tl.load(in_ptr0 + (4160 + x0 + 128 * x1 + 8192 * x2), None)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + x3, tmp6, None)
tl.store(out_ptr1 + x3, tmp16, None)
@triton.jit
def triton_poi_fused_convolution_relu_12(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 128
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_13(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 128
x1 = xindex // 128 % 16
x2 = xindex // 2048
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 256 * x1 + 8192 * x2), None)
tmp1 = tl.load(in_ptr0 + (128 + x0 + 256 * x1 + 8192 * x2), None)
tmp3 = tl.load(in_ptr0 + (4096 + x0 + 256 * x1 + 8192 * x2), None)
tmp5 = tl.load(in_ptr0 + (4224 + x0 + 256 * x1 + 8192 * x2), None)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + x3, tmp6, None)
tl.store(out_ptr1 + x3, tmp16, None)
@triton.jit
def triton_poi_fused_convolution_relu_14(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 256
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_15(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 256
x1 = xindex // 256 % 8
x2 = xindex // 2048
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 512 * x1 + 8192 * x2), None)
tmp1 = tl.load(in_ptr0 + (256 + x0 + 512 * x1 + 8192 * x2), None)
tmp3 = tl.load(in_ptr0 + (4096 + x0 + 512 * x1 + 8192 * x2), None)
tmp5 = tl.load(in_ptr0 + (4352 + x0 + 512 * x1 + 8192 * x2), None)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + x3, tmp6, None)
tl.store(out_ptr1 + x3, tmp16, None)
@triton.jit
def triton_poi_fused_convolution_relu_16(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 512
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused_convolution_relu_17(in_ptr0, in_ptr1, out_ptr0, ynumel,
xnumel, YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
xnumel = 64
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 512
y1 = yindex // 512
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 512 * x2 + 32768 * y1), xmask,
eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + y0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1, 1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(out_ptr0 + (x2 + 64 * y3), tmp4, xmask)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_18(in_ptr0, out_ptr0, out_ptr1,
ynumel, xnumel, YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex % 4
x3 = xindex // 4
y4 = yindex
x5 = xindex
y0 = yindex % 512
y1 = yindex // 512
tmp0 = tl.load(in_ptr0 + (2 * x2 + 16 * x3 + 64 * y4), xmask,
eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 2 * x2 + 16 * x3 + 64 * y4), xmask,
eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (8 + 2 * x2 + 16 * x3 + 64 * y4), xmask,
eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (9 + 2 * x2 + 16 * x3 + 64 * y4), xmask,
eviction_policy='evict_last')
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1, 1], 1, tl.int8)
tmp9 = tl.full([1, 1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1, 1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1, 1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + (y0 + 512 * x5 + 8192 * y1), tmp6, xmask)
tl.store(out_ptr1 + (y0 + 512 * x5 + 8192 * y1), tmp16, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_19(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 512
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_20(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex // 2048 % 4
x1 = xindex // 512 % 4
x6 = xindex
tmp0 = -1 + x2
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = -1 + x1
tmp7 = tmp6 >= tmp1
tmp8 = tmp6 < tmp3
tmp9 = tmp7 & tmp8
tmp10 = tmp5 & tmp9
tmp11 = tl.load(in_ptr0 + (-2560 + x6), tmp10, other=float('-inf'))
tmp12 = x1
tmp13 = tmp12 >= tmp1
tmp14 = tmp12 < tmp3
tmp15 = tmp13 & tmp14
tmp16 = tmp5 & tmp15
tmp17 = tl.load(in_ptr0 + (-2048 + x6), tmp16, other=float('-inf'))
tmp18 = triton_helpers.maximum(tmp17, tmp11)
tmp19 = 1 + x1
tmp20 = tmp19 >= tmp1
tmp21 = tmp19 < tmp3
tmp22 = tmp20 & tmp21
tmp23 = tmp5 & tmp22
tmp24 = tl.load(in_ptr0 + (-1536 + x6), tmp23, other=float('-inf'))
tmp25 = triton_helpers.maximum(tmp24, tmp18)
tmp26 = x2
tmp27 = tmp26 >= tmp1
tmp28 = tmp26 < tmp3
tmp29 = tmp27 & tmp28
tmp30 = tmp29 & tmp9
tmp31 = tl.load(in_ptr0 + (-512 + x6), tmp30, other=float('-inf'))
tmp32 = triton_helpers.maximum(tmp31, tmp25)
tmp33 = tmp29 & tmp15
tmp34 = tl.load(in_ptr0 + x6, tmp33, other=float('-inf'))
tmp35 = triton_helpers.maximum(tmp34, tmp32)
tmp36 = tmp29 & tmp22
tmp37 = tl.load(in_ptr0 + (512 + x6), tmp36, other=float('-inf'))
tmp38 = triton_helpers.maximum(tmp37, tmp35)
tmp39 = 1 + x2
tmp40 = tmp39 >= tmp1
tmp41 = tmp39 < tmp3
tmp42 = tmp40 & tmp41
tmp43 = tmp42 & tmp9
tmp44 = tl.load(in_ptr0 + (1536 + x6), tmp43, other=float('-inf'))
tmp45 = triton_helpers.maximum(tmp44, tmp38)
tmp46 = tmp42 & tmp15
tmp47 = tl.load(in_ptr0 + (2048 + x6), tmp46, other=float('-inf'))
tmp48 = triton_helpers.maximum(tmp47, tmp45)
tmp49 = tmp42 & tmp22
tmp50 = tl.load(in_ptr0 + (2560 + x6), tmp49, other=float('-inf'))
tmp51 = triton_helpers.maximum(tmp50, tmp48)
tmp52 = tmp17 > tmp11
tmp53 = tl.full([1], 1, tl.int8)
tmp54 = tl.full([1], 0, tl.int8)
tmp55 = tl.where(tmp52, tmp53, tmp54)
tmp56 = tmp24 > tmp18
tmp57 = tl.full([1], 2, tl.int8)
tmp58 = tl.where(tmp56, tmp57, tmp55)
tmp59 = tmp31 > tmp25
tmp60 = tl.full([1], 3, tl.int8)
tmp61 = tl.where(tmp59, tmp60, tmp58)
tmp62 = tmp34 > tmp32
tmp63 = tl.full([1], 4, tl.int8)
tmp64 = tl.where(tmp62, tmp63, tmp61)
tmp65 = tmp37 > tmp35
tmp66 = tl.full([1], 5, tl.int8)
tmp67 = tl.where(tmp65, tmp66, tmp64)
tmp68 = tmp44 > tmp38
tmp69 = tl.full([1], 6, tl.int8)
tmp70 = tl.where(tmp68, tmp69, tmp67)
tmp71 = tmp47 > tmp45
tmp72 = tl.full([1], 7, tl.int8)
tmp73 = tl.where(tmp71, tmp72, tmp70)
tmp74 = tmp50 > tmp48
tmp75 = tl.full([1], 8, tl.int8)
tmp76 = tl.where(tmp74, tmp75, tmp73)
tl.store(out_ptr0 + x6, tmp51, None)
tl.store(out_ptr1 + x6, tmp76, None)
@triton.jit
def triton_poi_fused_convolution_relu_21(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 1024
tmp0 = tl.load(in_out_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, None)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_22(in_ptr0,
in_ptr1, out_ptr0, out_ptr1, ynumel, xnumel, YBLOCK: tl.constexpr,
XBLOCK: tl.constexpr):
xnumel = 16
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 1024
y1 = yindex // 1024
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 1024 * x2 + 16384 * y1), xmask,
eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + y0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1, 1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x2 + 16 * y3), tmp4, xmask)
tl.store(out_ptr1 + (y0 + 1024 * x2 + 16384 * y1), tmp6, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13, primals_14, primals_15, primals_16, primals_17,
primals_18, primals_19, primals_20, primals_21, primals_22,
primals_23, primals_24, primals_25, primals_26, primals_27,
primals_28, primals_29, primals_30, primals_31) = args
args.clear()
assert_size_stride(primals_1, (64, 3, 3, 3), (27, 9, 3, 1))
assert_size_stride(primals_2, (64,), (1,))
assert_size_stride(primals_3, (4, 3, 64, 64), (12288, 4096, 64, 1))
assert_size_stride(primals_4, (64, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_5, (64,), (1,))
assert_size_stride(primals_6, (128, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_7, (128,), (1,))
assert_size_stride(primals_8, (128, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_9, (128,), (1,))
assert_size_stride(primals_10, (256, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_11, (256,), (1,))
assert_size_stride(primals_12, (256, 256, 3, 3), (2304, 9, 3, 1))
assert_size_stride(primals_13, (256,), (1,))
assert_size_stride(primals_14, (256, 256, 3, 3), (2304, 9, 3, 1))
assert_size_stride(primals_15, (256,), (1,))
assert_size_stride(primals_16, (512, 256, 3, 3), (2304, 9, 3, 1))
assert_size_stride(primals_17, (512,), (1,))
assert_size_stride(primals_18, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_19, (512,), (1,))
assert_size_stride(primals_20, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_21, (512,), (1,))
assert_size_stride(primals_22, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_23, (512,), (1,))
assert_size_stride(primals_24, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_25, (512,), (1,))
assert_size_stride(primals_26, (512, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_27, (512,), (1,))
assert_size_stride(primals_28, (1024, 512, 3, 3), (4608, 9, 3, 1))
assert_size_stride(primals_29, (1024,), (1,))
assert_size_stride(primals_30, (1024, 1024, 1, 1), (1024, 1, 1, 1))
assert_size_stride(primals_31, (1024,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 3, 3, 3), (27, 1, 9, 3), torch.float32)
get_raw_stream(0)
triton_poi_fused_0[grid(192, 9)](primals_1, buf0, 192, 9, XBLOCK=16,
YBLOCK=64, num_warps=4, num_stages=1)
del primals_1
buf1 = empty_strided_cuda((4, 3, 64, 64), (12288, 1, 192, 3), torch
.float32)
triton_poi_fused_1[grid(12, 4096)](primals_3, buf1, 12, 4096,
XBLOCK=64, YBLOCK=16, num_warps=4, num_stages=1)
del primals_3
buf2 = empty_strided_cuda((64, 64, 3, 3), (576, 1, 192, 64), torch.
float32)
triton_poi_fused_2[grid(4096, 9)](primals_4, buf2, 4096, 9, XBLOCK=
16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_4
buf3 = empty_strided_cuda((128, 64, 3, 3), (576, 1, 192, 64), torch
.float32)
triton_poi_fused_3[grid(8192, 9)](primals_6, buf3, 8192, 9, XBLOCK=
16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_6
buf4 = empty_strided_cuda((128, 128, 3, 3), (1152, 1, 384, 128),
torch.float32)
triton_poi_fused_4[grid(16384, 9)](primals_8, buf4, 16384, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_8
buf5 = empty_strided_cuda((256, 128, 3, 3), (1152, 1, 384, 128),
torch.float32)
triton_poi_fused_5[grid(32768, 9)](primals_10, buf5, 32768, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_10
buf6 = empty_strided_cuda((256, 256, 3, 3), (2304, 1, 768, 256),
torch.float32)
triton_poi_fused_6[grid(65536, 9)](primals_12, buf6, 65536, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_12
buf7 = empty_strided_cuda((256, 256, 3, 3), (2304, 1, 768, 256),
torch.float32)
triton_poi_fused_6[grid(65536, 9)](primals_14, buf7, 65536, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_14
buf8 = empty_strided_cuda((512, 256, 3, 3), (2304, 1, 768, 256),
torch.float32)
triton_poi_fused_7[grid(131072, 9)](primals_16, buf8, 131072, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_16
buf9 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512),
torch.float32)
triton_poi_fused_8[grid(262144, 9)](primals_18, buf9, 262144, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_18
buf10 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512),
torch.float32)
triton_poi_fused_8[grid(262144, 9)](primals_20, buf10, 262144, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_20
buf11 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512),
torch.float32)
triton_poi_fused_8[grid(262144, 9)](primals_22, buf11, 262144, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_22
buf12 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512),
torch.float32)
triton_poi_fused_8[grid(262144, 9)](primals_24, buf12, 262144, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_24
buf13 = empty_strided_cuda((512, 512, 3, 3), (4608, 1, 1536, 512),
torch.float32)
triton_poi_fused_8[grid(262144, 9)](primals_26, buf13, 262144, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_26
buf14 = empty_strided_cuda((1024, 512, 3, 3), (4608, 1, 1536, 512),
torch.float32)
triton_poi_fused_9[grid(524288, 9)](primals_28, buf14, 524288, 9,
XBLOCK=16, YBLOCK=64, num_warps=4, num_stages=1)
del primals_28
buf15 = extern_kernels.convolution(buf1, buf0, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf15, (4, 64, 64, 64), (262144, 1, 4096, 64))
buf16 = buf15
del buf15
triton_poi_fused_convolution_relu_10[grid(1048576)](buf16,
primals_2, 1048576, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_2
buf17 = extern_kernels.convolution(buf16, buf2, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf17, (4, 64, 64, 64), (262144, 1, 4096, 64))
buf18 = buf17
del buf17
triton_poi_fused_convolution_relu_10[grid(1048576)](buf18,
primals_5, 1048576, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_5
buf19 = empty_strided_cuda((4, 64, 32, 32), (65536, 1, 2048, 64),
torch.float32)
buf20 = empty_strided_cuda((4, 64, 32, 32), (65536, 1, 2048, 64),
torch.int8)
triton_poi_fused_max_pool2d_with_indices_11[grid(262144)](buf18,
buf19, buf20, 262144, XBLOCK=512, num_warps=8, num_stages=1)
buf21 = extern_kernels.convolution(buf19, buf3, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf21, (4, 128, 32, 32), (131072, 1, 4096, 128))
buf22 = buf21
del buf21
triton_poi_fused_convolution_relu_12[grid(524288)](buf22, primals_7,
524288, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_7
buf23 = extern_kernels.convolution(buf22, buf4, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf23, (4, 128, 32, 32), (131072, 1, 4096, 128))
buf24 = buf23
del buf23
triton_poi_fused_convolution_relu_12[grid(524288)](buf24, primals_9,
524288, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_9
buf25 = empty_strided_cuda((4, 128, 16, 16), (32768, 1, 2048, 128),
torch.float32)
buf26 = empty_strided_cuda((4, 128, 16, 16), (32768, 1, 2048, 128),
torch.int8)
triton_poi_fused_max_pool2d_with_indices_13[grid(131072)](buf24,
buf25, buf26, 131072, XBLOCK=512, num_warps=8, num_stages=1)
buf27 = extern_kernels.convolution(buf25, buf5, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf27, (4, 256, 16, 16), (65536, 1, 4096, 256))
buf28 = buf27
del buf27
triton_poi_fused_convolution_relu_14[grid(262144)](buf28,
primals_11, 262144, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_11
buf29 = extern_kernels.convolution(buf28, buf6, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf29, (4, 256, 16, 16), (65536, 1, 4096, 256))
buf30 = buf29
del buf29
triton_poi_fused_convolution_relu_14[grid(262144)](buf30,
primals_13, 262144, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_13
buf31 = extern_kernels.convolution(buf30, buf7, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf31, (4, 256, 16, 16), (65536, 1, 4096, 256))
buf32 = buf31
del buf31
triton_poi_fused_convolution_relu_14[grid(262144)](buf32,
primals_15, 262144, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_15
buf33 = empty_strided_cuda((4, 256, 8, 8), (16384, 1, 2048, 256),
torch.float32)
buf34 = empty_strided_cuda((4, 256, 8, 8), (16384, 1, 2048, 256),
torch.int8)
triton_poi_fused_max_pool2d_with_indices_15[grid(65536)](buf32,
buf33, buf34, 65536, XBLOCK=512, num_warps=4, num_stages=1)
buf35 = extern_kernels.convolution(buf33, buf8, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf35, (4, 512, 8, 8), (32768, 1, 4096, 512))
buf36 = buf35
del buf35
triton_poi_fused_convolution_relu_16[grid(131072)](buf36,
primals_17, 131072, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_17
buf37 = extern_kernels.convolution(buf36, buf9, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf37, (4, 512, 8, 8), (32768, 1, 4096, 512))
buf38 = buf37
del buf37
triton_poi_fused_convolution_relu_16[grid(131072)](buf38,
primals_19, 131072, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_19
buf39 = extern_kernels.convolution(buf38, buf10, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf39, (4, 512, 8, 8), (32768, 1, 4096, 512))
buf40 = empty_strided_cuda((4, 512, 8, 8), (32768, 64, 8, 1), torch
.float32)
triton_poi_fused_convolution_relu_17[grid(2048, 64)](buf39,
primals_21, buf40, 2048, 64, XBLOCK=32, YBLOCK=32, num_warps=4,
num_stages=1)
del buf39
del primals_21
buf41 = empty_strided_cuda((4, 512, 4, 4), (8192, 1, 2048, 512),
torch.float32)
buf42 = empty_strided_cuda((4, 512, 4, 4), (8192, 1, 2048, 512),
torch.int8)
triton_poi_fused_max_pool2d_with_indices_18[grid(2048, 16)](buf40,
buf41, buf42, 2048, 16, XBLOCK=16, YBLOCK=16, num_warps=4,
num_stages=1)
buf43 = extern_kernels.convolution(buf41, buf11, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf43, (4, 512, 4, 4), (8192, 1, 2048, 512))
buf44 = buf43
del buf43
triton_poi_fused_convolution_relu_19[grid(32768)](buf44, primals_23,
32768, XBLOCK=256, num_warps=4, num_stages=1)
del primals_23
buf45 = extern_kernels.convolution(buf44, buf12, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf45, (4, 512, 4, 4), (8192, 1, 2048, 512))
buf46 = buf45
del buf45
triton_poi_fused_convolution_relu_19[grid(32768)](buf46, primals_25,
32768, XBLOCK=256, num_warps=4, num_stages=1)
del primals_25
buf47 = extern_kernels.convolution(buf46, buf13, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf47, (4, 512, 4, 4), (8192, 1, 2048, 512))
buf48 = buf47
del buf47
triton_poi_fused_convolution_relu_19[grid(32768)](buf48, primals_27,
32768, XBLOCK=256, num_warps=4, num_stages=1)
del primals_27
buf49 = empty_strided_cuda((4, 512, 4, 4), (8192, 1, 2048, 512),
torch.float32)
buf50 = empty_strided_cuda((4, 512, 4, 4), (8192, 1, 2048, 512),
torch.int8)
triton_poi_fused_max_pool2d_with_indices_20[grid(32768)](buf48,
buf49, buf50, 32768, XBLOCK=256, num_warps=4, num_stages=1)
buf51 = extern_kernels.convolution(buf49, buf14, stride=(1, 1),
padding=(6, 6), dilation=(6, 6), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf51, (4, 1024, 4, 4), (16384, 1, 4096, 1024))
buf52 = buf51
del buf51
triton_poi_fused_convolution_relu_21[grid(65536)](buf52, primals_29,
65536, XBLOCK=512, num_warps=4, num_stages=1)
del primals_29
buf53 = extern_kernels.convolution(buf52, primals_30, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf53, (4, 1024, 4, 4), (16384, 1, 4096, 1024))
buf54 = empty_strided_cuda((4, 1024, 4, 4), (16384, 16, 4, 1),
torch.float32)
buf55 = empty_strided_cuda((4, 1024, 4, 4), (16384, 1, 4096, 1024),
torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_22[grid(4096, 16)
](buf53, primals_31, buf54, buf55, 4096, 16, XBLOCK=16, YBLOCK=
64, num_warps=4, num_stages=1)
del buf53
del primals_31
return (buf40, buf54, buf0, buf1, buf2, buf3, buf4, buf5, buf6, buf7,
buf8, buf9, buf10, buf11, buf12, buf13, buf14, primals_30, buf16,
buf18, buf19, buf20, buf22, buf24, buf25, buf26, buf28, buf30,
buf32, buf33, buf34, buf36, buf38, buf40, buf41, buf42, buf44,
buf46, buf48, buf49, buf50, buf52, buf55)
def decimate(tensor, m):
"""
Decimate a tensor by a factor 'm', i.e. downsample by keeping every 'm'th value.
This is used when we convert FC layers to equivalent Convolutional layers, BUT of a smaller size.
:param tensor: tensor to be decimated
:param m: list of decimation factors for each dimension of the tensor; None if not to be decimated along a dimension
:return: decimated tensor
"""
assert tensor.dim() == len(m)
for d in range(tensor.dim()):
if m[d] is not None:
tensor = tensor.index_select(dim=d, index=torch.arange(start=0,
end=tensor.size(d), step=m[d]).long())
return tensor
class VGGBaseNew(nn.Module):
"""
VGG base convolutions to produce lower-level feature maps.
"""
def __init__(self):
super(VGGBaseNew, self).__init__()
self.conv1_1 = nn.Conv2d(3, 64, kernel_size=3, padding=1)
self.conv1_2 = nn.Conv2d(64, 64, kernel_size=3, padding=1)
self.pool1 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv2_1 = nn.Conv2d(64, 128, kernel_size=3, padding=1)
self.conv2_2 = nn.Conv2d(128, 128, kernel_size=3, padding=1)
self.pool2 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv3_1 = nn.Conv2d(128, 256, kernel_size=3, padding=1)
self.conv3_2 = nn.Conv2d(256, 256, kernel_size=3, padding=1)
self.conv3_3 = nn.Conv2d(256, 256, kernel_size=3, padding=1)
self.pool3 = nn.MaxPool2d(kernel_size=2, stride=2, ceil_mode=True)
self.conv4_1 = nn.Conv2d(256, 512, kernel_size=3, padding=1)
self.conv4_2 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv4_3 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.pool4 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv5_1 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv5_2 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv5_3 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.pool5 = nn.MaxPool2d(kernel_size=3, stride=1, padding=1)
self.conv6 = nn.Conv2d(512, 1024, kernel_size=3, padding=6, dilation=6)
self.conv7 = nn.Conv2d(1024, 1024, kernel_size=1)
self.load_pretrained_layers()
def load_pretrained_layers(self):
"""
As in the paper, we use a VGG-16 pretrained on the ImageNet task as the base network.
There's one available in PyTorch, see https://pytorch.org/docs/stable/torchvision/models.html#torchvision.models.vgg16
We copy these parameters into our network. It's straightforward for conv1 to conv5.
However, the original VGG-16 does not contain the conv6 and con7 layers.
Therefore, we convert fc6 and fc7 into convolutional layers, and subsample by decimation. See 'decimate' in utils.py.
"""
state_dict = self.state_dict()
param_names = list(state_dict.keys())
pretrained_state_dict = torchvision.models.vgg16(pretrained=True
).state_dict()
pretrained_param_names = list(pretrained_state_dict.keys())
for i, param in enumerate(param_names[:-4]):
state_dict[param] = pretrained_state_dict[pretrained_param_names[i]
]
conv_fc6_weight = pretrained_state_dict['classifier.0.weight'].view(
4096, 512, 7, 7)
conv_fc6_bias = pretrained_state_dict['classifier.0.bias']
state_dict['conv6.weight'] = decimate(conv_fc6_weight, m=[4, None,
3, 3])
state_dict['conv6.bias'] = decimate(conv_fc6_bias, m=[4])
conv_fc7_weight = pretrained_state_dict['classifier.3.weight'].view(
4096, 4096, 1, 1)
conv_fc7_bias = pretrained_state_dict['classifier.3.bias']
state_dict['conv7.weight'] = decimate(conv_fc7_weight, m=[4, 4,
None, None])
state_dict['conv7.bias'] = decimate(conv_fc7_bias, m=[4])
self.load_state_dict(state_dict)
None
def forward(self, input_0):
primals_1 = self.conv1_1.weight
primals_2 = self.conv1_1.bias
primals_4 = self.conv1_2.weight
primals_5 = self.conv1_2.bias
primals_6 = self.conv2_1.weight
primals_7 = self.conv2_1.bias
primals_8 = self.conv2_2.weight
primals_9 = self.conv2_2.bias
primals_10 = self.conv3_1.weight
primals_11 = self.conv3_1.bias
primals_12 = self.conv3_2.weight
primals_13 = self.conv3_2.bias
primals_14 = self.conv3_3.weight
primals_15 = self.conv3_3.bias
primals_16 = self.conv4_1.weight
primals_17 = self.conv4_1.bias
primals_18 = self.conv4_2.weight
primals_19 = self.conv4_2.bias
primals_20 = self.conv4_3.weight
primals_21 = self.conv4_3.bias
primals_22 = self.conv5_1.weight
primals_23 = self.conv5_1.bias
primals_24 = self.conv5_2.weight
primals_25 = self.conv5_2.bias
primals_26 = self.conv5_3.weight
primals_27 = self.conv5_3.bias
primals_28 = self.conv6.weight
primals_29 = self.conv6.bias
primals_30 = self.conv7.weight
primals_31 = self.conv7.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13, primals_14,
primals_15, primals_16, primals_17, primals_18, primals_19,
primals_20, primals_21, primals_22, primals_23, primals_24,
primals_25, primals_26, primals_27, primals_28, primals_29,
primals_30, primals_31])
return output[0], output[1]
| dee-walia20/SSD-Implementation-using-Pytorch | VGGBase | false | 7,169 | [
"MIT"
] | 1 | 2a7dcdcea2787f4bffd45f335819f08af2b525dd | https://github.com/dee-walia20/SSD-Implementation-using-Pytorch/tree/2a7dcdcea2787f4bffd45f335819f08af2b525dd | import torch
import torchvision
from torch import nn
import torch.nn.functional as F
from itertools import product as product
import torch.optim
import torch.utils.data
def decimate(tensor, m):
"""
Decimate a tensor by a factor 'm', i.e. downsample by keeping every 'm'th value.
This is used when we convert FC layers to equivalent Convolutional layers, BUT of a smaller size.
:param tensor: tensor to be decimated
:param m: list of decimation factors for each dimension of the tensor; None if not to be decimated along a dimension
:return: decimated tensor
"""
assert tensor.dim() == len(m)
for d in range(tensor.dim()):
if m[d] is not None:
tensor = tensor.index_select(dim=d, index=torch.arange(start=0,
end=tensor.size(d), step=m[d]).long())
return tensor
class Model(nn.Module):
"""
VGG base convolutions to produce lower-level feature maps.
"""
def __init__(self):
super().__init__()
self.conv1_1 = nn.Conv2d(3, 64, kernel_size=3, padding=1)
self.conv1_2 = nn.Conv2d(64, 64, kernel_size=3, padding=1)
self.pool1 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv2_1 = nn.Conv2d(64, 128, kernel_size=3, padding=1)
self.conv2_2 = nn.Conv2d(128, 128, kernel_size=3, padding=1)
self.pool2 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv3_1 = nn.Conv2d(128, 256, kernel_size=3, padding=1)
self.conv3_2 = nn.Conv2d(256, 256, kernel_size=3, padding=1)
self.conv3_3 = nn.Conv2d(256, 256, kernel_size=3, padding=1)
self.pool3 = nn.MaxPool2d(kernel_size=2, stride=2, ceil_mode=True)
self.conv4_1 = nn.Conv2d(256, 512, kernel_size=3, padding=1)
self.conv4_2 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv4_3 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.pool4 = nn.MaxPool2d(kernel_size=2, stride=2)
self.conv5_1 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv5_2 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.conv5_3 = nn.Conv2d(512, 512, kernel_size=3, padding=1)
self.pool5 = nn.MaxPool2d(kernel_size=3, stride=1, padding=1)
self.conv6 = nn.Conv2d(512, 1024, kernel_size=3, padding=6, dilation=6)
self.conv7 = nn.Conv2d(1024, 1024, kernel_size=1)
self.load_pretrained_layers()
def forward(self, image):
"""
Forward propagation.
:param image: images, a tensor of dimensions (N, 3, 300, 300)
:return: lower-level feature maps conv4_3 and conv7
"""
out = F.relu(self.conv1_1(image))
out = F.relu(self.conv1_2(out))
out = self.pool1(out)
out = F.relu(self.conv2_1(out))
out = F.relu(self.conv2_2(out))
out = self.pool2(out)
out = F.relu(self.conv3_1(out))
out = F.relu(self.conv3_2(out))
out = F.relu(self.conv3_3(out))
out = self.pool3(out)
out = F.relu(self.conv4_1(out))
out = F.relu(self.conv4_2(out))
out = F.relu(self.conv4_3(out))
conv4_3_feats = out
out = self.pool4(out)
out = F.relu(self.conv5_1(out))
out = F.relu(self.conv5_2(out))
out = F.relu(self.conv5_3(out))
out = self.pool5(out)
out = F.relu(self.conv6(out))
conv7_feats = F.relu(self.conv7(out))
return conv4_3_feats, conv7_feats
def load_pretrained_layers(self):
"""
As in the paper, we use a VGG-16 pretrained on the ImageNet task as the base network.
There's one available in PyTorch, see https://pytorch.org/docs/stable/torchvision/models.html#torchvision.models.vgg16
We copy these parameters into our network. It's straightforward for conv1 to conv5.
However, the original VGG-16 does not contain the conv6 and con7 layers.
Therefore, we convert fc6 and fc7 into convolutional layers, and subsample by decimation. See 'decimate' in utils.py.
# ... truncated (>4000 chars) for memory efficiency |
ComprehensionLayer_step1 | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ow/cowyljxxhhhrqkmxr23ihg4kcpprrlgqunzly74nv22djt5aqkj7.py
# Topologically Sorted Source Nodes: [cated_vectors], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cated_vectors => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%primals_1, %primals_2, %primals_3], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 12
x0 = xindex % 4
x2 = (xindex // 48)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (4*x1) + (16*x2)), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + (4*((-4) + x1)) + (16*x2)), tmp9 & xmask, other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 12, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tl.load(in_ptr2 + (x0 + (4*((-8) + x1)) + (16*x2)), tmp11 & xmask, other=0.0)
tmp15 = tl.where(tmp9, tmp10, tmp14)
tmp16 = tl.where(tmp4, tmp5, tmp15)
tl.store(out_ptr0 + (x3), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/fu/cfux5iy7gwfyitws74d2sqyzewfwia5syjzb4xq2c6sywtehanw5.py
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_1 = async_compile.triton('triton_poi_fused_clone_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (48*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/pq/cpqnfrogm4dnzim2vyszfmugd6fc43gfnmxicoezmiidejzudrdz.py
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# atte_weights => exp
# Graph fragment:
# %mul_tensor_2 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_17, 1), kwargs = {})
# %amax_default_2 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_tensor_2, [-1], True), kwargs = {})
# %sub_tensor_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_tensor_2, %amax_default_2), kwargs = {})
# %div_tensor_2 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_tensor_2, 1.0), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%div_tensor_2,), kwargs = {})
triton_poi_fused__softmax_2 = async_compile.triton('triton_poi_fused__softmax_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp3 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp14 * tmp1
tmp16 = tl_math.exp(tmp15)
tl.store(out_ptr0 + (x2), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ry/cryn7ntc2gpkbfzbre3xh7lffx7zkbskw6oihbzsekkgajmdbki6.py
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# atte_weights => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_3 = async_compile.triton('triton_poi_fused__softmax_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ik/cikz5geipmyqd3kxpjnofzkxsr4kjdrgxpkjbc7og7eef7udyeww.py
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul_2 => clone_4
# Graph fragment:
# %clone_4 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_4,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_4 = async_compile.triton('triton_poi_fused_clone_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_4(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (16 + y0 + (4*x2) + (48*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/6z/c6zedbmwm2d35e7yvn5vdphlli3jag3job4zogz7e4gvhmq5kh6l.py
# Topologically Sorted Source Nodes: [matmul_4], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul_4 => clone_8
# Graph fragment:
# %clone_8 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_8,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_5 = async_compile.triton('triton_poi_fused_clone_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_5(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (32 + y0 + (4*x2) + (48*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/hi/chiryqwbirf2zgxbhddu72p6gbwic76xwpvmrw5z73sc4dacnodq.py
# Topologically Sorted Source Nodes: [cat_1], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat_1 => cat_1
# Graph fragment:
# %cat_1 : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%view_33, %view_34, %view_35], 1), kwargs = {})
triton_poi_fused_cat_6 = async_compile.triton('triton_poi_fused_cat_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_6(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 12
x0 = xindex % 4
x2 = (xindex // 48)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + ((4*x0) + (16*x2) + x1), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + ((4*x0) + (16*x2) + ((-4) + x1)), tmp9 & xmask, eviction_policy='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 12, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tl.load(in_ptr2 + ((4*x0) + (16*x2) + ((-8) + x1)), tmp11 & xmask, eviction_policy='evict_last', other=0.0)
tmp15 = tl.where(tmp9, tmp10, tmp14)
tmp16 = tl.where(tmp4, tmp5, tmp15)
tl.store(out_ptr0 + (x3), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/gz/cgze2dqh77rq4l6otvwzef2pxeh5ld4ph4dekt5m2esoelqh6vvp.py
# Topologically Sorted Source Nodes: [low_vectors_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# low_vectors_1 => add
# Graph fragment:
# %add : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_1, %getitem_9), kwargs = {})
triton_poi_fused_add_7 = async_compile.triton('triton_poi_fused_add_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_7(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = (xindex // 16)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0 + (48*x1)), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/hg/chg5uwxkvmi4rtqt7thpdxsai7g2vlek3ymk6zqqs7s62bl3dhjf.py
# Topologically Sorted Source Nodes: [mid_vectors_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# mid_vectors_1 => add_1
# Graph fragment:
# %add_1 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %getitem_10), kwargs = {})
triton_poi_fused_add_8 = async_compile.triton('triton_poi_fused_add_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_8', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_8(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = (xindex // 16)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (16 + x0 + (48*x1)), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/5k/c5ks6x23whu2i3x3i5iuwqehxdelpbuooovoprqj3xtugqfhvgv6.py
# Topologically Sorted Source Nodes: [hig_vectors_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# hig_vectors_1 => add_2
# Graph fragment:
# %add_2 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_3, %getitem_11), kwargs = {})
triton_poi_fused_add_9 = async_compile.triton('triton_poi_fused_add_9', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_9', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_9(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = (xindex // 16)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (32 + x0 + (48*x1)), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/4n/c4nmkn7t2fqsvg3giqxpycghdjkwlyjfnumznxo5wouesupsbkoc.py
# Topologically Sorted Source Nodes: [low_vectors_2], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# low_vectors_2 => add_3, rsqrt, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add, [2]), kwargs = {correction: 0, keepdim: True})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_12, 1e-08), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_3,), kwargs = {})
triton_poi_fused_native_layer_norm_10 = async_compile.triton('triton_poi_fused_native_layer_norm_10', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_10', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_10(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-08
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x0), tmp8, xmask)
tl.store(out_ptr1 + (x0), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/73/c73z2hfes5izl473wn57vaku4rt2ae7swkdamlriywh5x5xt7g3z.py
# Topologically Sorted Source Nodes: [low_vectors_2], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# low_vectors_2 => add_3, add_4, mul, mul_1, rsqrt, sub_3, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add, [2]), kwargs = {correction: 0, keepdim: True})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_12, 1e-08), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_3,), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %getitem_13), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_3, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_8), kwargs = {})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_9), kwargs = {})
triton_poi_fused_native_layer_norm_11 = async_compile.triton('triton_poi_fused_native_layer_norm_11', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_11', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_11(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (4, ), (1, ))
assert_size_stride(primals_9, (4, ), (1, ))
assert_size_stride(primals_10, (4, ), (1, ))
assert_size_stride(primals_11, (4, ), (1, ))
assert_size_stride(primals_12, (4, ), (1, ))
assert_size_stride(primals_13, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 12, 4), (48, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cated_vectors], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(primals_1, primals_2, primals_3, buf0, 192, grid=grid(192), stream=stream0)
buf1 = empty_strided_cuda((48, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [query], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf0, (48, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((48, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [key], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf0, (48, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf2)
del primals_5
buf3 = empty_strided_cuda((48, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [value], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf0, (48, 4), (4, 1), 0), reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf3)
del primals_6
buf4 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
triton_poi_fused_clone_1.run(buf1, buf4, 16, 4, grid=grid(16, 4), stream=stream0)
buf5 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
triton_poi_fused_clone_1.run(buf2, buf5, 16, 4, grid=grid(16, 4), stream=stream0)
buf6 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf5, (16, 1, 4), (4, 0, 1), 0), out=buf6)
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf6, buf7, 256, grid=grid(256), stream=stream0)
buf8 = reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf6 # reuse
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf7, buf8, 256, grid=grid(256), stream=stream0)
buf9 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
triton_poi_fused_clone_1.run(buf3, buf9, 16, 4, grid=grid(16, 4), stream=stream0)
buf10 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf8, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf9, (16, 4, 1), (4, 1, 0), 0), out=buf10)
buf11 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.clone]
triton_poi_fused_clone_4.run(buf1, buf11, 16, 4, grid=grid(16, 4), stream=stream0)
buf12 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.clone]
triton_poi_fused_clone_4.run(buf2, buf12, 16, 4, grid=grid(16, 4), stream=stream0)
buf13 = reinterpret_tensor(buf7, (16, 4, 4), (16, 4, 1), 0); del buf7 # reuse
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf11, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf12, (16, 1, 4), (4, 0, 1), 0), out=buf13)
buf14 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [atte_weights_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf13, buf14, 256, grid=grid(256), stream=stream0)
buf15 = reinterpret_tensor(buf13, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf13 # reuse
# Topologically Sorted Source Nodes: [atte_weights_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf14, buf15, 256, grid=grid(256), stream=stream0)
buf16 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.clone]
triton_poi_fused_clone_4.run(buf3, buf16, 16, 4, grid=grid(16, 4), stream=stream0)
buf17 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf15, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf16, (16, 4, 1), (4, 1, 0), 0), out=buf17)
buf18 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_4], Original ATen: [aten.clone]
triton_poi_fused_clone_5.run(buf1, buf18, 16, 4, grid=grid(16, 4), stream=stream0)
del buf1
buf19 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_4], Original ATen: [aten.clone]
triton_poi_fused_clone_5.run(buf2, buf19, 16, 4, grid=grid(16, 4), stream=stream0)
buf20 = reinterpret_tensor(buf14, (16, 4, 4), (16, 4, 1), 0); del buf14 # reuse
# Topologically Sorted Source Nodes: [matmul_4], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf18, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf19, (16, 1, 4), (4, 0, 1), 0), out=buf20)
buf21 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [atte_weights_4], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf20, buf21, 256, grid=grid(256), stream=stream0)
buf22 = reinterpret_tensor(buf20, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf20 # reuse
# Topologically Sorted Source Nodes: [atte_weights_4], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf21, buf22, 256, grid=grid(256), stream=stream0)
del buf21
buf23 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output_2], Original ATen: [aten.clone]
triton_poi_fused_clone_5.run(buf3, buf23, 16, 4, grid=grid(16, 4), stream=stream0)
buf24 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output_2], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf22, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf23, (16, 4, 1), (4, 1, 0), 0), out=buf24)
buf25 = reinterpret_tensor(buf3, (4, 12, 4), (48, 4, 1), 0); del buf3 # reuse
# Topologically Sorted Source Nodes: [cat_1], Original ATen: [aten.cat]
triton_poi_fused_cat_6.run(buf10, buf17, buf24, buf25, 192, grid=grid(192), stream=stream0)
buf26 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf25, (48, 4), (4, 1), 0), reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf26)
buf27 = reinterpret_tensor(buf24, (4, 4, 4), (16, 4, 1), 0); del buf24 # reuse
# Topologically Sorted Source Nodes: [low_vectors_1], Original ATen: [aten.add]
triton_poi_fused_add_7.run(primals_1, buf26, buf27, 64, grid=grid(64), stream=stream0)
del primals_1
buf28 = reinterpret_tensor(buf17, (4, 4, 4), (16, 4, 1), 0); del buf17 # reuse
# Topologically Sorted Source Nodes: [mid_vectors_1], Original ATen: [aten.add]
triton_poi_fused_add_8.run(primals_2, buf26, buf28, 64, grid=grid(64), stream=stream0)
del primals_2
buf29 = reinterpret_tensor(buf10, (4, 4, 4), (16, 4, 1), 0); del buf10 # reuse
# Topologically Sorted Source Nodes: [hig_vectors_1], Original ATen: [aten.add]
triton_poi_fused_add_9.run(primals_3, buf26, buf29, 64, grid=grid(64), stream=stream0)
del buf26
del primals_3
buf30 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf31 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
# Topologically Sorted Source Nodes: [low_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_10.run(buf27, buf30, buf31, 16, grid=grid(16), stream=stream0)
buf32 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [low_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_11.run(buf27, buf30, buf31, primals_8, primals_9, buf32, 64, grid=grid(64), stream=stream0)
del primals_9
buf33 = buf31; del buf31 # reuse
buf34 = buf30; del buf30 # reuse
# Topologically Sorted Source Nodes: [mid_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_10.run(buf28, buf33, buf34, 16, grid=grid(16), stream=stream0)
buf35 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mid_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_11.run(buf28, buf33, buf34, primals_10, primals_11, buf35, 64, grid=grid(64), stream=stream0)
del primals_11
buf36 = buf34; del buf34 # reuse
buf37 = buf33; del buf33 # reuse
# Topologically Sorted Source Nodes: [hig_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_10.run(buf29, buf36, buf37, 16, grid=grid(16), stream=stream0)
buf38 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [hig_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_11.run(buf29, buf36, buf37, primals_12, primals_13, buf38, 64, grid=grid(64), stream=stream0)
del buf36
del buf37
del primals_13
return (buf32, buf35, buf38, buf8, buf15, buf22, primals_8, primals_10, primals_12, reinterpret_tensor(buf0, (48, 4), (4, 1), 0), buf8, buf15, buf22, reinterpret_tensor(buf25, (48, 4), (4, 1), 0), buf27, buf28, buf29, primals_7, reinterpret_tensor(buf23, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf18, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf19, (16, 4, 1), (4, 1, 4), 0), reinterpret_tensor(buf16, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf11, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf12, (16, 4, 1), (4, 1, 4), 0), reinterpret_tensor(buf9, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf4, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf5, (16, 4, 1), (4, 1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super(ScaledDotProductAttention, self).__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(MultiHeadAttention, self).__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class ComprehensionLayer_step1(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(ComprehensionLayer_step1, self).__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.low_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, low_vectors, mid_vectors, hig_vectors):
b = low_vectors.size()[0]
low_num, mid_num, hig_num = low_vectors.size()[1], mid_vectors.size()[1
], hig_vectors.size()[1]
low_residual = low_vectors
mid_residual = mid_vectors
hig_residual = hig_vectors
cated_vectors = torch.cat((low_vectors, mid_vectors, hig_vectors),
dim=1)
query = self.Wq(cated_vectors)
key = self.Wk(cated_vectors)
value = self.Wv(cated_vectors)
low_query, mid_query, hig_query = torch.split(query, [low_num,
mid_num, hig_num], dim=1)
low_key, mid_key, hig_key = torch.split(key, [low_num, mid_num,
hig_num], dim=1)
low_value, mid_value, hig_value = torch.split(value, [low_num,
mid_num, hig_num], dim=1)
low_query = low_query.reshape(b, low_num, self.n_head, self.
reduced_dim // self.n_head)
low_key = low_key.reshape(b, low_num, self.n_head, self.reduced_dim //
self.n_head)
low_value = low_value.reshape(b, low_num, self.n_head, self.
reduced_dim // self.n_head)
low_query = low_query.transpose(1, 2)
low_key = low_key.transpose(1, 2)
low_value = low_value.transpose(1, 2)
mid_query = mid_query.reshape(b, mid_num, self.n_head, self.
reduced_dim // self.n_head)
mid_key = mid_key.reshape(b, mid_num, self.n_head, self.reduced_dim //
self.n_head)
mid_value = mid_value.reshape(b, mid_num, self.n_head, self.
reduced_dim // self.n_head)
mid_query = mid_query.transpose(1, 2)
mid_key = mid_key.transpose(1, 2)
mid_value = mid_value.transpose(1, 2)
hig_query = hig_query.reshape(b, hig_num, self.n_head, self.
reduced_dim // self.n_head)
hig_key = hig_key.reshape(b, hig_num, self.n_head, self.reduced_dim //
self.n_head)
hig_value = hig_value.reshape(b, hig_num, self.n_head, self.
reduced_dim // self.n_head)
hig_query = hig_query.transpose(1, 2)
hig_key = hig_key.transpose(1, 2)
hig_value = hig_value.transpose(1, 2)
low_query, low_weights = self.inner_attention(low_query, low_key,
low_value)
mid_query, mid_weights = self.inner_attention(mid_query, mid_key,
mid_value)
hig_query, hig_weights = self.inner_attention(hig_query, hig_key,
hig_value)
low_query = low_query.transpose(1, 2).reshape(b, low_num, self.
reduced_dim)
mid_query = mid_query.transpose(1, 2).reshape(b, mid_num, self.
reduced_dim)
hig_query = hig_query.transpose(1, 2).reshape(b, hig_num, self.
reduced_dim)
output = self.dropout(self.Wo(torch.cat((low_query, mid_query,
hig_query), dim=1)))
low_vectors, mid_vectors, hig_vectors = torch.split(output, [
low_num, mid_num, hig_num], dim=1)
low_vectors = low_residual + low_vectors
mid_vectors = mid_residual + mid_vectors
hig_vectors = hig_residual + hig_vectors
low_vectors = self.low_ln(low_vectors)
mid_vectors = self.mid_ln(mid_vectors)
hig_vectors = self.hig_ln(hig_vectors)
return (low_vectors, mid_vectors, hig_vectors, low_weights,
mid_weights, hig_weights)
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4]), torch.rand([4, 4, 4])
]
def get_init_inputs():
return [[], {'embedding_dim': 4, 'reduced_dim': 4, 'n_head': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 12
x0 = xindex % 4
x2 = xindex // 48
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 4 * x1 + 16 * x2), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + 4 * (-4 + x1) + 16 * x2), tmp9 & xmask,
other=0.0)
tmp11 = tmp0 >= tmp7
tl.full([1], 12, tl.int64)
tmp14 = tl.load(in_ptr2 + (x0 + 4 * (-8 + x1) + 16 * x2), tmp11 & xmask,
other=0.0)
tmp15 = tl.where(tmp9, tmp10, tmp14)
tmp16 = tl.where(tmp4, tmp5, tmp15)
tl.store(out_ptr0 + x3, tmp16, xmask)
@triton.jit
def triton_poi_fused_clone_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 48 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp3 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp14 * tmp1
tmp16 = tl_math.exp(tmp15)
tl.store(out_ptr0 + x2, tmp16, xmask)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_clone_4(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (16 + y0 + 4 * x2 + 48 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_clone_5(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (32 + y0 + 4 * x2 + 48 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_cat_6(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 12
x0 = xindex % 4
x2 = xindex // 48
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (4 * x0 + 16 * x2 + x1), tmp4 & xmask,
eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (4 * x0 + 16 * x2 + (-4 + x1)), tmp9 & xmask,
eviction_policy='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tl.full([1], 12, tl.int64)
tmp14 = tl.load(in_ptr2 + (4 * x0 + 16 * x2 + (-8 + x1)), tmp11 & xmask,
eviction_policy='evict_last', other=0.0)
tmp15 = tl.where(tmp9, tmp10, tmp14)
tmp16 = tl.where(tmp4, tmp5, tmp15)
tl.store(out_ptr0 + x3, tmp16, xmask)
@triton.jit
def triton_poi_fused_add_7(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = xindex // 16
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (x0 + 48 * x1), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_add_8(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = xindex // 16
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (16 + x0 + 48 * x1), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_add_9(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = xindex // 16
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (32 + x0 + 48 * x1), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_10(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-08
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x0, tmp8, xmask)
tl.store(out_ptr1 + x0, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_11(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (4,), (1,))
assert_size_stride(primals_9, (4,), (1,))
assert_size_stride(primals_10, (4,), (1,))
assert_size_stride(primals_11, (4,), (1,))
assert_size_stride(primals_12, (4,), (1,))
assert_size_stride(primals_13, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 12, 4), (48, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(192)](primals_1, primals_2, primals_3,
buf0, 192, XBLOCK=128, num_warps=4, num_stages=1)
buf1 = empty_strided_cuda((48, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf0, (48, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((48, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf0, (48, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf2)
del primals_5
buf3 = empty_strided_cuda((48, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf0, (48, 4), (4, 1), 0),
reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf3)
del primals_6
buf4 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_1[grid(16, 4)](buf1, buf4, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf5 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
triton_poi_fused_clone_1[grid(16, 4)](buf2, buf5, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf6 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf5, (16, 1, 4), (4, 0, 1), 0), out=buf6)
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_2[grid(256)](buf6, buf7, 256, XBLOCK=256,
num_warps=4, num_stages=1)
buf8 = reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf6
triton_poi_fused__softmax_3[grid(256)](buf7, buf8, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf9 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_1[grid(16, 4)](buf3, buf9, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf10 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf8, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf9, (16, 4, 1), (4, 1, 0), 0), out=buf10)
buf11 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_4[grid(16, 4)](buf1, buf11, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf12 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
triton_poi_fused_clone_4[grid(16, 4)](buf2, buf12, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf13 = reinterpret_tensor(buf7, (16, 4, 4), (16, 4, 1), 0)
del buf7
extern_kernels.bmm(reinterpret_tensor(buf11, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf12, (16, 1, 4), (4, 0, 1), 0), out=buf13)
buf14 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_2[grid(256)](buf13, buf14, 256, XBLOCK=
256, num_warps=4, num_stages=1)
buf15 = reinterpret_tensor(buf13, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf13
triton_poi_fused__softmax_3[grid(256)](buf14, buf15, 256, XBLOCK=
128, num_warps=4, num_stages=1)
buf16 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_4[grid(16, 4)](buf3, buf16, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf17 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf15, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf16, (16, 4, 1), (4, 1, 0), 0), out=buf17)
buf18 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_5[grid(16, 4)](buf1, buf18, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
del buf1
buf19 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
triton_poi_fused_clone_5[grid(16, 4)](buf2, buf19, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf20 = reinterpret_tensor(buf14, (16, 4, 4), (16, 4, 1), 0)
del buf14
extern_kernels.bmm(reinterpret_tensor(buf18, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf19, (16, 1, 4), (4, 0, 1), 0), out=buf20)
buf21 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_2[grid(256)](buf20, buf21, 256, XBLOCK=
256, num_warps=4, num_stages=1)
buf22 = reinterpret_tensor(buf20, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf20
triton_poi_fused__softmax_3[grid(256)](buf21, buf22, 256, XBLOCK=
128, num_warps=4, num_stages=1)
del buf21
buf23 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_5[grid(16, 4)](buf3, buf23, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf24 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf22, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf23, (16, 4, 1), (4, 1, 0), 0), out=buf24)
buf25 = reinterpret_tensor(buf3, (4, 12, 4), (48, 4, 1), 0)
del buf3
triton_poi_fused_cat_6[grid(192)](buf10, buf17, buf24, buf25, 192,
XBLOCK=128, num_warps=4, num_stages=1)
buf26 = buf2
del buf2
extern_kernels.mm(reinterpret_tensor(buf25, (48, 4), (4, 1), 0),
reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf26)
buf27 = reinterpret_tensor(buf24, (4, 4, 4), (16, 4, 1), 0)
del buf24
triton_poi_fused_add_7[grid(64)](primals_1, buf26, buf27, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_1
buf28 = reinterpret_tensor(buf17, (4, 4, 4), (16, 4, 1), 0)
del buf17
triton_poi_fused_add_8[grid(64)](primals_2, buf26, buf28, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_2
buf29 = reinterpret_tensor(buf10, (4, 4, 4), (16, 4, 1), 0)
del buf10
triton_poi_fused_add_9[grid(64)](primals_3, buf26, buf29, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del buf26
del primals_3
buf30 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf31 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
triton_poi_fused_native_layer_norm_10[grid(16)](buf27, buf30, buf31,
16, XBLOCK=16, num_warps=1, num_stages=1)
buf32 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_11[grid(64)](buf27, buf30, buf31,
primals_8, primals_9, buf32, 64, XBLOCK=64, num_warps=1,
num_stages=1)
del primals_9
buf33 = buf31
del buf31
buf34 = buf30
del buf30
triton_poi_fused_native_layer_norm_10[grid(16)](buf28, buf33, buf34,
16, XBLOCK=16, num_warps=1, num_stages=1)
buf35 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_11[grid(64)](buf28, buf33, buf34,
primals_10, primals_11, buf35, 64, XBLOCK=64, num_warps=1,
num_stages=1)
del primals_11
buf36 = buf34
del buf34
buf37 = buf33
del buf33
triton_poi_fused_native_layer_norm_10[grid(16)](buf29, buf36, buf37,
16, XBLOCK=16, num_warps=1, num_stages=1)
buf38 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_11[grid(64)](buf29, buf36, buf37,
primals_12, primals_13, buf38, 64, XBLOCK=64, num_warps=1,
num_stages=1)
del buf36
del buf37
del primals_13
return (buf32, buf35, buf38, buf8, buf15, buf22, primals_8, primals_10,
primals_12, reinterpret_tensor(buf0, (48, 4), (4, 1), 0), buf8,
buf15, buf22, reinterpret_tensor(buf25, (48, 4), (4, 1), 0), buf27,
buf28, buf29, primals_7, reinterpret_tensor(buf23, (16, 1, 4), (4,
1, 1), 0), reinterpret_tensor(buf18, (16, 1, 4), (4, 1, 1), 0),
reinterpret_tensor(buf19, (16, 4, 1), (4, 1, 4), 0),
reinterpret_tensor(buf16, (16, 1, 4), (4, 1, 1), 0),
reinterpret_tensor(buf11, (16, 1, 4), (4, 1, 1), 0),
reinterpret_tensor(buf12, (16, 4, 1), (4, 1, 4), 0),
reinterpret_tensor(buf9, (16, 1, 4), (4, 1, 1), 0),
reinterpret_tensor(buf4, (16, 1, 4), (4, 1, 1), 0),
reinterpret_tensor(buf5, (16, 4, 1), (4, 1, 4), 0))
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super(ScaledDotProductAttention, self).__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(MultiHeadAttention, self).__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class ComprehensionLayer_step1New(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(ComprehensionLayer_step1New, self).__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.low_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, input_0, input_1, input_2):
primals_4 = self.Wq.weight
primals_5 = self.Wk.weight
primals_6 = self.Wv.weight
primals_7 = self.Wo.weight
primals_8 = self.low_ln.weight
primals_9 = self.low_ln.bias
primals_10 = self.mid_ln.weight
primals_11 = self.mid_ln.bias
primals_12 = self.hig_ln.weight
primals_13 = self.hig_ln.bias
primals_1 = input_0
primals_2 = input_1
primals_3 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13])
return output[0], output[1], output[2], output[3], output[4], output[5]
| luyu-fan/LRCM | ComprehensionLayer_step1 | false | 7,170 | [
"MIT"
] | 1 | 6b0e4d7998bc4969afa764eb753077e3f858f1ba | https://github.com/luyu-fan/LRCM/tree/6b0e4d7998bc4969afa764eb753077e3f858f1ba | import math
import torch
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super().__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super().__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class Model(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super().__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.low_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, low_vectors, mid_vectors, hig_vectors):
b = low_vectors.size()[0]
low_num, mid_num, hig_num = low_vectors.size()[1], mid_vectors.size()[1
], hig_vectors.size()[1]
low_residual = low_vectors
mid_residual = mid_vectors
hig_residual = hig_vectors
cated_vectors = torch.cat((low_vectors, mid_vectors, hig_vectors),
dim=1)
query = self.Wq(cated_vectors)
key = self.Wk(cated_vectors)
value = self.Wv(cated_vectors)
low_query, mid_query, hig_query = torch.split(query, [low_num,
mid_num, hig_num], dim=1)
low_key, mid_key, hig_key = torch.split(key, [low_num, mid_num,
hig_num], dim=1)
low_value, mid_value, hig_value = torch.split(value, [low_num,
mid_num, hig_num], dim=1)
low_query = low_query.reshape(b, low_num, self.n_head, self.
reduced_dim // self.n_head)
low_key = low_key.reshape(b, low_num, self.n_head, self.reduced_dim //
self.n_head)
low_value = low_value.reshape(b, low_num, self.n_head, self.
reduced_dim // self.n_head)
low_query = low_query.transpose(1, 2)
low_key = low_key.transpose(1, 2)
low_value = low_value.transpose(1, 2)
mid_que
# ... truncated (>4000 chars) for memory efficiency |
PyramidUp | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/oj/cojl5mb3pzv5jbmfzjkbac5hekbmpvb72kof6ouyyasitrogdd6n.py
# Topologically Sorted Source Nodes: [upsample], Original ATen: [aten._unsafe_index]
# Source node to ATen node mapping:
# upsample => _unsafe_index
# Graph fragment:
# %_unsafe_index : [num_users=4] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%arg0_1, [None, None, %unsqueeze, %convert_element_type_3]), kwargs = {})
triton_poi_fused__unsafe_index_0 = async_compile.triton('triton_poi_fused__unsafe_index_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__unsafe_index_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__unsafe_index_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 8) % 8
x0 = xindex % 8
x2 = (xindex // 64)
x4 = xindex
tmp0 = x1
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 * tmp2
tmp4 = tmp3.to(tl.int32)
tmp5 = x0
tmp6 = tmp5.to(tl.float32)
tmp7 = tmp6 * tmp2
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.load(in_ptr0 + (tmp8 + (4*tmp4) + (16*x2)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x4), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/tf/ctfpfu7wmre7nsn3ualydv64rwcj6t2a5ldiko4qeb7hyfhiqx56.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%convolution, %convolution_1, %convolution_2, %convolution_3], 1), kwargs = {})
triton_poi_fused_cat_1 = async_compile.triton('triton_poi_fused_cat_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 64) % 4
x0 = xindex % 64
x2 = (xindex // 256)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (64*x2)), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + (64*x2)), tmp9 & xmask, eviction_policy='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x0 + (64*x2)), tmp14 & xmask, eviction_policy='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tmp17 = tl.full([1], 4, tl.int64)
tmp18 = tmp0 < tmp17
tmp19 = tl.load(in_ptr3 + (x0 + (64*x2)), tmp16 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + (x3), tmp22, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (1, 1, 5, 5), (25, 25, 5, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 8, 8), (256, 64, 8, 1), torch.float32)
# Topologically Sorted Source Nodes: [upsample], Original ATen: [aten._unsafe_index]
stream0 = get_raw_stream(0)
triton_poi_fused__unsafe_index_0.run(arg0_1, buf0, 1024, grid=grid(1024), stream=stream0)
del arg0_1
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(reinterpret_tensor(buf0, (4, 1, 8, 8), (256, 0, 8, 1), 0), arg1_1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 1, 8, 8), (64, 64, 8, 1))
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(reinterpret_tensor(buf0, (4, 1, 8, 8), (256, 64, 8, 1), 64), arg1_1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 8, 8), (64, 64, 8, 1))
# Topologically Sorted Source Nodes: [conv2d_2], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(reinterpret_tensor(buf0, (4, 1, 8, 8), (256, 64, 8, 1), 128), arg1_1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 1, 8, 8), (64, 64, 8, 1))
# Topologically Sorted Source Nodes: [conv2d_3], Original ATen: [aten.convolution]
buf4 = extern_kernels.convolution(reinterpret_tensor(buf0, (4, 1, 8, 8), (256, 64, 8, 1), 192), arg1_1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf4, (4, 1, 8, 8), (64, 64, 8, 1))
del arg1_1
buf5 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
triton_poi_fused_cat_1.run(buf1, buf2, buf3, buf4, buf5, 1024, grid=grid(1024), stream=stream0)
del buf1
del buf2
del buf3
del buf4
return (buf5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((1, 1, 5, 5), (25, 25, 5, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
from torch.nn import functional as F
class PyramidUp(nn.Module):
def __init__(self) ->None:
super(PyramidUp, self).__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
upsample = F.interpolate(x, scale_factor=2)
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(upsample[:, i:i + 1, :, :], self.filter,
padding=2))
return torch.cat(results, dim=1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__unsafe_index_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 8 % 8
x0 = xindex % 8
x2 = xindex // 64
x4 = xindex
tmp0 = x1
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 * tmp2
tmp4 = tmp3.to(tl.int32)
tmp5 = x0
tmp6 = tmp5.to(tl.float32)
tmp7 = tmp6 * tmp2
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.load(in_ptr0 + (tmp8 + 4 * tmp4 + 16 * x2), xmask,
eviction_policy='evict_last')
tl.store(out_ptr0 + x4, tmp9, xmask)
@triton.jit
def triton_poi_fused_cat_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 64 % 4
x0 = xindex % 64
x2 = xindex // 256
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 64 * x2), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + 64 * x2), tmp9 & xmask, eviction_policy
='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x0 + 64 * x2), tmp14 & xmask,
eviction_policy='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tl.full([1], 4, tl.int64)
tmp19 = tl.load(in_ptr3 + (x0 + 64 * x2), tmp16 & xmask,
eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + x3, tmp22, xmask)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (1, 1, 5, 5), (25, 25, 5, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 8, 8), (256, 64, 8, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__unsafe_index_0[grid(1024)](arg0_1, buf0, 1024,
XBLOCK=256, num_warps=4, num_stages=1)
del arg0_1
buf1 = extern_kernels.convolution(reinterpret_tensor(buf0, (4, 1, 8,
8), (256, 0, 8, 1), 0), arg1_1, stride=(1, 1), padding=(2, 2),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf1, (4, 1, 8, 8), (64, 64, 8, 1))
buf2 = extern_kernels.convolution(reinterpret_tensor(buf0, (4, 1, 8,
8), (256, 64, 8, 1), 64), arg1_1, stride=(1, 1), padding=(2, 2),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 8, 8), (64, 64, 8, 1))
buf3 = extern_kernels.convolution(reinterpret_tensor(buf0, (4, 1, 8,
8), (256, 64, 8, 1), 128), arg1_1, stride=(1, 1), padding=(2, 2
), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf3, (4, 1, 8, 8), (64, 64, 8, 1))
buf4 = extern_kernels.convolution(reinterpret_tensor(buf0, (4, 1, 8,
8), (256, 64, 8, 1), 192), arg1_1, stride=(1, 1), padding=(2, 2
), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf4, (4, 1, 8, 8), (64, 64, 8, 1))
del arg1_1
buf5 = buf0
del buf0
triton_poi_fused_cat_1[grid(1024)](buf1, buf2, buf3, buf4, buf5,
1024, XBLOCK=128, num_warps=4, num_stages=1)
del buf1
del buf2
del buf3
del buf4
return buf5,
class PyramidUpNew(nn.Module):
def __init__(self) ->None:
super(PyramidUpNew, self).__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, input_0):
arg1_1 = self.filter
arg0_1 = input_0
output = call([arg0_1, arg1_1])
return output[0]
| masanorihirano/pytorch_extra_mhirano | PyramidUp | false | 7,171 | [
"MIT"
] | 1 | d19e07445567c069793b7ca1a22a846d7cbce58d | https://github.com/masanorihirano/pytorch_extra_mhirano/tree/d19e07445567c069793b7ca1a22a846d7cbce58d | import torch
import torch.nn as nn
from torch.nn import functional as F
class Model(nn.Module):
def __init__(self) ->None:
super().__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
upsample = F.interpolate(x, scale_factor=2)
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(upsample[:, i:i + 1, :, :], self.filter,
padding=2))
return torch.cat(results, dim=1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
ComprehensionLayer_step2 | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/7j/c7jury2hid4lpdpl3qjbeeviqbffqxg2vp7bn64rpbpyjzvft5k3.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%primals_2, %primals_3], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 8
x0 = xindex % 4
x2 = (xindex // 32)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (4*x1) + (16*x2)), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr1 + (x0 + (4*((-4) + x1)) + (16*x2)), tmp6 & xmask, other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + (x3), tmp10, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/qg/cqgxf25farkr5u4ohhxxpb4hhbzxrbznilkzcx5b7euz6n5daguo.py
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_1 = async_compile.triton('triton_poi_fused_clone_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x2) + (32*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/pq/cpqnfrogm4dnzim2vyszfmugd6fc43gfnmxicoezmiidejzudrdz.py
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# atte_weights => exp
# Graph fragment:
# %mul_tensor_1 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_14, 1), kwargs = {})
# %amax_default_1 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_tensor_1, [-1], True), kwargs = {})
# %sub_tensor_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_tensor_1, %amax_default_1), kwargs = {})
# %div_tensor_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_tensor_1, 1.0), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%div_tensor_1,), kwargs = {})
triton_poi_fused__softmax_2 = async_compile.triton('triton_poi_fused__softmax_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp3 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp14 * tmp1
tmp16 = tl_math.exp(tmp15)
tl.store(out_ptr0 + (x2), tmp16, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ry/cryn7ntc2gpkbfzbre3xh7lffx7zkbskw6oihbzsekkgajmdbki6.py
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# atte_weights => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_3 = async_compile.triton('triton_poi_fused__softmax_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/hn/chn3ye3i7w5cpzxsptflgjdgyjx6j2ypvogbnh4bhpvf7lszwu6k.py
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# matmul_2 => clone_4
# Graph fragment:
# %clone_4 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_4,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_4 = async_compile.triton('triton_poi_fused_clone_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_4(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = (yindex // 4)
y3 = yindex
tmp0 = tl.load(in_ptr0 + (16 + y0 + (4*x2) + (32*y1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + (4*y3)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/2z/c2zikqwxmvg44bd3fza5ifapibefnqdnry3euzznggqb3w4ts7cs.py
# Topologically Sorted Source Nodes: [cat_3], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat_3 => cat_3
# Graph fragment:
# %cat_3 : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%view_24, %view_25], 1), kwargs = {})
triton_poi_fused_cat_5 = async_compile.triton('triton_poi_fused_cat_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_5(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 8
x0 = xindex % 4
x2 = (xindex // 32)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + ((4*x0) + (16*x2) + x1), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr1 + ((4*x0) + (16*x2) + ((-4) + x1)), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + (x3), tmp10, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/3u/c3u4j73fpz5jymyhzjjmmrfvwqhshzjtzll7i6khdsrgkpakyxvs.py
# Topologically Sorted Source Nodes: [mid_vectors_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# mid_vectors_1 => add
# Graph fragment:
# %add : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %getitem_6), kwargs = {})
triton_poi_fused_add_6 = async_compile.triton('triton_poi_fused_add_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_6(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = (xindex // 16)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0 + (32*x1)), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/7z/c7zq3k45gqeqgraar3vhubrfnclwsr2hnjpvf2pjxrlrjw3z5alj.py
# Topologically Sorted Source Nodes: [hig_vectors_1], Original ATen: [aten.add]
# Source node to ATen node mapping:
# hig_vectors_1 => add_1
# Graph fragment:
# %add_1 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_3, %getitem_7), kwargs = {})
triton_poi_fused_add_7 = async_compile.triton('triton_poi_fused_add_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_7(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = (xindex // 16)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (16 + x0 + (32*x1)), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ej/cej3lg77y73a2isa3ob65cyo32xdfs6l7ngaykptyx5ifjbpupjn.py
# Topologically Sorted Source Nodes: [mid_vectors_2], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# mid_vectors_2 => add_2, rsqrt, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add, [2]), kwargs = {correction: 0, keepdim: True})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_8, 1e-08), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_2,), kwargs = {})
triton_poi_fused_native_layer_norm_8 = async_compile.triton('triton_poi_fused_native_layer_norm_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_8', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_8(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-08
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x0), tmp8, xmask)
tl.store(out_ptr1 + (x0), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/4d/c4dqmytly5zolisxgude22db5p3pbommvxualmvn6ercwtoppkm7.py
# Topologically Sorted Source Nodes: [mid_vectors_2], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# mid_vectors_2 => add_2, add_3, mul, mul_1, rsqrt, sub_2, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add, [2]), kwargs = {correction: 0, keepdim: True})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_8, 1e-08), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_2,), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %getitem_9), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_2, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_8), kwargs = {})
# %add_3 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_9), kwargs = {})
triton_poi_fused_native_layer_norm_9 = async_compile.triton('triton_poi_fused_native_layer_norm_9', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_9', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_9(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (4, ), (1, ))
assert_size_stride(primals_9, (4, ), (1, ))
assert_size_stride(primals_10, (4, ), (1, ))
assert_size_stride(primals_11, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(primals_2, primals_3, buf0, 128, grid=grid(128), stream=stream0)
buf1 = empty_strided_cuda((32, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [query], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf0, (32, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat_1], Original ATen: [aten.cat]
triton_poi_fused_cat_0.run(primals_1, primals_2, buf2, 128, grid=grid(128), stream=stream0)
del primals_1
buf3 = empty_strided_cuda((32, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [key], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf2, (32, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf3)
del primals_5
buf4 = empty_strided_cuda((32, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [value], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf2, (32, 4), (4, 1), 0), reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf4)
del primals_6
buf5 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
triton_poi_fused_clone_1.run(buf1, buf5, 16, 4, grid=grid(16, 4), stream=stream0)
buf6 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.clone]
triton_poi_fused_clone_1.run(buf3, buf6, 16, 4, grid=grid(16, 4), stream=stream0)
buf7 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf5, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf6, (16, 1, 4), (4, 0, 1), 0), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf7, buf8, 256, grid=grid(256), stream=stream0)
buf9 = reinterpret_tensor(buf7, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf7 # reuse
# Topologically Sorted Source Nodes: [atte_weights], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf8, buf9, 256, grid=grid(256), stream=stream0)
buf10 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.clone]
triton_poi_fused_clone_1.run(buf4, buf10, 16, 4, grid=grid(16, 4), stream=stream0)
buf11 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf9, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf10, (16, 4, 1), (4, 1, 0), 0), out=buf11)
buf12 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.clone]
triton_poi_fused_clone_4.run(buf1, buf12, 16, 4, grid=grid(16, 4), stream=stream0)
del buf1
buf13 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.clone]
triton_poi_fused_clone_4.run(buf3, buf13, 16, 4, grid=grid(16, 4), stream=stream0)
buf14 = reinterpret_tensor(buf8, (16, 4, 4), (16, 4, 1), 0); del buf8 # reuse
# Topologically Sorted Source Nodes: [matmul_2], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf12, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf13, (16, 1, 4), (4, 0, 1), 0), out=buf14)
buf15 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [atte_weights_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf14, buf15, 256, grid=grid(256), stream=stream0)
buf16 = reinterpret_tensor(buf14, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf14 # reuse
# Topologically Sorted Source Nodes: [atte_weights_2], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf15, buf16, 256, grid=grid(256), stream=stream0)
del buf15
buf17 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.clone]
triton_poi_fused_clone_4.run(buf4, buf17, 16, 4, grid=grid(16, 4), stream=stream0)
buf18 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output_1], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf16, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf17, (16, 4, 1), (4, 1, 0), 0), out=buf18)
buf19 = reinterpret_tensor(buf4, (4, 8, 4), (32, 4, 1), 0); del buf4 # reuse
# Topologically Sorted Source Nodes: [cat_3], Original ATen: [aten.cat]
triton_poi_fused_cat_5.run(buf11, buf18, buf19, 128, grid=grid(128), stream=stream0)
buf20 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [linear_3], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf19, (32, 4), (4, 1), 0), reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf20)
buf21 = reinterpret_tensor(buf18, (4, 4, 4), (16, 4, 1), 0); del buf18 # reuse
# Topologically Sorted Source Nodes: [mid_vectors_1], Original ATen: [aten.add]
triton_poi_fused_add_6.run(primals_2, buf20, buf21, 64, grid=grid(64), stream=stream0)
del primals_2
buf22 = reinterpret_tensor(buf11, (4, 4, 4), (16, 4, 1), 0); del buf11 # reuse
# Topologically Sorted Source Nodes: [hig_vectors_1], Original ATen: [aten.add]
triton_poi_fused_add_7.run(primals_3, buf20, buf22, 64, grid=grid(64), stream=stream0)
del buf20
del primals_3
buf23 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf24 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
# Topologically Sorted Source Nodes: [mid_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_8.run(buf21, buf23, buf24, 16, grid=grid(16), stream=stream0)
buf25 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mid_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_9.run(buf21, buf23, buf24, primals_8, primals_9, buf25, 64, grid=grid(64), stream=stream0)
del primals_9
buf26 = buf24; del buf24 # reuse
buf27 = buf23; del buf23 # reuse
# Topologically Sorted Source Nodes: [hig_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_8.run(buf22, buf26, buf27, 16, grid=grid(16), stream=stream0)
buf28 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [hig_vectors_2], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_9.run(buf22, buf26, buf27, primals_10, primals_11, buf28, 64, grid=grid(64), stream=stream0)
del buf26
del buf27
del primals_11
return (buf25, buf28, buf9, buf16, primals_8, primals_10, reinterpret_tensor(buf0, (32, 4), (4, 1), 0), reinterpret_tensor(buf2, (32, 4), (4, 1), 0), buf9, buf16, reinterpret_tensor(buf19, (32, 4), (4, 1), 0), buf21, buf22, primals_7, reinterpret_tensor(buf17, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf12, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf13, (16, 4, 1), (4, 1, 4), 0), reinterpret_tensor(buf10, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf5, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf6, (16, 4, 1), (4, 1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super(ScaledDotProductAttention, self).__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(MultiHeadAttention, self).__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class ComprehensionLayer_step2(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(ComprehensionLayer_step2, self).__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, low_vectors, mid_vectors, hig_vectors):
b = low_vectors.size()[0]
low_num, mid_num, hig_num = low_vectors.size()[1], mid_vectors.size()[1
], hig_vectors.size()[1]
mid_residual = mid_vectors
hig_residual = hig_vectors
query = self.Wq(torch.cat((mid_vectors, hig_vectors), dim=1))
key = self.Wk(torch.cat((low_vectors, mid_vectors), dim=1))
value = self.Wv(torch.cat((low_vectors, mid_vectors), dim=1))
mid_query, hig_query = torch.split(query, [mid_num, hig_num], dim=1)
low_key, mid_key = torch.split(key, [low_num, mid_num], dim=1)
low_value, mid_value = torch.split(value, [low_num, mid_num], dim=1)
low_key = low_key.reshape(b, low_num, self.n_head, self.reduced_dim //
self.n_head)
low_value = low_value.reshape(b, low_num, self.n_head, self.
reduced_dim // self.n_head)
low_key = low_key.transpose(1, 2)
low_value = low_value.transpose(1, 2)
mid_query = mid_query.reshape(b, mid_num, self.n_head, self.
reduced_dim // self.n_head)
mid_key = mid_key.reshape(b, mid_num, self.n_head, self.reduced_dim //
self.n_head)
mid_value = mid_value.reshape(b, mid_num, self.n_head, self.
reduced_dim // self.n_head)
mid_query = mid_query.transpose(1, 2)
mid_key = mid_key.transpose(1, 2)
mid_value = mid_value.transpose(1, 2)
hig_query = hig_query.reshape(b, hig_num, self.n_head, self.
reduced_dim // self.n_head)
hig_query = hig_query.transpose(1, 2)
mid_query, mid_low_weights = self.inner_attention(mid_query,
low_key, low_value)
hig_query, hig_mid_weights = self.inner_attention(hig_query,
mid_key, mid_value)
mid_query = mid_query.transpose(1, 2).reshape(b, mid_num, self.
reduced_dim)
hig_query = hig_query.transpose(1, 2).reshape(b, hig_num, self.
reduced_dim)
output = self.dropout(self.Wo(torch.cat((mid_query, hig_query), dim=1))
)
mid_vectors, hig_vectors = torch.split(output, [mid_num, hig_num],
dim=1)
mid_vectors = mid_residual + mid_vectors
hig_vectors = hig_residual + hig_vectors
mid_vectors = self.mid_ln(mid_vectors)
hig_vectors = self.hig_ln(hig_vectors)
return mid_vectors, hig_vectors, mid_low_weights, hig_mid_weights
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4]), torch.rand([4, 4, 4])
]
def get_init_inputs():
return [[], {'embedding_dim': 4, 'reduced_dim': 4, 'n_head': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 8
x0 = xindex % 4
x2 = xindex // 32
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 4 * x1 + 16 * x2), tmp4 & xmask, other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1], 8, tl.int64)
tmp9 = tl.load(in_ptr1 + (x0 + 4 * (-4 + x1) + 16 * x2), tmp6 & xmask,
other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + x3, tmp10, xmask)
@triton.jit
def triton_poi_fused_clone_1(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x2 + 32 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp3 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = tmp14 * tmp1
tmp16 = tl_math.exp(tmp15)
tl.store(out_ptr0 + x2, tmp16, xmask)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_clone_4(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x2 = xindex
y0 = yindex % 4
y1 = yindex // 4
y3 = yindex
tmp0 = tl.load(in_ptr0 + (16 + y0 + 4 * x2 + 32 * y1), xmask & ymask,
eviction_policy='evict_last')
tl.store(out_ptr0 + (x2 + 4 * y3), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_cat_5(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 8
x0 = xindex % 4
x2 = xindex // 32
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (4 * x0 + 16 * x2 + x1), tmp4 & xmask,
eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1], 8, tl.int64)
tmp9 = tl.load(in_ptr1 + (4 * x0 + 16 * x2 + (-4 + x1)), tmp6 & xmask,
eviction_policy='evict_last', other=0.0)
tmp10 = tl.where(tmp4, tmp5, tmp9)
tl.store(out_ptr0 + x3, tmp10, xmask)
@triton.jit
def triton_poi_fused_add_6(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = xindex // 16
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (x0 + 32 * x1), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_add_7(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 16
x1 = xindex // 16
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (16 + x0 + 32 * x1), xmask)
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_8(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-08
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x0, tmp8, xmask)
tl.store(out_ptr1 + x0, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_9(in_ptr0, in_ptr1, in_ptr2, in_ptr3,
in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (4,), (1,))
assert_size_stride(primals_9, (4,), (1,))
assert_size_stride(primals_10, (4,), (1,))
assert_size_stride(primals_11, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(128)](primals_2, primals_3, buf0, 128,
XBLOCK=128, num_warps=4, num_stages=1)
buf1 = empty_strided_cuda((32, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf0, (32, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((4, 8, 4), (32, 4, 1), torch.float32)
triton_poi_fused_cat_0[grid(128)](primals_1, primals_2, buf2, 128,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
buf3 = empty_strided_cuda((32, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf2, (32, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf3)
del primals_5
buf4 = empty_strided_cuda((32, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf2, (32, 4), (4, 1), 0),
reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf4)
del primals_6
buf5 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_1[grid(16, 4)](buf1, buf5, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf6 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
triton_poi_fused_clone_1[grid(16, 4)](buf3, buf6, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf7 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf5, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf6, (16, 1, 4), (4, 0, 1), 0), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_2[grid(256)](buf7, buf8, 256, XBLOCK=256,
num_warps=4, num_stages=1)
buf9 = reinterpret_tensor(buf7, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf7
triton_poi_fused__softmax_3[grid(256)](buf8, buf9, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf10 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_1[grid(16, 4)](buf4, buf10, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf11 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf9, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf10, (16, 4, 1), (4, 1, 0), 0), out=buf11)
buf12 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_4[grid(16, 4)](buf1, buf12, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
del buf1
buf13 = empty_strided_cuda((4, 4, 1, 4), (16, 4, 4, 1), torch.float32)
triton_poi_fused_clone_4[grid(16, 4)](buf3, buf13, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf14 = reinterpret_tensor(buf8, (16, 4, 4), (16, 4, 1), 0)
del buf8
extern_kernels.bmm(reinterpret_tensor(buf12, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf13, (16, 1, 4), (4, 0, 1), 0), out=buf14)
buf15 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_2[grid(256)](buf14, buf15, 256, XBLOCK=
256, num_warps=4, num_stages=1)
buf16 = reinterpret_tensor(buf14, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf14
triton_poi_fused__softmax_3[grid(256)](buf15, buf16, 256, XBLOCK=
128, num_warps=4, num_stages=1)
del buf15
buf17 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 1), torch.float32)
triton_poi_fused_clone_4[grid(16, 4)](buf4, buf17, 16, 4, XBLOCK=4,
YBLOCK=16, num_warps=1, num_stages=1)
buf18 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf16, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf17, (16, 4, 1), (4, 1, 0), 0), out=buf18)
buf19 = reinterpret_tensor(buf4, (4, 8, 4), (32, 4, 1), 0)
del buf4
triton_poi_fused_cat_5[grid(128)](buf11, buf18, buf19, 128, XBLOCK=
128, num_warps=4, num_stages=1)
buf20 = buf3
del buf3
extern_kernels.mm(reinterpret_tensor(buf19, (32, 4), (4, 1), 0),
reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf20)
buf21 = reinterpret_tensor(buf18, (4, 4, 4), (16, 4, 1), 0)
del buf18
triton_poi_fused_add_6[grid(64)](primals_2, buf20, buf21, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_2
buf22 = reinterpret_tensor(buf11, (4, 4, 4), (16, 4, 1), 0)
del buf11
triton_poi_fused_add_7[grid(64)](primals_3, buf20, buf22, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del buf20
del primals_3
buf23 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
buf24 = empty_strided_cuda((4, 4, 1), (4, 1, 16), torch.float32)
triton_poi_fused_native_layer_norm_8[grid(16)](buf21, buf23, buf24,
16, XBLOCK=16, num_warps=1, num_stages=1)
buf25 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_9[grid(64)](buf21, buf23, buf24,
primals_8, primals_9, buf25, 64, XBLOCK=64, num_warps=1,
num_stages=1)
del primals_9
buf26 = buf24
del buf24
buf27 = buf23
del buf23
triton_poi_fused_native_layer_norm_8[grid(16)](buf22, buf26, buf27,
16, XBLOCK=16, num_warps=1, num_stages=1)
buf28 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_9[grid(64)](buf22, buf26, buf27,
primals_10, primals_11, buf28, 64, XBLOCK=64, num_warps=1,
num_stages=1)
del buf26
del buf27
del primals_11
return (buf25, buf28, buf9, buf16, primals_8, primals_10,
reinterpret_tensor(buf0, (32, 4), (4, 1), 0), reinterpret_tensor(
buf2, (32, 4), (4, 1), 0), buf9, buf16, reinterpret_tensor(buf19, (
32, 4), (4, 1), 0), buf21, buf22, primals_7, reinterpret_tensor(
buf17, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf12, (16, 1,
4), (4, 1, 1), 0), reinterpret_tensor(buf13, (16, 4, 1), (4, 1, 4),
0), reinterpret_tensor(buf10, (16, 1, 4), (4, 1, 1), 0),
reinterpret_tensor(buf5, (16, 1, 4), (4, 1, 1), 0),
reinterpret_tensor(buf6, (16, 4, 1), (4, 1, 4), 0))
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super(ScaledDotProductAttention, self).__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(MultiHeadAttention, self).__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class ComprehensionLayer_step2New(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super(ComprehensionLayer_step2New, self).__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, input_0, input_1, input_2):
primals_4 = self.Wq.weight
primals_5 = self.Wk.weight
primals_6 = self.Wv.weight
primals_7 = self.Wo.weight
primals_8 = self.mid_ln.weight
primals_9 = self.mid_ln.bias
primals_10 = self.hig_ln.weight
primals_11 = self.hig_ln.bias
primals_1 = input_0
primals_2 = input_1
primals_3 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11])
return output[0], output[1], output[2], output[3]
| luyu-fan/LRCM | ComprehensionLayer_step2 | false | 7,172 | [
"MIT"
] | 1 | 6b0e4d7998bc4969afa764eb753077e3f858f1ba | https://github.com/luyu-fan/LRCM/tree/6b0e4d7998bc4969afa764eb753077e3f858f1ba | import math
import torch
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, dropout=0.0):
super().__init__()
self.dropout = nn.Dropout(dropout)
def forward(self, query, key, value):
assert query.size()[-1] == key.size()[-1]
dim = query.size()[-1]
tmp_raw_scores = torch.div(torch.matmul(query, key.transpose(-2, -1
)), math.sqrt(dim))
atte_weights = torch.softmax(tmp_raw_scores, dim=-1)
atte_weights = self.dropout(atte_weights)
output = torch.matmul(atte_weights, value)
return output, atte_weights
class MultiHeadAttention(nn.Module):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super().__init__()
assert reduced_dim % n_head == 0
self.n_head = n_head
self.embedding_dim = embedding_dim
self.reduced_dim = reduced_dim
self.Wq = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wk = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.Wv = nn.Linear(embedding_dim, reduced_dim, bias=False)
self.inner_attention = ScaledDotProductAttention(dropout)
self.Wo = nn.Linear(reduced_dim, embedding_dim, bias=False)
self.dropout = nn.Dropout(dropout)
self.ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, query):
residual = query
value = key = query
query = self.Wq(query)
key = self.Wk(key)
value = self.Wv(value)
b, n, _ = query.size()
query = query.reshape(b, n, self.n_head, self.reduced_dim // self.
n_head)
b, m, _ = key.size()
key = key.reshape(b, m, self.n_head, self.reduced_dim // self.n_head)
value = value.reshape(b, m, self.n_head, self.reduced_dim // self.
n_head)
query = query.transpose(1, 2)
key = key.transpose(1, 2)
value = value.transpose(1, 2)
query, atte_weights = self.inner_attention(query, key, value)
query = query.transpose(1, 2).reshape(b, n, self.reduced_dim)
query = self.dropout(self.Wo(query))
query = query + residual
query = self.ln(query)
return query, atte_weights
class Model(MultiHeadAttention):
def __init__(self, embedding_dim, reduced_dim, n_head, dropout=0.0, eps
=1e-08):
super().__init__(embedding_dim,
reduced_dim, n_head, dropout)
del self.ln
self.mid_ln = nn.LayerNorm(embedding_dim, eps=eps)
self.hig_ln = nn.LayerNorm(embedding_dim, eps=eps)
def forward(self, low_vectors, mid_vectors, hig_vectors):
b = low_vectors.size()[0]
low_num, mid_num, hig_num = low_vectors.size()[1], mid_vectors.size()[1
], hig_vectors.size()[1]
mid_residual = mid_vectors
hig_residual = hig_vectors
query = self.Wq(torch.cat((mid_vectors, hig_vectors), dim=1))
key = self.Wk(torch.cat((low_vectors, mid_vectors), dim=1))
value = self.Wv(torch.cat((low_vectors, mid_vectors), dim=1))
mid_query, hig_query = torch.split(query, [mid_num, hig_num], dim=1)
low_key, mid_key = torch.split(key, [low_num, mid_num], dim=1)
low_value, mid_value = torch.split(value, [low_num, mid_num], dim=1)
low_key = low_key.reshape(b, low_num, self.n_head, self.reduced_dim //
self.n_head)
low_value = low_value.reshape(b, low_num, self.n_head, self.
reduced_dim // self.n_head)
low_key = low_key.transpose(1, 2)
low_value = low_value.transpose(1, 2)
mid_query = mid_query.reshape(b, mid_num, self.n_head, self.
reduced_dim // self.n_head)
mid_key = mid_key.reshape(b, mid_num, self.n_head, self.reduced_dim //
self.n_head)
mid_value = mid_value.reshape(b, mid_num, self.n_head, self.
reduced_dim // self.n_head)
mid_query = mid_query.transpo
# ... truncated (>4000 chars) for memory efficiency |
ClassHead | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/u3/cu3litezfpnwhpnfnfuj6dtimz6ml42wmcwnwxlnovd4p5lvyin4.py
# Unsorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
triton_poi_fused_0 = async_compile.triton('triton_poi_fused_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048, 4096], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 2048
xnumel = 4096
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, YBLOCK], True, tl.int1)
x2 = xindex
y3 = yindex
y0 = yindex % 512
y1 = (yindex // 512)
tmp0 = tl.load(in_ptr0 + (x2 + (4096*y3)), None, eviction_policy='evict_last')
tl.store(out_ptr0 + (y0 + (512*x2) + (2097152*y1)), tmp0, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/tn/ctngadf5wjmxyiyxcthxpsc746ohbupgtib27hmksylwrnku36ja.py
# Topologically Sorted Source Nodes: [out_3], Original ATen: [aten._log_softmax]
# Source node to ATen node mapping:
# out_3 => amax, clone, sub
# Graph fragment:
# %clone : [num_users=2] = call_function[target=torch.ops.aten.clone.default](args = (%view,), kwargs = {memory_format: torch.contiguous_format})
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%clone, [-1], True), kwargs = {})
# %sub : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%clone, %amax), kwargs = {})
triton_poi_fused__log_softmax_1 = async_compile.triton('triton_poi_fused__log_softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[131072],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax_1(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 98304
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x4 = xindex % 6
x5 = (xindex // 2)
x1 = (xindex // 2) % 3
tmp0 = tl.load(in_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr1 + (x4), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2*x5), None, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (2*x1), None, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (1 + (2*x5)), None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (1 + (2*x1)), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp8 = tmp6 + tmp7
tmp9 = triton_helpers.maximum(tmp5, tmp8)
tmp10 = tmp2 - tmp9
tl.store(out_ptr0 + (x3), tmp10, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/2j/c2jyydmw7qz6h2ff5q2usux5vna5jexqqq5irspqybkrepnonpty.py
# Topologically Sorted Source Nodes: [out_3], Original ATen: [aten._log_softmax, aten._log_softmax_backward_data]
# Source node to ATen node mapping:
# out_3 => exp, log, sub_1, sum_1
# Graph fragment:
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%sum_1,), kwargs = {})
# %sub_1 : [num_users=2] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %log), kwargs = {})
# %exp_1 : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%sub_1,), kwargs = {})
triton_poi_fused__log_softmax__log_softmax_backward_data_2 = async_compile.triton('triton_poi_fused__log_softmax__log_softmax_backward_data_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[131072],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__log_softmax__log_softmax_backward_data_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__log_softmax__log_softmax_backward_data_2(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 98304
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x1 = (xindex // 2)
tmp0 = tl.load(in_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr0 + (2*x1), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (2*x1)), None, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp6 = tl_math.log(tmp5)
tmp7 = tmp0 - tmp6
tmp8 = tl_math.exp(tmp7)
tl.store(out_ptr0 + (x2), tmp7, None)
tl.store(out_ptr1 + (x2), tmp8, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (6, 512, 1, 1), (512, 1, 1, 1))
assert_size_stride(primals_2, (6, ), (1, ))
assert_size_stride(primals_3, (4, 512, 64, 64), (2097152, 4096, 64, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 512, 64, 64), (2097152, 1, 32768, 512), torch.float32)
# Unsorted Source Nodes: [], Original ATen: []
stream0 = get_raw_stream(0)
triton_poi_fused_0.run(primals_3, buf0, 2048, 4096, grid=grid(2048, 4096), stream=stream0)
del primals_3
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_1, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 6, 64, 64), (24576, 1, 384, 6))
buf2 = empty_strided_cuda((4, 64, 64, 3, 2), (24576, 384, 6, 2, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_3], Original ATen: [aten._log_softmax]
triton_poi_fused__log_softmax_1.run(buf1, primals_2, buf2, 98304, grid=grid(98304), stream=stream0)
del primals_2
buf3 = reinterpret_tensor(buf1, (4, 64, 64, 3, 2), (24576, 384, 6, 2, 1), 0); del buf1 # reuse
buf4 = empty_strided_cuda((4, 64, 64, 3, 2), (24576, 384, 6, 2, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_3], Original ATen: [aten._log_softmax, aten._log_softmax_backward_data]
triton_poi_fused__log_softmax__log_softmax_backward_data_2.run(buf2, buf3, buf4, 98304, grid=grid(98304), stream=stream0)
del buf2
return (reinterpret_tensor(buf3, (4, 12288, 2), (24576, 2, 1), 0), primals_1, buf0, buf4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((6, 512, 1, 1), (512, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((6, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 512, 64, 64), (2097152, 4096, 64, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class ClassHead(nn.Module):
def __init__(self, inchannels=512, num_anchors=3):
super(ClassHead, self).__init__()
self.num_anchors = num_anchors
self.conv1x1 = nn.Conv2d(inchannels, self.num_anchors * 2,
kernel_size=(1, 1), stride=1, padding=0)
self.output_act = nn.LogSoftmax(dim=-1)
def forward(self, x):
out = self.conv1x1(x)
out = out.permute(0, 2, 3, 1)
b, h, w, _c = out.shape
out = out.view(b, h, w, self.num_anchors, 2)
out = self.output_act(out)
return out.contiguous().view(out.shape[0], -1, 2)
def get_inputs():
return [torch.rand([4, 512, 64, 64])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK: tl.
constexpr, XBLOCK: tl.constexpr):
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, YBLOCK], True, tl.int1)
x2 = xindex
y3 = yindex
y0 = yindex % 512
y1 = yindex // 512
tmp0 = tl.load(in_ptr0 + (x2 + 4096 * y3), None, eviction_policy=
'evict_last')
tl.store(out_ptr0 + (y0 + 512 * x2 + 2097152 * y1), tmp0, None)
@triton.jit
def triton_poi_fused__log_softmax_1(in_ptr0, in_ptr1, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x4 = xindex % 6
x5 = xindex // 2
x1 = xindex // 2 % 3
tmp0 = tl.load(in_ptr0 + x3, None)
tmp1 = tl.load(in_ptr1 + x4, None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + 2 * x5, None, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + 2 * x1, None, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (1 + 2 * x5), None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (1 + 2 * x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp8 = tmp6 + tmp7
tmp9 = triton_helpers.maximum(tmp5, tmp8)
tmp10 = tmp2 - tmp9
tl.store(out_ptr0 + x3, tmp10, None)
@triton.jit
def triton_poi_fused__log_softmax__log_softmax_backward_data_2(in_ptr0,
out_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x1 = xindex // 2
tmp0 = tl.load(in_ptr0 + x2, None)
tmp1 = tl.load(in_ptr0 + 2 * x1, None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 2 * x1), None, eviction_policy='evict_last')
tmp2 = tl_math.exp(tmp1)
tmp4 = tl_math.exp(tmp3)
tmp5 = tmp2 + tmp4
tmp6 = tl_math.log(tmp5)
tmp7 = tmp0 - tmp6
tmp8 = tl_math.exp(tmp7)
tl.store(out_ptr0 + x2, tmp7, None)
tl.store(out_ptr1 + x2, tmp8, None)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (6, 512, 1, 1), (512, 1, 1, 1))
assert_size_stride(primals_2, (6,), (1,))
assert_size_stride(primals_3, (4, 512, 64, 64), (2097152, 4096, 64, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 512, 64, 64), (2097152, 1, 32768, 512
), torch.float32)
get_raw_stream(0)
triton_poi_fused_0[grid(2048, 4096)](primals_3, buf0, 2048, 4096,
XBLOCK=32, YBLOCK=32, num_warps=4, num_stages=1)
del primals_3
buf1 = extern_kernels.convolution(buf0, primals_1, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 6, 64, 64), (24576, 1, 384, 6))
buf2 = empty_strided_cuda((4, 64, 64, 3, 2), (24576, 384, 6, 2, 1),
torch.float32)
triton_poi_fused__log_softmax_1[grid(98304)](buf1, primals_2, buf2,
98304, XBLOCK=512, num_warps=8, num_stages=1)
del primals_2
buf3 = reinterpret_tensor(buf1, (4, 64, 64, 3, 2), (24576, 384, 6,
2, 1), 0)
del buf1
buf4 = empty_strided_cuda((4, 64, 64, 3, 2), (24576, 384, 6, 2, 1),
torch.float32)
triton_poi_fused__log_softmax__log_softmax_backward_data_2[grid(98304)
](buf2, buf3, buf4, 98304, XBLOCK=512, num_warps=8, num_stages=1)
del buf2
return reinterpret_tensor(buf3, (4, 12288, 2), (24576, 2, 1), 0
), primals_1, buf0, buf4
class ClassHeadNew(nn.Module):
def __init__(self, inchannels=512, num_anchors=3):
super(ClassHeadNew, self).__init__()
self.num_anchors = num_anchors
self.conv1x1 = nn.Conv2d(inchannels, self.num_anchors * 2,
kernel_size=(1, 1), stride=1, padding=0)
self.output_act = nn.LogSoftmax(dim=-1)
def forward(self, input_0):
primals_1 = self.conv1x1.weight
primals_2 = self.conv1x1.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| lurenjia307/RetinaPedestrian_Pytorch | ClassHead | false | 7,173 | [
"MIT"
] | 1 | 59c4aa50f3ef2ecb1113ad3b9950e8bbbff1206f | https://github.com/lurenjia307/RetinaPedestrian_Pytorch/tree/59c4aa50f3ef2ecb1113ad3b9950e8bbbff1206f | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self, inchannels=512, num_anchors=3):
super().__init__()
self.num_anchors = num_anchors
self.conv1x1 = nn.Conv2d(inchannels, self.num_anchors * 2,
kernel_size=(1, 1), stride=1, padding=0)
self.output_act = nn.LogSoftmax(dim=-1)
def forward(self, x):
out = self.conv1x1(x)
out = out.permute(0, 2, 3, 1)
b, h, w, _c = out.shape
out = out.view(b, h, w, self.num_anchors, 2)
out = self.output_act(out)
return out.contiguous().view(out.shape[0], -1, 2)
def get_inputs():
return [torch.rand([4, 512, 64, 64])]
def get_init_inputs():
return []
|
LaplacianPyramidLayer | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/gy/cgyfx47penjbxrnlqdpur6wznrc2npiddss2rmhvsk53kjjd4wdb.py
# Topologically Sorted Source Nodes: [down], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# down => cat
# Graph fragment:
# %cat : [num_users=2] = call_function[target=torch.ops.aten.cat.default](args = ([%convolution, %convolution_1, %convolution_2, %convolution_3], 1), kwargs = {})
triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x0 = xindex % 4
x2 = (xindex // 16)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (4*x2)), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + (4*x2)), tmp9 & xmask, eviction_policy='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x0 + (4*x2)), tmp14 & xmask, eviction_policy='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tmp17 = tl.full([1], 4, tl.int64)
tmp18 = tmp0 < tmp17
tmp19 = tl.load(in_ptr3 + (x0 + (4*x2)), tmp16 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + (x3), tmp22, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/un/cun4zkovw4qwlct3fch6tiatbs5f3sfjlujal64kew2sdgtghgii.py
# Topologically Sorted Source Nodes: [upsample], Original ATen: [aten._unsafe_index]
# Source node to ATen node mapping:
# upsample => _unsafe_index
# Graph fragment:
# %_unsafe_index : [num_users=4] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%cat, [None, None, %unsqueeze, %convert_element_type_3]), kwargs = {})
triton_poi_fused__unsafe_index_1 = async_compile.triton('triton_poi_fused__unsafe_index_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__unsafe_index_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__unsafe_index_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x0 = xindex % 4
x2 = (xindex // 16)
x4 = xindex
tmp0 = x1
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 * tmp2
tmp4 = tmp3.to(tl.int32)
tmp5 = x0
tmp6 = tmp5.to(tl.float32)
tmp7 = tmp6 * tmp2
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.load(in_ptr0 + (tmp8 + (2*tmp4) + (4*x2)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x4), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/vm/cvmkwryfo4cwodqac64c6pdxt5wkrnc62udspire6hpqqwyhfk4t.py
# Topologically Sorted Source Nodes: [remade, diff], Original ATen: [aten.cat, aten.sub]
# Source node to ATen node mapping:
# diff => sub
# remade => cat_1
# Graph fragment:
# %cat_1 : [num_users=2] = call_function[target=torch.ops.aten.cat.default](args = ([%convolution_4, %convolution_5, %convolution_6, %convolution_7], 1), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %cat_1), kwargs = {})
triton_poi_fused_cat_sub_2 = async_compile.triton('triton_poi_fused_cat_sub_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_sub_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_sub_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 16) % 4
x0 = xindex % 16
x2 = (xindex // 64)
x3 = xindex
tmp23 = tl.load(in_ptr4 + (x3), xmask)
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (16*x2)), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + (16*x2)), tmp9 & xmask, eviction_policy='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x0 + (16*x2)), tmp14 & xmask, eviction_policy='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tmp17 = tl.full([1], 4, tl.int64)
tmp18 = tmp0 < tmp17
tmp19 = tl.load(in_ptr3 + (x0 + (16*x2)), tmp16 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tmp24 = tmp23 - tmp22
tl.store(out_ptr0 + (x3), tmp22, xmask)
tl.store(out_ptr1 + (x3), tmp24, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (1, 1, 5, 5), (25, 25, 5, 1))
assert_size_stride(arg2_1, (1, 1, 5, 5), (25, 25, 5, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [conv2d], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 16, 4, 1), 0), arg1_1, stride=(2, 2), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 1, 2, 2), (4, 4, 2, 1))
# Topologically Sorted Source Nodes: [conv2d_1], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 16, 4, 1), 16), arg1_1, stride=(2, 2), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 1, 2, 2), (4, 4, 2, 1))
# Topologically Sorted Source Nodes: [conv2d_2], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 16, 4, 1), 32), arg1_1, stride=(2, 2), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 2, 2), (4, 4, 2, 1))
# Topologically Sorted Source Nodes: [conv2d_3], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1, 4, 4), (64, 16, 4, 1), 48), arg1_1, stride=(2, 2), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf3, (4, 1, 2, 2), (4, 4, 2, 1))
del arg1_1
buf4 = empty_strided_cuda((4, 4, 2, 2), (16, 4, 2, 1), torch.float32)
# Topologically Sorted Source Nodes: [down], Original ATen: [aten.cat]
stream0 = get_raw_stream(0)
triton_poi_fused_cat_0.run(buf0, buf1, buf2, buf3, buf4, 64, grid=grid(64), stream=stream0)
del buf0
del buf1
del buf2
del buf3
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [upsample], Original ATen: [aten._unsafe_index]
triton_poi_fused__unsafe_index_1.run(buf4, buf5, 256, grid=grid(256), stream=stream0)
# Topologically Sorted Source Nodes: [conv2d_4], Original ATen: [aten.convolution]
buf6 = extern_kernels.convolution(reinterpret_tensor(buf5, (4, 1, 4, 4), (64, 0, 4, 1), 0), arg2_1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf6, (4, 1, 4, 4), (16, 16, 4, 1))
# Topologically Sorted Source Nodes: [conv2d_5], Original ATen: [aten.convolution]
buf7 = extern_kernels.convolution(reinterpret_tensor(buf5, (4, 1, 4, 4), (64, 16, 4, 1), 16), arg2_1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf7, (4, 1, 4, 4), (16, 16, 4, 1))
# Topologically Sorted Source Nodes: [conv2d_6], Original ATen: [aten.convolution]
buf8 = extern_kernels.convolution(reinterpret_tensor(buf5, (4, 1, 4, 4), (64, 16, 4, 1), 32), arg2_1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf8, (4, 1, 4, 4), (16, 16, 4, 1))
# Topologically Sorted Source Nodes: [conv2d_7], Original ATen: [aten.convolution]
buf9 = extern_kernels.convolution(reinterpret_tensor(buf5, (4, 1, 4, 4), (64, 16, 4, 1), 48), arg2_1, stride=(1, 1), padding=(2, 2), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf9, (4, 1, 4, 4), (16, 16, 4, 1))
del arg2_1
buf10 = buf5; del buf5 # reuse
buf11 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [remade, diff], Original ATen: [aten.cat, aten.sub]
triton_poi_fused_cat_sub_2.run(buf6, buf7, buf8, buf9, arg0_1, buf10, buf11, 256, grid=grid(256), stream=stream0)
del arg0_1
del buf6
del buf7
del buf8
del buf9
return (buf11, buf4, buf10, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((1, 1, 5, 5), (25, 25, 5, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((1, 1, 5, 5), (25, 25, 5, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from typing import Tuple
import torch.nn as nn
from torch.nn import functional as F
class PyramidDown(nn.Module):
def __init__(self) ->None:
super(PyramidDown, self).__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(x[:, i:i + 1, :, :], self.filter,
padding=2, stride=2))
return torch.cat(results, dim=1)
class PyramidUp(nn.Module):
def __init__(self) ->None:
super(PyramidUp, self).__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
upsample = F.interpolate(x, scale_factor=2)
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(upsample[:, i:i + 1, :, :], self.filter,
padding=2))
return torch.cat(results, dim=1)
class LaplacianPyramidLayer(nn.Module):
def __init__(self) ->None:
super(LaplacianPyramidLayer, self).__init__()
self.pyramid_down = PyramidDown()
self.pyramid_up = PyramidUp()
def forward(self, x: 'torch.Tensor') ->Tuple[torch.Tensor, torch.Tensor,
torch.Tensor]:
y = x
if x.shape[-1] % 2 != 0:
y = torch.cat([y, torch.zeros(y.shape[:-1]).unsqueeze(dim=-1)],
dim=-1)
if x.shape[-2] % 2 != 0:
y = y.transpose(-1, -2)
y = torch.cat([y, torch.zeros(y.shape[:-1]).unsqueeze(dim=-1)],
dim=-1)
y = y.transpose(-1, -2)
down: 'torch.Tensor' = self.pyramid_down(y)
remade: 'torch.Tensor' = self.pyramid_up(down)
diff: 'torch.Tensor' = y - remade
if x.shape[-1] % 2 != 0:
diff = diff[:, :, :, :-1]
if x.shape[-1] % 2 != 0:
diff = diff[:, :, :-1, :]
return diff, down, remade
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
from torch.nn import functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_cat_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x0 = xindex % 4
x2 = xindex // 16
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 4 * x2), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + 4 * x2), tmp9 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x0 + 4 * x2), tmp14 & xmask, eviction_policy
='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tl.full([1], 4, tl.int64)
tmp19 = tl.load(in_ptr3 + (x0 + 4 * x2), tmp16 & xmask, eviction_policy
='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tl.store(out_ptr0 + x3, tmp22, xmask)
@triton.jit
def triton_poi_fused__unsafe_index_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x0 = xindex % 4
x2 = xindex // 16
x4 = xindex
tmp0 = x1
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 * tmp2
tmp4 = tmp3.to(tl.int32)
tmp5 = x0
tmp6 = tmp5.to(tl.float32)
tmp7 = tmp6 * tmp2
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.load(in_ptr0 + (tmp8 + 2 * tmp4 + 4 * x2), xmask,
eviction_policy='evict_last')
tl.store(out_ptr0 + x4, tmp9, xmask)
@triton.jit
def triton_poi_fused_cat_sub_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4,
out_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 16 % 4
x0 = xindex % 16
x2 = xindex // 64
x3 = xindex
tmp23 = tl.load(in_ptr4 + x3, xmask)
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 16 * x2), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 2, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tmp6 & tmp8
tmp10 = tl.load(in_ptr1 + (x0 + 16 * x2), tmp9 & xmask, eviction_policy
='evict_last', other=0.0)
tmp11 = tmp0 >= tmp7
tmp12 = tl.full([1], 3, tl.int64)
tmp13 = tmp0 < tmp12
tmp14 = tmp11 & tmp13
tmp15 = tl.load(in_ptr2 + (x0 + 16 * x2), tmp14 & xmask,
eviction_policy='evict_last', other=0.0)
tmp16 = tmp0 >= tmp12
tl.full([1], 4, tl.int64)
tmp19 = tl.load(in_ptr3 + (x0 + 16 * x2), tmp16 & xmask,
eviction_policy='evict_last', other=0.0)
tmp20 = tl.where(tmp14, tmp15, tmp19)
tmp21 = tl.where(tmp9, tmp10, tmp20)
tmp22 = tl.where(tmp4, tmp5, tmp21)
tmp24 = tmp23 - tmp22
tl.store(out_ptr0 + x3, tmp22, xmask)
tl.store(out_ptr1 + x3, tmp24, xmask)
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (1, 1, 5, 5), (25, 25, 5, 1))
assert_size_stride(arg2_1, (1, 1, 5, 5), (25, 25, 5, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 16, 4, 1), 0), arg1_1, stride=(2, 2), padding=(2, 2
), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf0, (4, 1, 2, 2), (4, 4, 2, 1))
buf1 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 16, 4, 1), 16), arg1_1, stride=(2, 2), padding=(2,
2), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf1, (4, 1, 2, 2), (4, 4, 2, 1))
buf2 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 16, 4, 1), 32), arg1_1, stride=(2, 2), padding=(2,
2), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf2, (4, 1, 2, 2), (4, 4, 2, 1))
buf3 = extern_kernels.convolution(reinterpret_tensor(arg0_1, (4, 1,
4, 4), (64, 16, 4, 1), 48), arg1_1, stride=(2, 2), padding=(2,
2), dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf3, (4, 1, 2, 2), (4, 4, 2, 1))
del arg1_1
buf4 = empty_strided_cuda((4, 4, 2, 2), (16, 4, 2, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_cat_0[grid(64)](buf0, buf1, buf2, buf3, buf4, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del buf0
del buf1
del buf2
del buf3
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__unsafe_index_1[grid(256)](buf4, buf5, 256, XBLOCK
=128, num_warps=4, num_stages=1)
buf6 = extern_kernels.convolution(reinterpret_tensor(buf5, (4, 1, 4,
4), (64, 0, 4, 1), 0), arg2_1, stride=(1, 1), padding=(2, 2),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf6, (4, 1, 4, 4), (16, 16, 4, 1))
buf7 = extern_kernels.convolution(reinterpret_tensor(buf5, (4, 1, 4,
4), (64, 16, 4, 1), 16), arg2_1, stride=(1, 1), padding=(2, 2),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf7, (4, 1, 4, 4), (16, 16, 4, 1))
buf8 = extern_kernels.convolution(reinterpret_tensor(buf5, (4, 1, 4,
4), (64, 16, 4, 1), 32), arg2_1, stride=(1, 1), padding=(2, 2),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf8, (4, 1, 4, 4), (16, 16, 4, 1))
buf9 = extern_kernels.convolution(reinterpret_tensor(buf5, (4, 1, 4,
4), (64, 16, 4, 1), 48), arg2_1, stride=(1, 1), padding=(2, 2),
dilation=(1, 1), transposed=False, output_padding=(0, 0),
groups=1, bias=None)
assert_size_stride(buf9, (4, 1, 4, 4), (16, 16, 4, 1))
del arg2_1
buf10 = buf5
del buf5
buf11 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_cat_sub_2[grid(256)](buf6, buf7, buf8, buf9,
arg0_1, buf10, buf11, 256, XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
del buf6
del buf7
del buf8
del buf9
return buf11, buf4, buf10
class PyramidDown(nn.Module):
def __init__(self) ->None:
super(PyramidDown, self).__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(x[:, i:i + 1, :, :], self.filter,
padding=2, stride=2))
return torch.cat(results, dim=1)
class PyramidUp(nn.Module):
def __init__(self) ->None:
super(PyramidUp, self).__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
upsample = F.interpolate(x, scale_factor=2)
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(upsample[:, i:i + 1, :, :], self.filter,
padding=2))
return torch.cat(results, dim=1)
class LaplacianPyramidLayerNew(nn.Module):
def __init__(self) ->None:
super(LaplacianPyramidLayerNew, self).__init__()
self.pyramid_down = PyramidDown()
self.pyramid_up = PyramidUp()
def forward(self, input_0):
arg1_1 = self.pyramid_down.filter
arg2_1 = self.pyramid_up.filter
arg0_1 = input_0
output = call([arg0_1, arg1_1, arg2_1])
return output[0], output[1], output[2]
| masanorihirano/pytorch_extra_mhirano | LaplacianPyramidLayer | false | 7,174 | [
"MIT"
] | 1 | d19e07445567c069793b7ca1a22a846d7cbce58d | https://github.com/masanorihirano/pytorch_extra_mhirano/tree/d19e07445567c069793b7ca1a22a846d7cbce58d | import torch
from typing import Tuple
import torch.nn as nn
from torch.nn import functional as F
class PyramidDown(nn.Module):
def __init__(self) ->None:
super().__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(x[:, i:i + 1, :, :], self.filter,
padding=2, stride=2))
return torch.cat(results, dim=1)
class PyramidUp(nn.Module):
def __init__(self) ->None:
super().__init__()
self.filter = nn.Parameter(torch.tensor([[1, 4, 6, 4, 1], [4, 16,
24, 16, 4], [6, 24, 36, 24, 6], [4, 16, 24, 16, 4], [1, 4, 6, 4,
1]], dtype=torch.float).reshape(1, 1, 5, 5) / 256,
requires_grad=False)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
upsample = F.interpolate(x, scale_factor=2)
results = []
for i in range(x.shape[1]):
results.append(F.conv2d(upsample[:, i:i + 1, :, :], self.filter,
padding=2))
return torch.cat(results, dim=1)
class Model(nn.Module):
def __init__(self) ->None:
super().__init__()
self.pyramid_down = PyramidDown()
self.pyramid_up = PyramidUp()
def forward(self, x: 'torch.Tensor') ->Tuple[torch.Tensor, torch.Tensor,
torch.Tensor]:
y = x
if x.shape[-1] % 2 != 0:
y = torch.cat([y, torch.zeros(y.shape[:-1]).unsqueeze(dim=-1)],
dim=-1)
if x.shape[-2] % 2 != 0:
y = y.transpose(-1, -2)
y = torch.cat([y, torch.zeros(y.shape[:-1]).unsqueeze(dim=-1)],
dim=-1)
y = y.transpose(-1, -2)
down: 'torch.Tensor' = self.pyramid_down(y)
remade: 'torch.Tensor' = self.pyramid_up(down)
diff: 'torch.Tensor' = y - remade
if x.shape[-1] % 2 != 0:
diff = diff[:, :, :, :-1]
if x.shape[-1] % 2 != 0:
diff = diff[:, :, :-1, :]
return diff, down, remade
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
ActorNet | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/xs/cxsx2jlz7qi3eerscrmq6daj5pyplc3rcuj6cilxq2zofbhntrow.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x_1 => gt, mul, where
# Graph fragment:
# %gt : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_3, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_3, 0.01), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %view_3, %mul), kwargs = {})
triton_poi_fused_leaky_relu_0 = async_compile.triton('triton_poi_fused_leaky_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 2560
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 40
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ui/cui6sxfzttgecmgtvnw2gevr54jsmp7y3t6fwxqjhor3rpwdmuui.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x_2 => gt_1, mul_1, where_1
# Graph fragment:
# %gt_1 : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_5, 0), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_5, 0.01), kwargs = {})
# %where_1 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %view_5, %mul_1), kwargs = {})
triton_poi_fused_leaky_relu_1 = async_compile.triton('triton_poi_fused_leaky_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 3200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 50
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/2o/c2osggm77pqxxx5r32gyjemu3ryhpfi3fx5zzrm7y43ln3qjxt2k.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x_3 => gt_2, mul_2, where_2
# Graph fragment:
# %gt_2 : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_7, 0), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_7, 0.01), kwargs = {})
# %where_2 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_2, %view_7, %mul_2), kwargs = {})
triton_poi_fused_leaky_relu_2 = async_compile.triton('triton_poi_fused_leaky_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_2(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 1920
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 30
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/gc/cgcgqaqngs7z7n2pbstvkxhguyvvx7auk4jaz3wlw25y3neeo5vn.py
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x_4 => gt_3, mul_3, where_3
# Graph fragment:
# %gt_3 : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_9, 0), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_9, 0.01), kwargs = {})
# %where_3 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_3, %view_9, %mul_3), kwargs = {})
triton_poi_fused_leaky_relu_3 = async_compile.triton('triton_poi_fused_leaky_relu_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_3(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 768
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 12
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13 = args
args.clear()
assert_size_stride(primals_1, (20, 4), (4, 1))
assert_size_stride(primals_2, (20, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (40, 20), (20, 1))
assert_size_stride(primals_5, (40, ), (1, ))
assert_size_stride(primals_6, (50, 40), (40, 1))
assert_size_stride(primals_7, (50, ), (1, ))
assert_size_stride(primals_8, (30, 50), (50, 1))
assert_size_stride(primals_9, (30, ), (1, ))
assert_size_stride(primals_10, (12, 30), (30, 1))
assert_size_stride(primals_11, (12, ), (1, ))
assert_size_stride(primals_12, (2, 12), (12, 1))
assert_size_stride(primals_13, (2, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 20), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 40), (40, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf0, reinterpret_tensor(primals_4, (20, 40), (1, 20), 0), out=buf1)
buf2 = empty_strided_cuda((4, 4, 4, 40), (640, 160, 40, 1), torch.bool)
buf3 = empty_strided_cuda((4, 4, 4, 40), (640, 160, 40, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_0.run(buf1, primals_5, buf2, buf3, 2560, grid=grid(2560), stream=stream0)
del buf1
del primals_5
buf4 = empty_strided_cuda((64, 50), (50, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf3, (64, 40), (40, 1), 0), reinterpret_tensor(primals_6, (40, 50), (1, 40), 0), out=buf4)
buf5 = empty_strided_cuda((4, 4, 4, 50), (800, 200, 50, 1), torch.bool)
buf6 = empty_strided_cuda((4, 4, 4, 50), (800, 200, 50, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_1.run(buf4, primals_7, buf5, buf6, 3200, grid=grid(3200), stream=stream0)
del buf4
del primals_7
buf7 = empty_strided_cuda((64, 30), (30, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf6, (64, 50), (50, 1), 0), reinterpret_tensor(primals_8, (50, 30), (1, 50), 0), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4, 30), (480, 120, 30, 1), torch.bool)
buf9 = empty_strided_cuda((4, 4, 4, 30), (480, 120, 30, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_2.run(buf7, primals_9, buf8, buf9, 1920, grid=grid(1920), stream=stream0)
del buf7
del primals_9
buf10 = empty_strided_cuda((64, 12), (12, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf9, (64, 30), (30, 1), 0), reinterpret_tensor(primals_10, (30, 12), (1, 30), 0), out=buf10)
buf11 = empty_strided_cuda((4, 4, 4, 12), (192, 48, 12, 1), torch.bool)
buf12 = empty_strided_cuda((4, 4, 4, 12), (192, 48, 12, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_3.run(buf10, primals_11, buf11, buf12, 768, grid=grid(768), stream=stream0)
del buf10
del primals_11
buf13 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_13, reinterpret_tensor(buf12, (64, 12), (12, 1), 0), reinterpret_tensor(primals_12, (12, 2), (1, 12), 0), alpha=1, beta=1, out=buf13)
del primals_13
return (reinterpret_tensor(buf13, (4, 4, 4, 2), (32, 8, 2, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf0, buf2, reinterpret_tensor(buf3, (64, 40), (40, 1), 0), buf5, reinterpret_tensor(buf6, (64, 50), (50, 1), 0), buf8, reinterpret_tensor(buf9, (64, 30), (30, 1), 0), buf11, reinterpret_tensor(buf12, (64, 12), (12, 1), 0), primals_12, primals_10, primals_8, primals_6, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((20, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((20, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((40, 20), (20, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((40, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((50, 40), (40, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((50, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((30, 50), (50, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((30, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((12, 30), (30, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((12, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((2, 12), (12, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class ActorNet(nn.Module):
def __init__(self):
super(ActorNet, self).__init__()
self.fc1 = nn.Linear(4, 20)
self.fc2 = nn.Linear(20, 40)
self.fc3 = nn.Linear(40, 50)
self.fc4 = nn.Linear(50, 30)
self.fc5 = nn.Linear(30, 12)
self.fc6 = nn.Linear(12, 2)
def forward(self, x):
x = self.fc1(x)
x = F.leaky_relu(self.fc2(x))
x = F.leaky_relu(self.fc3(x))
x = F.leaky_relu(self.fc4(x))
x = F.leaky_relu(self.fc5(x))
x = self.fc6(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 2560
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 40
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
@triton.jit
def triton_poi_fused_leaky_relu_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 3200
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 50
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
@triton.jit
def triton_poi_fused_leaky_relu_2(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 1920
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 30
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
@triton.jit
def triton_poi_fused_leaky_relu_3(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 768
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 12
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13) = args
args.clear()
assert_size_stride(primals_1, (20, 4), (4, 1))
assert_size_stride(primals_2, (20,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (40, 20), (20, 1))
assert_size_stride(primals_5, (40,), (1,))
assert_size_stride(primals_6, (50, 40), (40, 1))
assert_size_stride(primals_7, (50,), (1,))
assert_size_stride(primals_8, (30, 50), (50, 1))
assert_size_stride(primals_9, (30,), (1,))
assert_size_stride(primals_10, (12, 30), (30, 1))
assert_size_stride(primals_11, (12,), (1,))
assert_size_stride(primals_12, (2, 12), (12, 1))
assert_size_stride(primals_13, (2,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 20), (1, 4),
0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 40), (40, 1), torch.float32)
extern_kernels.mm(buf0, reinterpret_tensor(primals_4, (20, 40), (1,
20), 0), out=buf1)
buf2 = empty_strided_cuda((4, 4, 4, 40), (640, 160, 40, 1), torch.bool)
buf3 = empty_strided_cuda((4, 4, 4, 40), (640, 160, 40, 1), torch.
float32)
get_raw_stream(0)
triton_poi_fused_leaky_relu_0[grid(2560)](buf1, primals_5, buf2,
buf3, 2560, XBLOCK=128, num_warps=4, num_stages=1)
del buf1
del primals_5
buf4 = empty_strided_cuda((64, 50), (50, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf3, (64, 40), (40, 1), 0),
reinterpret_tensor(primals_6, (40, 50), (1, 40), 0), out=buf4)
buf5 = empty_strided_cuda((4, 4, 4, 50), (800, 200, 50, 1), torch.bool)
buf6 = empty_strided_cuda((4, 4, 4, 50), (800, 200, 50, 1), torch.
float32)
triton_poi_fused_leaky_relu_1[grid(3200)](buf4, primals_7, buf5,
buf6, 3200, XBLOCK=256, num_warps=4, num_stages=1)
del buf4
del primals_7
buf7 = empty_strided_cuda((64, 30), (30, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf6, (64, 50), (50, 1), 0),
reinterpret_tensor(primals_8, (50, 30), (1, 50), 0), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4, 30), (480, 120, 30, 1), torch.bool)
buf9 = empty_strided_cuda((4, 4, 4, 30), (480, 120, 30, 1), torch.
float32)
triton_poi_fused_leaky_relu_2[grid(1920)](buf7, primals_9, buf8,
buf9, 1920, XBLOCK=128, num_warps=4, num_stages=1)
del buf7
del primals_9
buf10 = empty_strided_cuda((64, 12), (12, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf9, (64, 30), (30, 1), 0),
reinterpret_tensor(primals_10, (30, 12), (1, 30), 0), out=buf10)
buf11 = empty_strided_cuda((4, 4, 4, 12), (192, 48, 12, 1), torch.bool)
buf12 = empty_strided_cuda((4, 4, 4, 12), (192, 48, 12, 1), torch.
float32)
triton_poi_fused_leaky_relu_3[grid(768)](buf10, primals_11, buf11,
buf12, 768, XBLOCK=128, num_warps=4, num_stages=1)
del buf10
del primals_11
buf13 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
extern_kernels.addmm(primals_13, reinterpret_tensor(buf12, (64, 12),
(12, 1), 0), reinterpret_tensor(primals_12, (12, 2), (1, 12), 0
), alpha=1, beta=1, out=buf13)
del primals_13
return reinterpret_tensor(buf13, (4, 4, 4, 2), (32, 8, 2, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf0, buf2, reinterpret_tensor(buf3, (64, 40), (40, 1), 0
), buf5, reinterpret_tensor(buf6, (64, 50), (50, 1), 0
), buf8, reinterpret_tensor(buf9, (64, 30), (30, 1), 0
), buf11, reinterpret_tensor(buf12, (64, 12), (12, 1), 0
), primals_12, primals_10, primals_8, primals_6, primals_4
class ActorNetNew(nn.Module):
def __init__(self):
super(ActorNetNew, self).__init__()
self.fc1 = nn.Linear(4, 20)
self.fc2 = nn.Linear(20, 40)
self.fc3 = nn.Linear(40, 50)
self.fc4 = nn.Linear(50, 30)
self.fc5 = nn.Linear(30, 12)
self.fc6 = nn.Linear(12, 2)
def forward(self, input_0):
primals_1 = self.fc1.weight
primals_2 = self.fc1.bias
primals_4 = self.fc2.weight
primals_5 = self.fc2.bias
primals_6 = self.fc3.weight
primals_7 = self.fc3.bias
primals_8 = self.fc4.weight
primals_9 = self.fc4.bias
primals_10 = self.fc5.weight
primals_11 = self.fc5.bias
primals_12 = self.fc6.weight
primals_13 = self.fc6.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13])
return output[0]
| mathildebadoual/RL_power_systems | ActorNet | false | 7,175 | [
"MIT"
] | 1 | 825e60bad16129e0a0229d15af5110b26e0a1577 | https://github.com/mathildebadoual/RL_power_systems/tree/825e60bad16129e0a0229d15af5110b26e0a1577 | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self):
super().__init__()
self.fc1 = nn.Linear(4, 20)
self.fc2 = nn.Linear(20, 40)
self.fc3 = nn.Linear(40, 50)
self.fc4 = nn.Linear(50, 30)
self.fc5 = nn.Linear(30, 12)
self.fc6 = nn.Linear(12, 2)
def forward(self, x):
x = self.fc1(x)
x = F.leaky_relu(self.fc2(x))
x = F.leaky_relu(self.fc3(x))
x = F.leaky_relu(self.fc4(x))
x = F.leaky_relu(self.fc5(x))
x = self.fc6(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
MyKernelTorch | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/yb/cybsjfmgf75kwyq3kyez46wzwjgjffwtsqe2uwa7bdzwlb6l22gt.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1280
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 20
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (20, 4), (4, 1))
assert_size_stride(primals_2, (20, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (2, 20), (20, 1))
assert_size_stride(primals_5, (2, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 20), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 20), (320, 80, 20, 1), 0); del buf0 # reuse
buf3 = empty_strided_cuda((4, 4, 4, 20), (320, 80, 20, 1), torch.bool)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_2, buf3, 1280, grid=grid(1280), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 20), (20, 1), 0), reinterpret_tensor(primals_4, (20, 2), (1, 20), 0), alpha=1, beta=1, out=buf2)
del primals_5
return (reinterpret_tensor(buf2, (4, 4, 4, 2), (32, 8, 2, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf1, (64, 20), (20, 1), 0), primals_4, buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((20, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((20, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((2, 20), (20, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class MyKernelTorch(nn.Module):
def __init__(self, n_features: 'int'):
super().__init__()
self.dense1 = nn.Linear(n_features, 20)
self.dense2 = nn.Linear(20, 2)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
x = nn.ReLU()(self.dense1(x))
return self.dense2(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'n_features': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 1280
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 20
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (20, 4), (4, 1))
assert_size_stride(primals_2, (20,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (2, 20), (20, 1))
assert_size_stride(primals_5, (2,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 20), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 20), (320, 80, 20, 1), 0)
del buf0
buf3 = empty_strided_cuda((4, 4, 4, 20), (320, 80, 20, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(1280)](buf1,
primals_2, buf3, 1280, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 20),
(20, 1), 0), reinterpret_tensor(primals_4, (20, 2), (1, 20), 0),
alpha=1, beta=1, out=buf2)
del primals_5
return reinterpret_tensor(buf2, (4, 4, 4, 2), (32, 8, 2, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), reinterpret_tensor(buf1, (64, 20), (20, 1), 0), primals_4, buf3
class MyKernelTorchNew(nn.Module):
def __init__(self, n_features: 'int'):
super().__init__()
self.dense1 = nn.Linear(n_features, 20)
self.dense2 = nn.Linear(20, 2)
def forward(self, input_0):
primals_1 = self.dense1.weight
primals_2 = self.dense1.bias
primals_4 = self.dense2.weight
primals_5 = self.dense2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| maxpark/alibi-detect | MyKernelTorch | false | 7,176 | [
"Apache-2.0"
] | 1 | 84384297a85764c18537aa1c8699c4ad040cf7cd | https://github.com/maxpark/alibi-detect/tree/84384297a85764c18537aa1c8699c4ad040cf7cd | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self, n_features: 'int'):
super().__init__()
self.dense1 = nn.Linear(n_features, 20)
self.dense2 = nn.Linear(20, 2)
def forward(self, x: 'torch.Tensor') ->torch.Tensor:
x = nn.ReLU()(self.dense1(x))
return self.dense2(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
ResidualConnection | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/n6/cn6mgjcngocfze5qgo2hju7qfnotg6mtpbuctnnlrhye2luxouz5.py
# Topologically Sorted Source Nodes: [add, truediv], Original ATen: [aten.add, aten.div]
# Source node to ATen node mapping:
# add => add
# truediv => div
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, %arg0_1), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%add, 2.0), kwargs = {})
triton_poi_fused_add_div_0 = async_compile.triton('triton_poi_fused_add_div_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tmp0 + tmp0
tmp2 = 0.5
tmp3 = tmp1 * tmp2
tl.store(out_ptr0 + (x0), tmp3, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [add, truediv], Original ATen: [aten.add, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class ResidualConnection(nn.Module):
def __init__(self, *layers):
super(ResidualConnection, self).__init__()
self.layers = nn.Sequential(*layers)
def forward(self, input):
return (input + self.layers(input)) / 2.0
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tmp0 + tmp0
tmp2 = 0.5
tmp3 = tmp1 * tmp2
tl.store(out_ptr0 + x0, tmp3, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_0[grid(256)](arg0_1, buf0, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del arg0_1
return buf0,
class ResidualConnectionNew(nn.Module):
def __init__(self, *layers):
super(ResidualConnectionNew, self).__init__()
self.layers = nn.Sequential(*layers)
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| maxkvant/LinearizedNNs | ResidualConnection | false | 7,177 | [
"Apache-2.0"
] | 1 | eb0198be70ca55e7463b97a5023d2f6ffe0f8ba6 | https://github.com/maxkvant/LinearizedNNs/tree/eb0198be70ca55e7463b97a5023d2f6ffe0f8ba6 | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self, *layers):
super().__init__()
self.layers = nn.Sequential(*layers)
def forward(self, input):
return (input + self.layers(input)) / 2.0
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
NormalizeImages | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/vk/cvktkyqkbrs4hp44g2v5fsjqeerqa54umgdv2pbxazbcvuw73y73.py
# Topologically Sorted Source Nodes: [mp, sub, std, truediv], Original ATen: [aten.mean, aten.sub, aten.std, aten.div]
# Source node to ATen node mapping:
# mp => mean
# std => var
# sub => sub
# truediv => div
# Graph fragment:
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%view, [1]), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %expand), kwargs = {})
# %var : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%view, [1]), kwargs = {correction: 1.0})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub, %expand_1), kwargs = {})
triton_per_fused_div_mean_std_sub_0 = async_compile.triton('triton_per_fused_div_mean_std_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[4, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_div_mean_std_sub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 4, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_div_mean_std_sub_0(in_ptr0, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 4
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (64*x0)), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 64, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 64.0
tmp20 = tmp4 / tmp19
tmp21 = tmp0 - tmp20
tmp22 = 63.0
tmp23 = tmp18 / tmp22
tmp24 = libdevice.sqrt(tmp23)
tmp25 = 1e-07
tmp26 = tmp24 + tmp25
tmp27 = tmp21 / tmp26
tl.store(out_ptr2 + (r1 + (64*x0)), tmp27, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mp, sub, std, truediv], Original ATen: [aten.mean, aten.sub, aten.std, aten.div]
stream0 = get_raw_stream(0)
triton_per_fused_div_mean_std_sub_0.run(arg0_1, buf4, 4, 64, grid=grid(4), stream=stream0)
del arg0_1
return (buf4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class NormalizeImages(nn.Module):
def __init__(self):
super().__init__()
def forward(self, x):
flat = x.view(x.size(0), -1)
mp = torch.mean(flat, dim=1)
sp = torch.std(flat, dim=1) + 1e-07
return (x - mp.detach().unsqueeze(-1).unsqueeze(-1).unsqueeze(-1).
expand_as(x)) / sp.detach().unsqueeze(-1).unsqueeze(-1).unsqueeze(1
).expand_as(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_div_mean_std_sub_0(in_ptr0, out_ptr2, xnumel, rnumel,
XBLOCK: tl.constexpr):
xnumel = 4
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 64 * x0), xmask, other=0.0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp6 = tl.broadcast_to(tmp1, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = tl.full([XBLOCK, 1], 64, tl.int32)
tmp11 = tmp10.to(tl.float32)
tmp12 = tmp9 / tmp11
tmp13 = tmp1 - tmp12
tmp14 = tmp13 * tmp13
tmp15 = tl.broadcast_to(tmp14, [XBLOCK, RBLOCK])
tmp17 = tl.where(xmask, tmp15, 0)
tmp18 = tl.sum(tmp17, 1)[:, None]
tmp19 = 64.0
tmp20 = tmp4 / tmp19
tmp21 = tmp0 - tmp20
tmp22 = 63.0
tmp23 = tmp18 / tmp22
tmp24 = libdevice.sqrt(tmp23)
tmp25 = 1e-07
tmp26 = tmp24 + tmp25
tmp27 = tmp21 / tmp26
tl.store(out_ptr2 + (r1 + 64 * x0), tmp27, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_per_fused_div_mean_std_sub_0[grid(4)](arg0_1, buf4, 4, 64,
XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
return buf4,
class NormalizeImagesNew(nn.Module):
def __init__(self):
super().__init__()
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| matteo-ronchetti/IKA | NormalizeImages | false | 7,178 | [
"MIT"
] | 1 | 29d1752a059c3ab7659b332b72bf8c1506e7dd20 | https://github.com/matteo-ronchetti/IKA/tree/29d1752a059c3ab7659b332b72bf8c1506e7dd20 | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self):
super().__init__()
def forward(self, x):
flat = x.view(x.size(0), -1)
mp = torch.mean(flat, dim=1)
sp = torch.std(flat, dim=1) + 1e-07
return (x - mp.detach().unsqueeze(-1).unsqueeze(-1).unsqueeze(-1).
expand_as(x)) / sp.detach().unsqueeze(-1).unsqueeze(-1).unsqueeze(1
).expand_as(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
SoftmaxAttention | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/b3/cb35lfgojdtaj2fnq2zxtf4ul7ez7i5xdvfuqztkpfoiqg5brilt.py
# Topologically Sorted Source Nodes: [contiguous], Original ATen: [aten.clone]
# Source node to ATen node mapping:
# contiguous => clone
# Graph fragment:
# %clone : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%permute,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_clone_0 = async_compile.triton('triton_poi_fused_clone_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tl.store(out_ptr0 + (x0), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/6v/c6v5qkna4b7ktybiouwzycows6tqpfi3a57j64smwq7gu44oewoh.py
# Topologically Sorted Source Nodes: [mul, result, result_1, sum_1], Original ATen: [aten.mul, aten._softmax, aten.sum]
# Source node to ATen node mapping:
# mul => mul
# result => amax, div, exp, sub, sum_1
# result_1 => mul_1
# sum_1 => sum_2
# Graph fragment:
# %mul : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view, %view_1), kwargs = {})
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul, [-1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
# %mul_1 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div, %view_1), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul_1, [-1], True), kwargs = {})
triton_poi_fused__softmax_mul_sum_1 = async_compile.triton('triton_poi_fused__softmax_mul_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_mul_sum_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_mul_sum_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4*(x0 // 4)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*(x0 // 4))), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + (4*(x0 // 4))), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (3 + (4*(x0 // 4))), xmask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp5 = tmp3 * tmp4
tmp6 = triton_helpers.maximum(tmp2, tmp5)
tmp9 = tmp7 * tmp8
tmp10 = triton_helpers.maximum(tmp6, tmp9)
tmp13 = tmp11 * tmp12
tmp14 = triton_helpers.maximum(tmp10, tmp13)
tmp15 = tmp2 - tmp14
tmp16 = tl_math.exp(tmp15)
tmp17 = tmp5 - tmp14
tmp18 = tl_math.exp(tmp17)
tmp19 = tmp16 + tmp18
tmp20 = tmp9 - tmp14
tmp21 = tl_math.exp(tmp20)
tmp22 = tmp19 + tmp21
tmp23 = tmp13 - tmp14
tmp24 = tl_math.exp(tmp23)
tmp25 = tmp22 + tmp24
tmp26 = tmp16 / tmp25
tmp27 = tmp26 * tmp1
tmp28 = tmp18 / tmp25
tmp29 = tmp28 * tmp4
tmp30 = tmp27 + tmp29
tmp31 = tmp21 / tmp25
tmp32 = tmp31 * tmp8
tmp33 = tmp30 + tmp32
tmp34 = tmp24 / tmp25
tmp35 = tmp34 * tmp12
tmp36 = tmp33 + tmp35
tl.store(out_ptr0 + (x0), tmp14, xmask)
tl.store(out_ptr1 + (x0), tmp25, xmask)
tl.store(out_ptr2 + (x0), tmp36, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/is/ciswv4nv4tnmcgltqwrg467dpzyhzsbsfuhmwaqhrfirgvy2wlj3.py
# Topologically Sorted Source Nodes: [mul, result, result_1, add, result_2], Original ATen: [aten.mul, aten._softmax, aten.add, aten.div]
# Source node to ATen node mapping:
# add => add
# mul => mul
# result => amax, div, exp, sub
# result_1 => mul_1
# result_2 => div_1
# Graph fragment:
# %mul : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view, %view_1), kwargs = {})
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul, [-1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
# %mul_1 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div, %view_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sum_2, 1e-13), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul_1, %add), kwargs = {})
triton_poi_fused__softmax_add_div_mul_2 = async_compile.triton('triton_poi_fused__softmax_add_div_mul_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_add_div_mul_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_add_div_mul_2(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0 + (4*(x1 // 4))), xmask)
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr4 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp4 = tmp2 - tmp3
tmp5 = tl_math.exp(tmp4)
tmp7 = tmp5 / tmp6
tmp8 = tmp7 * tmp1
tmp10 = 1e-13
tmp11 = tmp9 + tmp10
tmp12 = tmp8 / tmp11
tl.store(out_ptr0 + (x2), tmp12, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/jp/cjprp4gkdednpeuxi4f7cltxj3u3gp6fwnvqfpxsisvgw5wmd2kv.py
# Topologically Sorted Source Nodes: [contiguous_4, attended_premises], Original ATen: [aten.clone, aten.mul]
# Source node to ATen node mapping:
# attended_premises => mul_4
# contiguous_4 => clone_4
# Graph fragment:
# %clone_4 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_2,), kwargs = {memory_format: torch.contiguous_format})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%bmm_1, %clone_4), kwargs = {})
triton_poi_fused_clone_mul_3 = async_compile.triton('triton_poi_fused_clone_mul_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_mul_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_clone_mul_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/un/cunrsbrp6jdeyt2khuxbnbybp45l4w5j3v646exq5ecwb4qtljaf.py
# Topologically Sorted Source Nodes: [mul_2, result_3, result_4, sum_2], Original ATen: [aten.mul, aten._softmax, aten.sum]
# Source node to ATen node mapping:
# mul_2 => mul_2
# result_3 => amax_1, div_2, exp_1, sub_1, sum_3
# result_4 => mul_3
# sum_2 => sum_4
# Graph fragment:
# %mul_2 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_3, %view_4), kwargs = {})
# %amax_1 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_2, [-1], True), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_2, %amax_1), kwargs = {})
# %exp_1 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_1,), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_1, [-1], True), kwargs = {})
# %div_2 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp_1, %sum_3), kwargs = {})
# %mul_3 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div_2, %view_4), kwargs = {})
# %sum_4 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul_3, [-1], True), kwargs = {})
triton_poi_fused__softmax_mul_sum_4 = async_compile.triton('triton_poi_fused__softmax_mul_sum_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_mul_sum_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_mul_sum_4(in_ptr0, in_ptr1, out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + ((16*(x0 // 4)) + (x0 % 4)), xmask)
tmp1 = tl.load(in_ptr1 + (4*(x0 // 4)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (4 + (16*(x0 // 4)) + (x0 % 4)), xmask)
tmp4 = tl.load(in_ptr1 + (1 + (4*(x0 // 4))), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (8 + (16*(x0 // 4)) + (x0 % 4)), xmask)
tmp8 = tl.load(in_ptr1 + (2 + (4*(x0 // 4))), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (12 + (16*(x0 // 4)) + (x0 % 4)), xmask)
tmp12 = tl.load(in_ptr1 + (3 + (4*(x0 // 4))), xmask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp5 = tmp3 * tmp4
tmp6 = triton_helpers.maximum(tmp2, tmp5)
tmp9 = tmp7 * tmp8
tmp10 = triton_helpers.maximum(tmp6, tmp9)
tmp13 = tmp11 * tmp12
tmp14 = triton_helpers.maximum(tmp10, tmp13)
tmp15 = tmp2 - tmp14
tmp16 = tl_math.exp(tmp15)
tmp17 = tmp5 - tmp14
tmp18 = tl_math.exp(tmp17)
tmp19 = tmp16 + tmp18
tmp20 = tmp9 - tmp14
tmp21 = tl_math.exp(tmp20)
tmp22 = tmp19 + tmp21
tmp23 = tmp13 - tmp14
tmp24 = tl_math.exp(tmp23)
tmp25 = tmp22 + tmp24
tmp26 = tmp16 / tmp25
tmp27 = tmp26 * tmp1
tmp28 = tmp18 / tmp25
tmp29 = tmp28 * tmp4
tmp30 = tmp27 + tmp29
tmp31 = tmp21 / tmp25
tmp32 = tmp31 * tmp8
tmp33 = tmp30 + tmp32
tmp34 = tmp24 / tmp25
tmp35 = tmp34 * tmp12
tmp36 = tmp33 + tmp35
tl.store(out_ptr0 + (x0), tmp14, xmask)
tl.store(out_ptr1 + (x0), tmp25, xmask)
tl.store(out_ptr2 + (x0), tmp36, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/lc/clclcxx7qvisvs673o3jouryicpphx7lqleeokcpqyqjj6bqri7s.py
# Topologically Sorted Source Nodes: [mul_2, result_3, result_4, add_1, result_5], Original ATen: [aten.mul, aten._softmax, aten.add, aten.div]
# Source node to ATen node mapping:
# add_1 => add_1
# mul_2 => mul_2
# result_3 => amax_1, div_2, exp_1, sub_1
# result_4 => mul_3
# result_5 => div_3
# Graph fragment:
# %mul_2 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_3, %view_4), kwargs = {})
# %amax_1 : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_2, [-1], True), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_2, %amax_1), kwargs = {})
# %exp_1 : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_1,), kwargs = {})
# %div_2 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp_1, %sum_3), kwargs = {})
# %mul_3 : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div_2, %view_4), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sum_4, 1e-13), kwargs = {})
# %div_3 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul_3, %add_1), kwargs = {})
triton_poi_fused__softmax_add_div_mul_5 = async_compile.triton('triton_poi_fused__softmax_add_div_mul_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16, 4], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_add_div_mul_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_add_div_mul_5(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + ((4*x1) + (16*(y0 // 4)) + (y0 % 4)), xmask & ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x1 + (4*(y0 // 4))), xmask & ymask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (y0), ymask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr3 + (y0), ymask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr4 + (y0), ymask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp4 = tmp2 - tmp3
tmp5 = tl_math.exp(tmp4)
tmp7 = tmp5 / tmp6
tmp8 = tmp7 * tmp1
tmp10 = 1e-13
tmp11 = tmp9 + tmp10
tmp12 = tmp8 / tmp11
tl.store(out_ptr0 + (x1 + (4*y0)), tmp12, xmask & ymask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(arg2_1, (4, 4), (4, 1))
assert_size_stride(arg3_1, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 1, 4), torch.float32)
# Topologically Sorted Source Nodes: [contiguous], Original ATen: [aten.clone]
stream0 = get_raw_stream(0)
triton_poi_fused_clone_0.run(arg1_1, buf0, 64, grid=grid(64), stream=stream0)
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [contiguous, similarity_matrix], Original ATen: [aten.clone, aten.bmm]
extern_kernels.bmm(arg0_1, buf0, out=buf1)
buf2 = empty_strided_cuda((16, 1), (1, 16), torch.float32)
buf3 = empty_strided_cuda((16, 1), (1, 16), torch.float32)
buf4 = empty_strided_cuda((16, 1), (1, 16), torch.float32)
# Topologically Sorted Source Nodes: [mul, result, result_1, sum_1], Original ATen: [aten.mul, aten._softmax, aten.sum]
triton_poi_fused__softmax_mul_sum_1.run(buf1, arg2_1, buf2, buf3, buf4, 16, grid=grid(16), stream=stream0)
buf5 = reinterpret_tensor(buf0, (16, 4), (4, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [mul, result, result_1, add, result_2], Original ATen: [aten.mul, aten._softmax, aten.add, aten.div]
triton_poi_fused__softmax_add_div_mul_2.run(buf1, arg2_1, buf2, buf3, buf4, buf5, 64, grid=grid(64), stream=stream0)
buf6 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [weighted_sum], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf5, (4, 4, 4), (16, 4, 1), 0), arg1_1, out=buf6)
del arg1_1
buf7 = buf6; del buf6 # reuse
# Topologically Sorted Source Nodes: [contiguous_4, attended_premises], Original ATen: [aten.clone, aten.mul]
triton_poi_fused_clone_mul_3.run(buf7, arg3_1, 64, grid=grid(64), stream=stream0)
buf8 = buf4; del buf4 # reuse
buf9 = buf3; del buf3 # reuse
buf10 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [mul_2, result_3, result_4, sum_2], Original ATen: [aten.mul, aten._softmax, aten.sum]
triton_poi_fused__softmax_mul_sum_4.run(buf1, arg3_1, buf8, buf9, buf10, 16, grid=grid(16), stream=stream0)
buf11 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul_2, result_3, result_4, add_1, result_5], Original ATen: [aten.mul, aten._softmax, aten.add, aten.div]
triton_poi_fused__softmax_add_div_mul_5.run(buf1, arg3_1, buf8, buf9, buf10, buf11, 16, 4, grid=grid(16, 4), stream=stream0)
del arg3_1
del buf10
del buf8
del buf9
buf12 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [weighted_sum_1], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf11, (4, 4, 4), (16, 4, 1), 0), arg0_1, out=buf12)
del arg0_1
buf13 = buf12; del buf12 # reuse
# Topologically Sorted Source Nodes: [contiguous_5, attended_hypotheses], Original ATen: [aten.clone, aten.mul]
triton_poi_fused_clone_mul_3.run(buf13, arg2_1, 64, grid=grid(64), stream=stream0)
del arg2_1
return (buf7, buf13, reinterpret_tensor(buf5, (4, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf11, (4, 4, 4), (16, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
arg3_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1, arg3_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
def masked_softmax(tensor, mask):
"""
Apply a masked softmax on the last dimension of a tensor.
The input tensor and mask should be of size (batch, *, sequence_length).
Args:
tensor: The tensor on which the softmax function must be applied along
the last dimension.
mask: A mask of the same size as the tensor with 0s in the positions of
the values that must be masked and 1s everywhere else.
Returns:
A tensor of the same size as the inputs containing the result of the
softmax.
"""
tensor_shape = tensor.size()
reshaped_tensor = tensor.view(-1, tensor_shape[-1])
while mask.dim() < tensor.dim():
mask = mask.unsqueeze(1)
mask = mask.expand_as(tensor).contiguous().float()
reshaped_mask = mask.view(-1, mask.size()[-1])
result = nn.functional.softmax(reshaped_tensor * reshaped_mask, dim=-1)
result = result * reshaped_mask
result = result / (result.sum(dim=-1, keepdim=True) + 1e-13)
return result.view(*tensor_shape)
def weighted_sum(tensor, weights, mask):
"""
Apply a weighted sum on the vectors along the last dimension of 'tensor',
and mask the vectors in the result with 'mask'.
Args:
tensor: A tensor of vectors on which a weighted sum must be applied.
weights: The weights to use in the weighted sum.
mask: A mask to apply on the result of the weighted sum.
Returns:
A new tensor containing the result of the weighted sum after the mask
has been applied on it.
"""
weighted_sum = weights.bmm(tensor)
while mask.dim() < weighted_sum.dim():
mask = mask.unsqueeze(1)
mask = mask.transpose(-1, -2)
mask = mask.expand_as(weighted_sum).contiguous().float()
return weighted_sum * mask
class SoftmaxAttention(nn.Module):
"""
Attention layer taking premises and hypotheses encoded by an RNN as input
and computing the soft attention between their elements.
The dot product of the encoded vectors in the premises and hypotheses is
first computed. The softmax of the result is then used in a weighted sum
of the vectors of the premises for each element of the hypotheses, and
conversely for the elements of the premises.
"""
def forward(self, premise_batch, premise_mask, hypothesis_batch,
hypothesis_mask):
"""
Args:
premise_batch: A batch of sequences of vectors representing the
premises in some NLI task. The batch is assumed to have the
size (batch, sequences, vector_dim).
premise_mask: A mask for the sequences in the premise batch, to
ignore padding data in the sequences during the computation of
the attention.
hypothesis_batch: A batch of sequences of vectors representing the
hypotheses in some NLI task. The batch is assumed to have the
size (batch, sequences, vector_dim).
hypothesis_mask: A mask for the sequences in the hypotheses batch,
to ignore padding data in the sequences during the computation
of the attention.
Returns:
attended_premises: The sequences of attention vectors for the
premises in the input batch.
attended_hypotheses: The sequences of attention vectors for the
hypotheses in the input batch.
prem_hyp_attn: TODO
hyp_prem_attn: TODO
"""
similarity_matrix = premise_batch.bmm(hypothesis_batch.transpose(2,
1).contiguous())
prem_hyp_attn = masked_softmax(similarity_matrix, hypothesis_mask)
hyp_prem_attn = masked_softmax(similarity_matrix.transpose(1, 2).
contiguous(), premise_mask)
attended_premises = weighted_sum(hypothesis_batch, prem_hyp_attn,
premise_mask)
attended_hypotheses = weighted_sum(premise_batch, hyp_prem_attn,
hypothesis_mask)
return (attended_premises, attended_hypotheses, prem_hyp_attn,
hyp_prem_attn)
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4]), torch.rand([4, 4, 4]
), torch.rand([4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_clone_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tl.store(out_ptr0 + x0, tmp0, xmask)
@triton.jit
def triton_poi_fused__softmax_mul_sum_1(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, out_ptr2, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + 4 * (x0 // 4), xmask, eviction_policy='evict_last'
)
tmp3 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * (x0 // 4)), xmask, eviction_policy=
'evict_last')
tmp7 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + 4 * (x0 // 4)), xmask, eviction_policy=
'evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp12 = tl.load(in_ptr1 + (3 + 4 * (x0 // 4)), xmask, eviction_policy=
'evict_last')
tmp2 = tmp0 * tmp1
tmp5 = tmp3 * tmp4
tmp6 = triton_helpers.maximum(tmp2, tmp5)
tmp9 = tmp7 * tmp8
tmp10 = triton_helpers.maximum(tmp6, tmp9)
tmp13 = tmp11 * tmp12
tmp14 = triton_helpers.maximum(tmp10, tmp13)
tmp15 = tmp2 - tmp14
tmp16 = tl_math.exp(tmp15)
tmp17 = tmp5 - tmp14
tmp18 = tl_math.exp(tmp17)
tmp19 = tmp16 + tmp18
tmp20 = tmp9 - tmp14
tmp21 = tl_math.exp(tmp20)
tmp22 = tmp19 + tmp21
tmp23 = tmp13 - tmp14
tmp24 = tl_math.exp(tmp23)
tmp25 = tmp22 + tmp24
tmp26 = tmp16 / tmp25
tmp27 = tmp26 * tmp1
tmp28 = tmp18 / tmp25
tmp29 = tmp28 * tmp4
tmp30 = tmp27 + tmp29
tmp31 = tmp21 / tmp25
tmp32 = tmp31 * tmp8
tmp33 = tmp30 + tmp32
tmp34 = tmp24 / tmp25
tmp35 = tmp34 * tmp12
tmp36 = tmp33 + tmp35
tl.store(out_ptr0 + x0, tmp14, xmask)
tl.store(out_ptr1 + x0, tmp25, xmask)
tl.store(out_ptr2 + x0, tmp36, xmask)
@triton.jit
def triton_poi_fused__softmax_add_div_mul_2(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (x0 + 4 * (x1 // 4)), xmask)
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr4 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp4 = tmp2 - tmp3
tmp5 = tl_math.exp(tmp4)
tmp7 = tmp5 / tmp6
tmp8 = tmp7 * tmp1
tmp10 = 1e-13
tmp11 = tmp9 + tmp10
tmp12 = tmp8 / tmp11
tl.store(out_ptr0 + x2, tmp12, xmask)
@triton.jit
def triton_poi_fused_clone_mul_3(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
@triton.jit
def triton_poi_fused__softmax_mul_sum_4(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, out_ptr2, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (16 * (x0 // 4) + x0 % 4), xmask)
tmp1 = tl.load(in_ptr1 + 4 * (x0 // 4), xmask, eviction_policy='evict_last'
)
tmp3 = tl.load(in_ptr0 + (4 + 16 * (x0 // 4) + x0 % 4), xmask)
tmp4 = tl.load(in_ptr1 + (1 + 4 * (x0 // 4)), xmask, eviction_policy=
'evict_last')
tmp7 = tl.load(in_ptr0 + (8 + 16 * (x0 // 4) + x0 % 4), xmask)
tmp8 = tl.load(in_ptr1 + (2 + 4 * (x0 // 4)), xmask, eviction_policy=
'evict_last')
tmp11 = tl.load(in_ptr0 + (12 + 16 * (x0 // 4) + x0 % 4), xmask)
tmp12 = tl.load(in_ptr1 + (3 + 4 * (x0 // 4)), xmask, eviction_policy=
'evict_last')
tmp2 = tmp0 * tmp1
tmp5 = tmp3 * tmp4
tmp6 = triton_helpers.maximum(tmp2, tmp5)
tmp9 = tmp7 * tmp8
tmp10 = triton_helpers.maximum(tmp6, tmp9)
tmp13 = tmp11 * tmp12
tmp14 = triton_helpers.maximum(tmp10, tmp13)
tmp15 = tmp2 - tmp14
tmp16 = tl_math.exp(tmp15)
tmp17 = tmp5 - tmp14
tmp18 = tl_math.exp(tmp17)
tmp19 = tmp16 + tmp18
tmp20 = tmp9 - tmp14
tmp21 = tl_math.exp(tmp20)
tmp22 = tmp19 + tmp21
tmp23 = tmp13 - tmp14
tmp24 = tl_math.exp(tmp23)
tmp25 = tmp22 + tmp24
tmp26 = tmp16 / tmp25
tmp27 = tmp26 * tmp1
tmp28 = tmp18 / tmp25
tmp29 = tmp28 * tmp4
tmp30 = tmp27 + tmp29
tmp31 = tmp21 / tmp25
tmp32 = tmp31 * tmp8
tmp33 = tmp30 + tmp32
tmp34 = tmp24 / tmp25
tmp35 = tmp34 * tmp12
tmp36 = tmp33 + tmp35
tl.store(out_ptr0 + x0, tmp14, xmask)
tl.store(out_ptr1 + x0, tmp25, xmask)
tl.store(out_ptr2 + x0, tmp36, xmask)
@triton.jit
def triton_poi_fused__softmax_add_div_mul_5(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, out_ptr0, ynumel, xnumel, YBLOCK: tl.constexpr,
XBLOCK: tl.constexpr):
ynumel = 16
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (4 * x1 + 16 * (y0 // 4) + y0 % 4), xmask &
ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (x1 + 4 * (y0 // 4)), xmask & ymask,
eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + y0, ymask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr3 + y0, ymask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr4 + y0, ymask, eviction_policy='evict_last')
tmp2 = tmp0 * tmp1
tmp4 = tmp2 - tmp3
tmp5 = tl_math.exp(tmp4)
tmp7 = tmp5 / tmp6
tmp8 = tmp7 * tmp1
tmp10 = 1e-13
tmp11 = tmp9 + tmp10
tmp12 = tmp8 / tmp11
tl.store(out_ptr0 + (x1 + 4 * y0), tmp12, xmask & ymask)
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(arg2_1, (4, 4), (4, 1))
assert_size_stride(arg3_1, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 1, 4), torch.float32)
get_raw_stream(0)
triton_poi_fused_clone_0[grid(64)](arg1_1, buf0, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(arg0_1, buf0, out=buf1)
buf2 = empty_strided_cuda((16, 1), (1, 16), torch.float32)
buf3 = empty_strided_cuda((16, 1), (1, 16), torch.float32)
buf4 = empty_strided_cuda((16, 1), (1, 16), torch.float32)
triton_poi_fused__softmax_mul_sum_1[grid(16)](buf1, arg2_1, buf2,
buf3, buf4, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf5 = reinterpret_tensor(buf0, (16, 4), (4, 1), 0)
del buf0
triton_poi_fused__softmax_add_div_mul_2[grid(64)](buf1, arg2_1,
buf2, buf3, buf4, buf5, 64, XBLOCK=64, num_warps=1, num_stages=1)
buf6 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf5, (4, 4, 4), (16, 4, 1),
0), arg1_1, out=buf6)
del arg1_1
buf7 = buf6
del buf6
triton_poi_fused_clone_mul_3[grid(64)](buf7, arg3_1, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf8 = buf4
del buf4
buf9 = buf3
del buf3
buf10 = buf2
del buf2
triton_poi_fused__softmax_mul_sum_4[grid(16)](buf1, arg3_1, buf8,
buf9, buf10, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf11 = empty_strided_cuda((16, 4), (4, 1), torch.float32)
triton_poi_fused__softmax_add_div_mul_5[grid(16, 4)](buf1, arg3_1,
buf8, buf9, buf10, buf11, 16, 4, XBLOCK=4, YBLOCK=8, num_warps=
1, num_stages=1)
del arg3_1
del buf10
del buf8
del buf9
buf12 = buf1
del buf1
extern_kernels.bmm(reinterpret_tensor(buf11, (4, 4, 4), (16, 4, 1),
0), arg0_1, out=buf12)
del arg0_1
buf13 = buf12
del buf12
triton_poi_fused_clone_mul_3[grid(64)](buf13, arg2_1, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del arg2_1
return buf7, buf13, reinterpret_tensor(buf5, (4, 4, 4), (16, 4, 1), 0
), reinterpret_tensor(buf11, (4, 4, 4), (16, 4, 1), 0)
def masked_softmax(tensor, mask):
"""
Apply a masked softmax on the last dimension of a tensor.
The input tensor and mask should be of size (batch, *, sequence_length).
Args:
tensor: The tensor on which the softmax function must be applied along
the last dimension.
mask: A mask of the same size as the tensor with 0s in the positions of
the values that must be masked and 1s everywhere else.
Returns:
A tensor of the same size as the inputs containing the result of the
softmax.
"""
tensor_shape = tensor.size()
reshaped_tensor = tensor.view(-1, tensor_shape[-1])
while mask.dim() < tensor.dim():
mask = mask.unsqueeze(1)
mask = mask.expand_as(tensor).contiguous().float()
reshaped_mask = mask.view(-1, mask.size()[-1])
result = nn.functional.softmax(reshaped_tensor * reshaped_mask, dim=-1)
result = result * reshaped_mask
result = result / (result.sum(dim=-1, keepdim=True) + 1e-13)
return result.view(*tensor_shape)
def weighted_sum(tensor, weights, mask):
"""
Apply a weighted sum on the vectors along the last dimension of 'tensor',
and mask the vectors in the result with 'mask'.
Args:
tensor: A tensor of vectors on which a weighted sum must be applied.
weights: The weights to use in the weighted sum.
mask: A mask to apply on the result of the weighted sum.
Returns:
A new tensor containing the result of the weighted sum after the mask
has been applied on it.
"""
weighted_sum = weights.bmm(tensor)
while mask.dim() < weighted_sum.dim():
mask = mask.unsqueeze(1)
mask = mask.transpose(-1, -2)
mask = mask.expand_as(weighted_sum).contiguous().float()
return weighted_sum * mask
class SoftmaxAttentionNew(nn.Module):
"""
Attention layer taking premises and hypotheses encoded by an RNN as input
and computing the soft attention between their elements.
The dot product of the encoded vectors in the premises and hypotheses is
first computed. The softmax of the result is then used in a weighted sum
of the vectors of the premises for each element of the hypotheses, and
conversely for the elements of the premises.
"""
def forward(self, input_0, input_1, input_2, input_3):
arg0_1 = input_0
arg2_1 = input_1
arg1_1 = input_2
arg3_1 = input_3
output = call([arg0_1, arg1_1, arg2_1, arg3_1])
return output[0], output[1], output[2], output[3]
| marvosyntactical/fs2018ex3viz | SoftmaxAttention | false | 7,179 | [
"Apache-2.0"
] | 1 | 9002133a45b52c596efa91d842f691fe1f066a6c | https://github.com/marvosyntactical/fs2018ex3viz/tree/9002133a45b52c596efa91d842f691fe1f066a6c | import torch
import torch.nn as nn
def masked_softmax(tensor, mask):
"""
Apply a masked softmax on the last dimension of a tensor.
The input tensor and mask should be of size (batch, *, sequence_length).
Args:
tensor: The tensor on which the softmax function must be applied along
the last dimension.
mask: A mask of the same size as the tensor with 0s in the positions of
the values that must be masked and 1s everywhere else.
Returns:
A tensor of the same size as the inputs containing the result of the
softmax.
"""
tensor_shape = tensor.size()
reshaped_tensor = tensor.view(-1, tensor_shape[-1])
while mask.dim() < tensor.dim():
mask = mask.unsqueeze(1)
mask = mask.expand_as(tensor).contiguous().float()
reshaped_mask = mask.view(-1, mask.size()[-1])
result = nn.functional.softmax(reshaped_tensor * reshaped_mask, dim=-1)
result = result * reshaped_mask
result = result / (result.sum(dim=-1, keepdim=True) + 1e-13)
return result.view(*tensor_shape)
def weighted_sum(tensor, weights, mask):
"""
Apply a weighted sum on the vectors along the last dimension of 'tensor',
and mask the vectors in the result with 'mask'.
Args:
tensor: A tensor of vectors on which a weighted sum must be applied.
weights: The weights to use in the weighted sum.
mask: A mask to apply on the result of the weighted sum.
Returns:
A new tensor containing the result of the weighted sum after the mask
has been applied on it.
"""
weighted_sum = weights.bmm(tensor)
while mask.dim() < weighted_sum.dim():
mask = mask.unsqueeze(1)
mask = mask.transpose(-1, -2)
mask = mask.expand_as(weighted_sum).contiguous().float()
return weighted_sum * mask
class Model(nn.Module):
"""
Attention layer taking premises and hypotheses encoded by an RNN as input
and computing the soft attention between their elements.
The dot product of the encoded vectors in the premises and hypotheses is
first computed. The softmax of the result is then used in a weighted sum
of the vectors of the premises for each element of the hypotheses, and
conversely for the elements of the premises.
"""
def forward(self, premise_batch, premise_mask, hypothesis_batch,
hypothesis_mask):
"""
Args:
premise_batch: A batch of sequences of vectors representing the
premises in some NLI task. The batch is assumed to have the
size (batch, sequences, vector_dim).
premise_mask: A mask for the sequences in the premise batch, to
ignore padding data in the sequences during the computation of
the attention.
hypothesis_batch: A batch of sequences of vectors representing the
hypotheses in some NLI task. The batch is assumed to have the
size (batch, sequences, vector_dim).
hypothesis_mask: A mask for the sequences in the hypotheses batch,
to ignore padding data in the sequences during the computation
of the attention.
Returns:
attended_premises: The sequences of attention vectors for the
premises in the input batch.
attended_hypotheses: The sequences of attention vectors for the
hypotheses in the input batch.
prem_hyp_attn: TODO
hyp_prem_attn: TODO
"""
similarity_matrix = premise_batch.bmm(hypothesis_batch.transpose(2,
1).contiguous())
prem_hyp_attn = masked_softmax(similarity_matrix, hypothesis_mask)
hyp_prem_attn = masked_softmax(similarity_matrix.transpose(1, 2).
contiguous(), premise_mask)
attended_premises = weighted_sum(hypothesis_batch, prem_hyp_attn,
premise_mask)
attended_hypotheses = weig
# ... truncated (>4000 chars) for memory efficiency |
_leaky_relu | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/js/cjszscodslnfyiv2d7igvqacsrgsys76rucgsdtlfb7q6ntp2vru.py
# Topologically Sorted Source Nodes: [x_neg, max_1], Original ATen: [aten.mul, aten.maximum]
# Source node to ATen node mapping:
# max_1 => maximum
# x_neg => mul
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, 0.1), kwargs = {})
# %maximum : [num_users=1] = call_function[target=torch.ops.aten.maximum.default](args = (%mul, %arg0_1), kwargs = {})
triton_poi_fused_maximum_mul_0 = async_compile.triton('triton_poi_fused_maximum_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_maximum_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_maximum_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.1
tmp2 = tmp0 * tmp1
tmp3 = triton_helpers.maximum(tmp2, tmp0)
tl.store(out_ptr0 + (x0), tmp3, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_neg, max_1], Original ATen: [aten.mul, aten.maximum]
stream0 = get_raw_stream(0)
triton_poi_fused_maximum_mul_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
class _leaky_relu(nn.Module):
def __init__(self):
super(_leaky_relu, self).__init__()
def forward(self, x):
x_neg = 0.1 * x
return torch.max(x_neg, x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_maximum_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.1
tmp2 = tmp0 * tmp1
tmp3 = triton_helpers.maximum(tmp2, tmp0)
tl.store(out_ptr0 + x0, tmp3, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_maximum_mul_0[grid(256)](arg0_1, buf0, 256, XBLOCK
=128, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class _leaky_reluNew(nn.Module):
def __init__(self):
super(_leaky_reluNew, self).__init__()
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| maxuanquang/SfmLearner-Redesign | _leaky_relu | false | 7,180 | [
"MIT"
] | 1 | 0250a9cc443b5754ba45f69153a03ca26f903a7b | https://github.com/maxuanquang/SfmLearner-Redesign/tree/0250a9cc443b5754ba45f69153a03ca26f903a7b | import torch
from torch import nn
class Model(nn.Module):
def __init__(self):
super().__init__()
def forward(self, x):
x_neg = 0.1 * x
return torch.max(x_neg, x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
CriticNet | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/xs/cxsx2jlz7qi3eerscrmq6daj5pyplc3rcuj6cilxq2zofbhntrow.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x_1 => gt, mul, where
# Graph fragment:
# %gt : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_3, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_3, 0.01), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %view_3, %mul), kwargs = {})
triton_poi_fused_leaky_relu_0 = async_compile.triton('triton_poi_fused_leaky_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 2560
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 40
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/5v/c5vn7uggpfqtlchl6swvsdgcgavhwaeod4idwzjydfczuqulpenk.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x_2 => gt_1, mul_1, where_1
# Graph fragment:
# %gt_1 : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_5, 0), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_5, 0.01), kwargs = {})
# %where_1 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_1, %view_5, %mul_1), kwargs = {})
triton_poi_fused_leaky_relu_1 = async_compile.triton('triton_poi_fused_leaky_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 1920
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 30
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/7u/c7uibvrhz722pvnfpzp3uhgihgkfifzjsn2rpnnsjorfgng2msxf.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x_3 => gt_2, mul_2, where_2
# Graph fragment:
# %gt_2 : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_7, 0), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_7, 0.01), kwargs = {})
# %where_2 : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt_2, %view_7, %mul_2), kwargs = {})
triton_poi_fused_leaky_relu_2 = async_compile.triton('triton_poi_fused_leaky_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[512],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_2(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 8
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11 = args
args.clear()
assert_size_stride(primals_1, (20, 4), (4, 1))
assert_size_stride(primals_2, (20, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (40, 20), (20, 1))
assert_size_stride(primals_5, (40, ), (1, ))
assert_size_stride(primals_6, (30, 40), (40, 1))
assert_size_stride(primals_7, (30, ), (1, ))
assert_size_stride(primals_8, (8, 30), (30, 1))
assert_size_stride(primals_9, (8, ), (1, ))
assert_size_stride(primals_10, (1, 8), (8, 1))
assert_size_stride(primals_11, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 20), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 40), (40, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf0, reinterpret_tensor(primals_4, (20, 40), (1, 20), 0), out=buf1)
buf2 = empty_strided_cuda((4, 4, 4, 40), (640, 160, 40, 1), torch.bool)
buf3 = empty_strided_cuda((4, 4, 4, 40), (640, 160, 40, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_0.run(buf1, primals_5, buf2, buf3, 2560, grid=grid(2560), stream=stream0)
del buf1
del primals_5
buf4 = empty_strided_cuda((64, 30), (30, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf3, (64, 40), (40, 1), 0), reinterpret_tensor(primals_6, (40, 30), (1, 40), 0), out=buf4)
buf5 = empty_strided_cuda((4, 4, 4, 30), (480, 120, 30, 1), torch.bool)
buf6 = empty_strided_cuda((4, 4, 4, 30), (480, 120, 30, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_1.run(buf4, primals_7, buf5, buf6, 1920, grid=grid(1920), stream=stream0)
del buf4
del primals_7
buf7 = empty_strided_cuda((64, 8), (8, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf6, (64, 30), (30, 1), 0), reinterpret_tensor(primals_8, (30, 8), (1, 30), 0), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4, 8), (128, 32, 8, 1), torch.bool)
buf9 = empty_strided_cuda((4, 4, 4, 8), (128, 32, 8, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_2.run(buf7, primals_9, buf8, buf9, 512, grid=grid(512), stream=stream0)
del buf7
del primals_9
buf11 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_11, reinterpret_tensor(buf9, (64, 8), (8, 1), 0), reinterpret_tensor(primals_10, (8, 1), (1, 8), 0), alpha=1, beta=1, out=buf11)
del primals_11
return (reinterpret_tensor(buf11, (4, 4, 4, 1), (16, 4, 1, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf0, buf2, reinterpret_tensor(buf3, (64, 40), (40, 1), 0), buf5, reinterpret_tensor(buf6, (64, 30), (30, 1), 0), buf8, reinterpret_tensor(buf9, (64, 8), (8, 1), 0), primals_10, primals_8, primals_6, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((20, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((20, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((40, 20), (20, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((40, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((30, 40), (40, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((30, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((8, 30), (30, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((8, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((1, 8), (8, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class CriticNet(nn.Module):
def __init__(self):
super(CriticNet, self).__init__()
self.fc1 = nn.Linear(4, 20)
self.fc2 = nn.Linear(20, 40)
self.fc3 = nn.Linear(40, 30)
self.fc4 = nn.Linear(30, 8)
self.fc5 = nn.Linear(8, 1)
def forward(self, x):
x = self.fc1(x)
x = F.leaky_relu(self.fc2(x))
x = F.leaky_relu(self.fc3(x))
x = F.leaky_relu(self.fc4(x))
x = self.fc5(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 2560
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 40
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
@triton.jit
def triton_poi_fused_leaky_relu_1(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 1920
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 30
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
@triton.jit
def triton_poi_fused_leaky_relu_2(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 512
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 8
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.01
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11) = args
args.clear()
assert_size_stride(primals_1, (20, 4), (4, 1))
assert_size_stride(primals_2, (20,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (40, 20), (20, 1))
assert_size_stride(primals_5, (40,), (1,))
assert_size_stride(primals_6, (30, 40), (40, 1))
assert_size_stride(primals_7, (30,), (1,))
assert_size_stride(primals_8, (8, 30), (30, 1))
assert_size_stride(primals_9, (8,), (1,))
assert_size_stride(primals_10, (1, 8), (8, 1))
assert_size_stride(primals_11, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 20), (20, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 20), (1, 4),
0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 40), (40, 1), torch.float32)
extern_kernels.mm(buf0, reinterpret_tensor(primals_4, (20, 40), (1,
20), 0), out=buf1)
buf2 = empty_strided_cuda((4, 4, 4, 40), (640, 160, 40, 1), torch.bool)
buf3 = empty_strided_cuda((4, 4, 4, 40), (640, 160, 40, 1), torch.
float32)
get_raw_stream(0)
triton_poi_fused_leaky_relu_0[grid(2560)](buf1, primals_5, buf2,
buf3, 2560, XBLOCK=256, num_warps=4, num_stages=1)
del buf1
del primals_5
buf4 = empty_strided_cuda((64, 30), (30, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf3, (64, 40), (40, 1), 0),
reinterpret_tensor(primals_6, (40, 30), (1, 40), 0), out=buf4)
buf5 = empty_strided_cuda((4, 4, 4, 30), (480, 120, 30, 1), torch.bool)
buf6 = empty_strided_cuda((4, 4, 4, 30), (480, 120, 30, 1), torch.
float32)
triton_poi_fused_leaky_relu_1[grid(1920)](buf4, primals_7, buf5,
buf6, 1920, XBLOCK=128, num_warps=4, num_stages=1)
del buf4
del primals_7
buf7 = empty_strided_cuda((64, 8), (8, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf6, (64, 30), (30, 1), 0),
reinterpret_tensor(primals_8, (30, 8), (1, 30), 0), out=buf7)
buf8 = empty_strided_cuda((4, 4, 4, 8), (128, 32, 8, 1), torch.bool)
buf9 = empty_strided_cuda((4, 4, 4, 8), (128, 32, 8, 1), torch.float32)
triton_poi_fused_leaky_relu_2[grid(512)](buf7, primals_9, buf8,
buf9, 512, XBLOCK=256, num_warps=4, num_stages=1)
del buf7
del primals_9
buf11 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
extern_kernels.addmm(primals_11, reinterpret_tensor(buf9, (64, 8),
(8, 1), 0), reinterpret_tensor(primals_10, (8, 1), (1, 8), 0),
alpha=1, beta=1, out=buf11)
del primals_11
return reinterpret_tensor(buf11, (4, 4, 4, 1), (16, 4, 1, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf0, buf2, reinterpret_tensor(buf3, (64, 40), (40, 1), 0
), buf5, reinterpret_tensor(buf6, (64, 30), (30, 1), 0
), buf8, reinterpret_tensor(buf9, (64, 8), (8, 1), 0
), primals_10, primals_8, primals_6, primals_4
class CriticNetNew(nn.Module):
def __init__(self):
super(CriticNetNew, self).__init__()
self.fc1 = nn.Linear(4, 20)
self.fc2 = nn.Linear(20, 40)
self.fc3 = nn.Linear(40, 30)
self.fc4 = nn.Linear(30, 8)
self.fc5 = nn.Linear(8, 1)
def forward(self, input_0):
primals_1 = self.fc1.weight
primals_2 = self.fc1.bias
primals_4 = self.fc2.weight
primals_5 = self.fc2.bias
primals_6 = self.fc3.weight
primals_7 = self.fc3.bias
primals_8 = self.fc4.weight
primals_9 = self.fc4.bias
primals_10 = self.fc5.weight
primals_11 = self.fc5.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11])
return output[0]
| mathildebadoual/RL_power_systems | CriticNet | false | 7,181 | [
"MIT"
] | 1 | 825e60bad16129e0a0229d15af5110b26e0a1577 | https://github.com/mathildebadoual/RL_power_systems/tree/825e60bad16129e0a0229d15af5110b26e0a1577 | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self):
super().__init__()
self.fc1 = nn.Linear(4, 20)
self.fc2 = nn.Linear(20, 40)
self.fc3 = nn.Linear(40, 30)
self.fc4 = nn.Linear(30, 8)
self.fc5 = nn.Linear(8, 1)
def forward(self, x):
x = self.fc1(x)
x = F.leaky_relu(self.fc2(x))
x = F.leaky_relu(self.fc3(x))
x = F.leaky_relu(self.fc4(x))
x = self.fc5(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
ZeroConv2d | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/tg/ctgaykn2ev5vsympbzywhnbmo5fz5ljycgt452vxtvv7ybt7r3gf.py
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.constant_pad_nd]
# Source node to ATen node mapping:
# out => constant_pad_nd
# Graph fragment:
# %constant_pad_nd : [num_users=2] = call_function[target=torch.ops.aten.constant_pad_nd.default](args = (%primals_1, [1, 1, 1, 1], 1.0), kwargs = {})
triton_poi_fused_constant_pad_nd_0 = async_compile.triton('triton_poi_fused_constant_pad_nd_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_constant_pad_nd_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_constant_pad_nd_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 576
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 6) % 6
x0 = xindex % 6
x2 = (xindex // 36)
x4 = xindex
tmp0 = (-1) + x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = (-1) + x0
tmp6 = tmp5 >= tmp1
tmp7 = tmp5 < tmp3
tmp8 = tmp2 & tmp4
tmp9 = tmp8 & tmp6
tmp10 = tmp9 & tmp7
tmp11 = tl.load(in_ptr0 + ((-5) + x0 + (4*x1) + (16*x2)), tmp10 & xmask, other=1.0)
tl.store(out_ptr0 + (x4), tmp11, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/3i/c3iaqwcts6h3wpjzfzg6woukwoyx2igoqvo3awmc4fkavsodfyss.py
# Topologically Sorted Source Nodes: [out_1, mul, exp, out_2], Original ATen: [aten.convolution, aten.mul, aten.exp]
# Source node to ATen node mapping:
# exp => exp
# mul => mul
# out_1 => convolution
# out_2 => mul_1
# Graph fragment:
# %convolution : [num_users=2] = call_function[target=torch.ops.aten.convolution.default](args = (%constant_pad_nd, %primals_2, %primals_3, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_4, 3), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%mul,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%convolution, %exp), kwargs = {})
triton_poi_fused_convolution_exp_mul_1 = async_compile.triton('triton_poi_fused_convolution_exp_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_exp_mul_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_exp_mul_1(in_out_ptr0, in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = 3.0
tmp5 = tmp3 * tmp4
tmp6 = tl_math.exp(tmp5)
tmp7 = tmp2 * tmp6
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
tl.store(out_ptr0 + (x3), tmp7, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (1, 4, 1, 1), (4, 1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 6, 6), (144, 36, 6, 1), torch.float32)
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.constant_pad_nd]
stream0 = get_raw_stream(0)
triton_poi_fused_constant_pad_nd_0.run(primals_1, buf0, 576, grid=grid(576), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [out_1], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4, 4), (64, 16, 4, 1))
buf2 = buf1; del buf1 # reuse
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [out_1, mul, exp, out_2], Original ATen: [aten.convolution, aten.mul, aten.exp]
triton_poi_fused_convolution_exp_mul_1.run(buf2, primals_3, primals_4, buf3, 256, grid=grid(256), stream=stream0)
del primals_3
return (buf3, primals_2, primals_4, buf0, buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((1, 4, 1, 1), (4, 1, 1, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
from torch.nn import functional as F
class ZeroConv2d(nn.Module):
def __init__(self, in_channel, out_channel, padding=1):
super().__init__()
self.conv = nn.Conv2d(in_channel, out_channel, 3, padding=0)
self.conv.weight.data.zero_()
self.conv.bias.data.zero_()
self.scale = nn.Parameter(torch.zeros(1, out_channel, 1, 1))
def forward(self, input):
out = F.pad(input, [1, 1, 1, 1], value=1)
out = self.conv(out)
out = out * torch.exp(self.scale * 3)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channel': 4, 'out_channel': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_constant_pad_nd_0(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 576
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 6 % 6
x0 = xindex % 6
x2 = xindex // 36
x4 = xindex
tmp0 = -1 + x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = -1 + x0
tmp6 = tmp5 >= tmp1
tmp7 = tmp5 < tmp3
tmp8 = tmp2 & tmp4
tmp9 = tmp8 & tmp6
tmp10 = tmp9 & tmp7
tmp11 = tl.load(in_ptr0 + (-5 + x0 + 4 * x1 + 16 * x2), tmp10 & xmask,
other=1.0)
tl.store(out_ptr0 + x4, tmp11, xmask)
@triton.jit
def triton_poi_fused_convolution_exp_mul_1(in_out_ptr0, in_ptr0, in_ptr1,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = 3.0
tmp5 = tmp3 * tmp4
tmp6 = tl_math.exp(tmp5)
tmp7 = tmp2 * tmp6
tl.store(in_out_ptr0 + x3, tmp2, xmask)
tl.store(out_ptr0 + x3, tmp7, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (1, 4, 1, 1), (4, 1, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 6, 6), (144, 36, 6, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_constant_pad_nd_0[grid(576)](primals_1, buf0, 576,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_1
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4, 4), (64, 16, 4, 1))
buf2 = buf1
del buf1
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_convolution_exp_mul_1[grid(256)](buf2, primals_3,
primals_4, buf3, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_3
return buf3, primals_2, primals_4, buf0, buf2
class ZeroConv2dNew(nn.Module):
def __init__(self, in_channel, out_channel, padding=1):
super().__init__()
self.conv = nn.Conv2d(in_channel, out_channel, 3, padding=0)
self.conv.weight.data.zero_()
self.conv.bias.data.zero_()
self.scale = nn.Parameter(torch.zeros(1, out_channel, 1, 1))
def forward(self, input_0):
primals_4 = self.scale
primals_2 = self.conv.weight
primals_3 = self.conv.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4])
return output[0]
| mbaddar1/glow-pytorch | ZeroConv2d | false | 7,182 | [
"MIT"
] | 1 | e07ca542ce4dd93ddf680c51eda25d1f9db252a1 | https://github.com/mbaddar1/glow-pytorch/tree/e07ca542ce4dd93ddf680c51eda25d1f9db252a1 | import torch
from torch import nn
from torch.nn import functional as F
class Model(nn.Module):
def __init__(self, in_channel, out_channel, padding=1):
super().__init__()
self.conv = nn.Conv2d(in_channel, out_channel, 3, padding=0)
self.conv.weight.data.zero_()
self.conv.bias.data.zero_()
self.scale = nn.Parameter(torch.zeros(1, out_channel, 1, 1))
def forward(self, input):
out = F.pad(input, [1, 1, 1, 1], value=1)
out = self.conv(out)
out = out * torch.exp(self.scale * 3)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
BasicGraphConvolutionLayer | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/jp/cjpt3hzytizdpfnexpdpxvh4az4p257tvla46eoequf2fpegj4du.py
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.add]
# Source node to ATen node mapping:
# output => add_1
# Graph fragment:
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_plus_mm, %primals_5), kwargs = {})
triton_poi_fused_add_0 = async_compile.triton('triton_poi_fused_add_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [potential_msgs], Original ATen: [aten.mm]
extern_kernels.mm(primals_2, primals_1, out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
extern_kernels._mm_plus_mm(primals_3, buf0, primals_2, primals_4, out=buf1)
del buf0
del primals_4
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_0.run(buf2, primals_5, 16, grid=grid(16), stream=stream0)
del primals_5
return (buf2, reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch.nn.parameter import Parameter
class BasicGraphConvolutionLayer(torch.nn.Module):
def __init__(self, in_channels, out_channels):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.W2 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.W1 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.bias = Parameter(torch.zeros(out_channels, dtype=torch.float32))
def forward(self, X, A):
potential_msgs = torch.mm(X, self.W2)
propagated_msgs = torch.mm(A, potential_msgs)
root_update = torch.mm(X, self.W1)
output = propagated_msgs + root_update + self.bias
return output
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn.parameter import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels.mm(primals_2, primals_1, out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
extern_kernels._mm_plus_mm(primals_3, buf0, primals_2, primals_4,
out=buf1)
del buf0
del primals_4
buf2 = buf1
del buf1
get_raw_stream(0)
triton_poi_fused_add_0[grid(16)](buf2, primals_5, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del primals_5
return buf2, reinterpret_tensor(primals_2, (4, 4), (1, 4), 0
), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0)
class BasicGraphConvolutionLayerNew(torch.nn.Module):
def __init__(self, in_channels, out_channels):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.W2 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.W1 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.bias = Parameter(torch.zeros(out_channels, dtype=torch.float32))
def forward(self, input_0, input_1):
primals_1 = self.W2
primals_2 = self.W1
primals_5 = self.bias
primals_3 = input_0
primals_4 = input_1
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| mbrukman/machine-learning-book | BasicGraphConvolutionLayer | false | 7,183 | [
"MIT"
] | 1 | f29a0f8aafa63a77081f3bcec68866e33dd41776 | https://github.com/mbrukman/machine-learning-book/tree/f29a0f8aafa63a77081f3bcec68866e33dd41776 | import torch
from torch.nn.parameter import Parameter
class Model(torch.nn.Module):
def __init__(self, in_channels, out_channels):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.W2 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.W1 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.bias = Parameter(torch.zeros(out_channels, dtype=torch.float32))
def forward(self, X, A):
potential_msgs = torch.mm(X, self.W2)
propagated_msgs = torch.mm(A, potential_msgs)
root_update = torch.mm(X, self.W1)
output = propagated_msgs + root_update + self.bias
return output
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [4, 4]
|
InvConv2d | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/2v/c2vp4wevd4mmk6p3qeilou7zqojjaarvm3pedrgkmrhhjbggkpqu.py
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# out => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_1, %primals_2, None, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 4
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (4*x1)), xmask & ymask)
tl.store(out_ptr0 + (x1 + (4*y0)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/4m/c4mesezrsrzdxpuuyqxvdzjuqjyrphys7ucdqbdgokmbugszibng.py
# Topologically Sorted Source Nodes: [double], Original ATen: [aten._to_copy]
# Source node to ATen node mapping:
# double => convert_element_type
# Graph fragment:
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%squeeze, torch.float64), kwargs = {})
triton_poi_fused__to_copy_1 = async_compile.triton('triton_poi_fused__to_copy_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp64', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tmp0.to(tl.float64)
tl.store(out_ptr0 + (x0), tmp1, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/lz/clzjrdoqlebi4pdm72l7bnqxgktieirugz3rrrldxhmugh4eiwfe.py
# Topologically Sorted Source Nodes: [float_1, logdet], Original ATen: [aten._to_copy, aten.mul]
# Source node to ATen node mapping:
# float_1 => convert_element_type_1
# logdet => mul
# Graph fragment:
# %convert_element_type_1 : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%getitem_1, torch.float32), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%convert_element_type_1, 16), kwargs = {})
triton_poi_fused__to_copy_mul_2 = async_compile.triton('triton_poi_fused__to_copy_mul_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1],
filename=__file__,
triton_meta={'signature': {0: '*fp64', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {2: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=(2,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_mul_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_mul_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
tmp0 = tl.load(in_ptr0 + (0))
tmp1 = tl.broadcast_to(tmp0, [XBLOCK])
tmp2 = tmp1.to(tl.float32)
tmp3 = 16.0
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + (tl.full([XBLOCK], 0, tl.int32)), tmp4, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 1, 1), (1, 4, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(primals_2, buf0, 4, 4, grid=grid(4, 4), stream=stream0)
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(primals_1, buf0, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4, 4), (64, 16, 4, 1))
del buf0
buf2 = empty_strided_cuda((4, 4), (1, 4), torch.float64)
# Topologically Sorted Source Nodes: [double], Original ATen: [aten._to_copy]
triton_poi_fused__to_copy_1.run(primals_2, buf2, 16, grid=grid(16), stream=stream0)
# Topologically Sorted Source Nodes: [double, slogdet], Original ATen: [aten._to_copy, aten._linalg_slogdet]
buf3 = torch.ops.aten._linalg_slogdet.default(buf2)
del buf2
buf5 = buf3[1]
buf6 = buf3[2]
buf7 = buf3[3]
del buf3
buf8 = empty_strided_cuda((), (), torch.float32)
# Topologically Sorted Source Nodes: [float_1, logdet], Original ATen: [aten._to_copy, aten.mul]
triton_poi_fused__to_copy_mul_2.run(buf5, buf8, 1, grid=grid(1), stream=stream0)
del buf5
return (buf1, buf8, primals_1, primals_2, buf6, buf7, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 1, 1), (1, 4, 1, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
from torch.nn import functional as F
class InvConv2d(nn.Module):
def __init__(self, in_channel):
super().__init__()
weight = torch.randn(in_channel, in_channel)
q, _ = torch.qr(weight)
weight = q.unsqueeze(2).unsqueeze(3)
self.weight = nn.Parameter(weight)
def forward(self, input):
_, _, height, width = input.shape
out = F.conv2d(input, self.weight)
logdet = height * width * torch.slogdet(self.weight.squeeze().double()
)[1].float()
return out, logdet
def reverse(self, output):
return F.conv2d(output, self.weight.squeeze().inverse().unsqueeze(2
).unsqueeze(3))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channel': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch import nn
from torch.nn import functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_convolution_0(in_ptr0, out_ptr0, ynumel, xnumel,
YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 4
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 4 * x1), xmask & ymask)
tl.store(out_ptr0 + (x1 + 4 * y0), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused__to_copy_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tmp0.to(tl.float64)
tl.store(out_ptr0 + x0, tmp1, xmask)
@triton.jit
def triton_poi_fused__to_copy_mul_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
tmp0 = tl.load(in_ptr0 + 0)
tmp1 = tl.broadcast_to(tmp0, [XBLOCK])
tmp2 = tmp1.to(tl.float32)
tmp3 = 16.0
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + tl.full([XBLOCK], 0, tl.int32), tmp4, None)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 1, 1), (1, 4, 1, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 1, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(4, 4)](primals_2, buf0, 4, 4,
XBLOCK=4, YBLOCK=4, num_warps=1, num_stages=1)
buf1 = extern_kernels.convolution(primals_1, buf0, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4, 4), (64, 16, 4, 1))
del buf0
buf2 = empty_strided_cuda((4, 4), (1, 4), torch.float64)
triton_poi_fused__to_copy_1[grid(16)](primals_2, buf2, 16, XBLOCK=
16, num_warps=1, num_stages=1)
buf3 = torch.ops.aten._linalg_slogdet.default(buf2)
del buf2
buf5 = buf3[1]
buf6 = buf3[2]
buf7 = buf3[3]
del buf3
buf8 = empty_strided_cuda((), (), torch.float32)
triton_poi_fused__to_copy_mul_2[grid(1)](buf5, buf8, 1, XBLOCK=1,
num_warps=1, num_stages=1)
del buf5
return buf1, buf8, primals_1, primals_2, buf6, buf7
class InvConv2dNew(nn.Module):
def __init__(self, in_channel):
super().__init__()
weight = torch.randn(in_channel, in_channel)
q, _ = torch.qr(weight)
weight = q.unsqueeze(2).unsqueeze(3)
self.weight = nn.Parameter(weight)
def reverse(self, output):
return F.conv2d(output, self.weight.squeeze().inverse().unsqueeze(2
).unsqueeze(3))
def forward(self, input_0):
primals_2 = self.weight
primals_1 = input_0
output = call([primals_1, primals_2])
return output[0], output[1]
| mbaddar1/glow-pytorch | InvConv2d | false | 7,184 | [
"MIT"
] | 1 | e07ca542ce4dd93ddf680c51eda25d1f9db252a1 | https://github.com/mbaddar1/glow-pytorch/tree/e07ca542ce4dd93ddf680c51eda25d1f9db252a1 | import torch
from torch import nn
from torch.nn import functional as F
class Model(nn.Module):
def __init__(self, in_channel):
super().__init__()
weight = torch.randn(in_channel, in_channel)
q, _ = torch.qr(weight)
weight = q.unsqueeze(2).unsqueeze(3)
self.weight = nn.Parameter(weight)
def forward(self, input):
_, _, height, width = input.shape
out = F.conv2d(input, self.weight)
logdet = height * width * torch.slogdet(self.weight.squeeze().double()
)[1].float()
return out, logdet
def reverse(self, output):
return F.conv2d(output, self.weight.squeeze().inverse().unsqueeze(2
).unsqueeze(3))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
ScaledDotProductAttention | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/fz/cfzmg4qtw6jgry4nhlwopodzjz62ll3n3ykfox77hwd2crdnlh2w.py
# Topologically Sorted Source Nodes: [attn_1], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attn_1 => exp
# Graph fragment:
# %mul_tensor : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%bmm, 1), kwargs = {})
# %amax_default : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%mul_tensor, [-1], True), kwargs = {})
# %sub_tensor : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_tensor, %amax_default), kwargs = {})
# %div_tensor : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_tensor, 2.0), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%div_tensor,), kwargs = {})
triton_poi_fused__softmax_0 = async_compile.triton('triton_poi_fused__softmax_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp3 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = 0.5
tmp16 = tmp14 * tmp15
tmp17 = tl_math.exp(tmp16)
tl.store(out_ptr0 + (x2), tmp17, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/kj/ckjtlefzavjukjsytvkak6ek26zmzexpcbnlwelx4k5kascjxlf3.py
# Topologically Sorted Source Nodes: [attn_1], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# attn_1 => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_1 = async_compile.triton('triton_poi_fused__softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4), (16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [bmm], Original ATen: [aten.bmm]
extern_kernels.bmm(arg1_1, reinterpret_tensor(arg0_1, (4, 4, 4), (16, 1, 4), 0), out=buf0)
del arg0_1
del arg1_1
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [attn_1], Original ATen: [aten._softmax]
stream0 = get_raw_stream(0)
triton_poi_fused__softmax_0.run(buf0, buf1, 64, grid=grid(64), stream=stream0)
buf2 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [attn_1], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf1, buf2, 64, grid=grid(64), stream=stream0)
buf3 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(buf2, arg2_1, out=buf3)
del arg2_1
return (buf3, buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.optim.lr_scheduler
import torch.nn as nn
class ScaledDotProductAttention(nn.Module):
def __init__(self, d_model, attention_dropout=0.1):
super(ScaledDotProductAttention, self).__init__()
self.temper = d_model ** 0.5
self.dropout = nn.Dropout(attention_dropout)
self.softmax = nn.Softmax(dim=-1)
def forward(self, q, k, v, attn_mask=None):
attn = torch.bmm(q, k.transpose(1, 2)) / self.temper
if attn_mask is not None:
assert attn_mask.size() == attn.size(
), 'Attention mask shape {} mismatch with Attention logit tensor shape {}.'.format(
attn_mask.size(), attn.size())
attn.data.masked_fill_(attn_mask, -float('inf'))
attn = self.softmax(attn)
attn = self.dropout(attn)
output = torch.bmm(attn, v)
return output, attn
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4]), torch.rand([4, 4, 4])
]
def get_init_inputs():
return [[], {'d_model': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.optim.lr_scheduler
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused__softmax_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp3 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = tmp3 * tmp1
tmp6 = tmp5 * tmp1
tmp7 = triton_helpers.maximum(tmp4, tmp6)
tmp9 = tmp8 * tmp1
tmp10 = triton_helpers.maximum(tmp7, tmp9)
tmp12 = tmp11 * tmp1
tmp13 = triton_helpers.maximum(tmp10, tmp12)
tmp14 = tmp2 - tmp13
tmp15 = 0.5
tmp16 = tmp14 * tmp15
tmp17 = tl_math.exp(tmp16)
tl.store(out_ptr0 + x2, tmp17, xmask)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
arg0_1, arg1_1, arg2_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4), (16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(arg1_1, reinterpret_tensor(arg0_1, (4, 4, 4), (
16, 1, 4), 0), out=buf0)
del arg0_1
del arg1_1
buf1 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused__softmax_0[grid(64)](buf0, buf1, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf2 = buf0
del buf0
triton_poi_fused__softmax_1[grid(64)](buf1, buf2, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf3 = buf1
del buf1
extern_kernels.bmm(buf2, arg2_1, out=buf3)
del arg2_1
return buf3, buf2
class ScaledDotProductAttentionNew(nn.Module):
def __init__(self, d_model, attention_dropout=0.1):
super(ScaledDotProductAttentionNew, self).__init__()
self.temper = d_model ** 0.5
self.dropout = nn.Dropout(attention_dropout)
self.softmax = nn.Softmax(dim=-1)
def forward(self, input_0, input_1, input_2):
arg0_1 = input_0
arg1_1 = input_1
arg2_1 = input_2
output = call([arg0_1, arg1_1, arg2_1])
return output[0], output[1]
| mcoavoux/self-attentive-parser | ScaledDotProductAttention | false | 7,185 | [
"MIT"
] | 1 | fa5814ecfdbf4fde329ea725e1d2ddaa55f247d6 | https://github.com/mcoavoux/self-attentive-parser/tree/fa5814ecfdbf4fde329ea725e1d2ddaa55f247d6 | import torch
import torch.optim.lr_scheduler
import torch.nn as nn
class Model(nn.Module):
def __init__(self, d_model, attention_dropout=0.1):
super().__init__()
self.temper = d_model ** 0.5
self.dropout = nn.Dropout(attention_dropout)
self.softmax = nn.Softmax(dim=-1)
def forward(self, q, k, v, attn_mask=None):
attn = torch.bmm(q, k.transpose(1, 2)) / self.temper
if attn_mask is not None:
assert attn_mask.size() == attn.size(
), 'Attention mask shape {} mismatch with Attention logit tensor shape {}.'.format(
attn_mask.size(), attn.size())
attn.data.masked_fill_(attn_mask, -float('inf'))
attn = self.softmax(attn)
attn = self.dropout(attn)
output = torch.bmm(attn, v)
return output, attn
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4]), torch.rand([4, 4, 4])
]
def get_init_inputs():
return [4]
|
LayerNorm | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/p5/cp56rg7mpbf3ekga7gqadrbxvyrlzzia7mkho3ecsunbsiae7n7a.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# x_1 => add, clone, rsqrt, var_mean
# Graph fragment:
# %clone : [num_users=2] = call_function[target=torch.ops.aten.clone.default](args = (%permute,), kwargs = {memory_format: torch.contiguous_format})
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%clone, [3]), kwargs = {correction: 0, keepdim: True})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-05), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add,), kwargs = {})
triton_poi_fused_native_layer_norm_0 = async_compile.triton('triton_poi_fused_native_layer_norm_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_0(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = (xindex // 16)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (64*x1)), xmask)
tmp1 = tl.load(in_ptr0 + (16 + x0 + (64*x1)), xmask)
tmp3 = tl.load(in_ptr0 + (32 + x0 + (64*x1)), xmask)
tmp5 = tl.load(in_ptr0 + (48 + x0 + (64*x1)), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x2), tmp8, xmask)
tl.store(out_ptr1 + (x2), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/74/c74lpbw2rq4iu6ghsieimw7wwph5kzk2iiojo5f22stwavema7ym.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# x_1 => add, add_1, clone, mul, mul_1, rsqrt, sub, var_mean
# Graph fragment:
# %clone : [num_users=2] = call_function[target=torch.ops.aten.clone.default](args = (%permute,), kwargs = {memory_format: torch.contiguous_format})
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%clone, [3]), kwargs = {correction: 0, keepdim: True})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-05), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%clone, %getitem_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_2), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_3), kwargs = {})
triton_poi_fused_native_layer_norm_1 = async_compile.triton('triton_poi_fused_native_layer_norm_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64, 4], tile_hint=TileHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x3 = xindex
y2 = (yindex // 16)
y4 = yindex % 16
y5 = yindex
y0 = yindex % 4
y1 = (yindex // 4) % 4
tmp0 = tl.load(in_ptr0 + (y4 + (16*x3) + (64*y2)), xmask & ymask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (y5), ymask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (y5), ymask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x3), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x3), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x3 + (4*y1) + (16*y0) + (64*y2)), tmp8, xmask & ymask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 1), (16, 1, 4, 64), torch.float32)
buf1 = empty_strided_cuda((4, 4, 4, 1), (16, 1, 4, 64), torch.float32)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.native_layer_norm]
stream0 = get_raw_stream(0)
triton_poi_fused_native_layer_norm_0.run(primals_1, buf0, buf1, 64, grid=grid(64), stream=stream0)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_1.run(primals_1, buf0, buf1, primals_2, primals_3, buf2, 64, 4, grid=grid(64, 4), stream=stream0)
del buf0
del buf1
del primals_2
del primals_3
return (reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 1, 4, 16), 0), primals_1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.multiprocessing
from torch import nn
from torch.nn import functional as F
import torch.optim
import torch.utils.data
import torch.distributed
class LayerNorm(nn.Module):
def __init__(self, channels: 'int', eps: 'float'=1e-05):
super().__init__()
self.channels = channels
self.eps = eps
self.gamma = nn.Parameter(torch.ones(channels))
self.beta = nn.Parameter(torch.zeros(channels))
def forward(self, x):
x = x.transpose(1, -1)
x = F.layer_norm(x, (self.channels,), self.gamma, self.beta, self.eps)
return x.transpose(1, -1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'channels': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.multiprocessing
from torch import nn
import torch.optim
import torch.utils.data
import torch.distributed
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_native_layer_norm_0(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 16
x1 = xindex // 16
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 64 * x1), xmask)
tmp1 = tl.load(in_ptr0 + (16 + x0 + 64 * x1), xmask)
tmp3 = tl.load(in_ptr0 + (32 + x0 + 64 * x1), xmask)
tmp5 = tl.load(in_ptr0 + (48 + x0 + 64 * x1), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x2, tmp8, xmask)
tl.store(out_ptr1 + x2, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3,
in_ptr4, out_ptr0, ynumel, xnumel, YBLOCK: tl.constexpr, XBLOCK: tl.
constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x3 = xindex
y2 = yindex // 16
y4 = yindex % 16
y5 = yindex
y0 = yindex % 4
y1 = yindex // 4 % 4
tmp0 = tl.load(in_ptr0 + (y4 + 16 * x3 + 64 * y2), xmask & ymask,
eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + y5, ymask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + y5, ymask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x3, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x3, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x3 + 4 * y1 + 16 * y0 + 64 * y2), tmp8, xmask & ymask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 1), (16, 1, 4, 64), torch.float32)
buf1 = empty_strided_cuda((4, 4, 4, 1), (16, 1, 4, 64), torch.float32)
get_raw_stream(0)
triton_poi_fused_native_layer_norm_0[grid(64)](primals_1, buf0,
buf1, 64, XBLOCK=64, num_warps=1, num_stages=1)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_1[grid(64, 4)](primals_1, buf0,
buf1, primals_2, primals_3, buf2, 64, 4, XBLOCK=4, YBLOCK=32,
num_warps=4, num_stages=1)
del buf0
del buf1
del primals_2
del primals_3
return reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 1, 4, 16), 0), primals_1
class LayerNormNew(nn.Module):
def __init__(self, channels: 'int', eps: 'float'=1e-05):
super().__init__()
self.channels = channels
self.eps = eps
self.gamma = nn.Parameter(torch.ones(channels))
self.beta = nn.Parameter(torch.zeros(channels))
def forward(self, input_0):
primals_2 = self.gamma
primals_3 = self.beta
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| mbarnig/vits-train | LayerNorm | false | 7,186 | [
"MIT"
] | 1 | cfb8a0fc91daad868fe3d062ebf85d62edbd7506 | https://github.com/mbarnig/vits-train/tree/cfb8a0fc91daad868fe3d062ebf85d62edbd7506 | import torch
import torch.multiprocessing
from torch import nn
from torch.nn import functional as F
import torch.optim
import torch.utils.data
import torch.distributed
class Model(nn.Module):
def __init__(self, channels: 'int', eps: 'float'=1e-05):
super().__init__()
self.channels = channels
self.eps = eps
self.gamma = nn.Parameter(torch.ones(channels))
self.beta = nn.Parameter(torch.zeros(channels))
def forward(self, x):
x = x.transpose(1, -1)
x = F.layer_norm(x, (self.channels,), self.gamma, self.beta, self.eps)
return x.transpose(1, -1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
AvgPoolShortening | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/yr/cyrgajo5a5abqha3m42zm7eckzjdefosqmg6mj3cpa2rfivywiav.py
# Topologically Sorted Source Nodes: [avg_pool1d], Original ATen: [aten.avg_pool2d]
# Source node to ATen node mapping:
# avg_pool1d => avg_pool2d
# Graph fragment:
# %avg_pool2d : [num_users=1] = call_function[target=torch.ops.aten.avg_pool2d.default](args = (%unsqueeze, [1, 4], [1, 4], [0, 0], True), kwargs = {})
triton_poi_fused_avg_pool2d_0 = async_compile.triton('triton_poi_fused_avg_pool2d_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_avg_pool2d_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_avg_pool2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (16 + x0), xmask)
tmp3 = tl.load(in_ptr0 + (32 + x0), xmask)
tmp5 = tl.load(in_ptr0 + (48 + x0), xmask)
tmp2 = tmp1 + tmp0
tmp4 = tmp3 + tmp2
tmp6 = tmp5 + tmp4
tmp7 = 0.25
tmp8 = tmp6 * tmp7
tl.store(out_ptr0 + (x0), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4), (16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
# Topologically Sorted Source Nodes: [avg_pool1d], Original ATen: [aten.avg_pool2d]
stream0 = get_raw_stream(0)
triton_poi_fused_avg_pool2d_0.run(arg0_1, buf0, 16, grid=grid(16), stream=stream0)
del arg0_1
return (reinterpret_tensor(buf0, (1, 4, 4), (1, 4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class AvgPoolShortening(Module):
"""
### Average pool shortening
This down-samples by a given factor with average pooling
"""
def __init__(self, k: 'int'):
"""
* `k` is the shortening factor
"""
super().__init__()
self.pool = nn.AvgPool1d(k, ceil_mode=True)
def forward(self, x: 'torch.Tensor'):
"""
* `x` is of shape `[seq_len, batch_size, d_model]`
"""
return self.pool(x.permute(1, 2, 0)).permute(2, 0, 1)
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'k': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_avg_pool2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + (16 + x0), xmask)
tmp3 = tl.load(in_ptr0 + (32 + x0), xmask)
tmp5 = tl.load(in_ptr0 + (48 + x0), xmask)
tmp2 = tmp1 + tmp0
tmp4 = tmp3 + tmp2
tmp6 = tmp5 + tmp4
tmp7 = 0.25
tmp8 = tmp6 * tmp7
tl.store(out_ptr0 + x0, tmp8, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4), (16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 1, 1), (4, 1, 16, 16), torch.float32)
get_raw_stream(0)
triton_poi_fused_avg_pool2d_0[grid(16)](arg0_1, buf0, 16, XBLOCK=16,
num_warps=1, num_stages=1)
del arg0_1
return reinterpret_tensor(buf0, (1, 4, 4), (1, 4, 1), 0),
class AvgPoolShorteningNew(Module):
"""
### Average pool shortening
This down-samples by a given factor with average pooling
"""
def __init__(self, k: 'int'):
"""
* `k` is the shortening factor
"""
super().__init__()
self.pool = nn.AvgPool1d(k, ceil_mode=True)
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | AvgPoolShortening | false | 7,187 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
### Average pool shortening
This down-samples by a given factor with average pooling
"""
def __init__(self, k: 'int'):
"""
* `k` is the shortening factor
"""
super().__init__()
self.pool = nn.AvgPool1d(k, ceil_mode=True)
def forward(self, x: 'torch.Tensor'):
"""
* `x` is of shape `[seq_len, batch_size, d_model]`
"""
return self.pool(x.permute(1, 2, 0)).permute(2, 0, 1)
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [4]
|
AttentionNet | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ej/cejfrwnzxinkchwn6symdb72fdtj7gix5hy2vuswodhbeh45mrae.py
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# x => convolution
# x_1 => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
triton_poi_fused_convolution_relu_0 = async_compile.triton('triton_poi_fused_convolution_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1048576],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1048576
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = (xindex // 4096) % 64
tmp0 = tl.load(in_out_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/7z/c7zuih2ysjtir5rh5seep5ijnhokjlgkyjw2edhf257ahvz4iipr.py
# Topologically Sorted Source Nodes: [y2], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# y2 => getitem, getitem_1
# Graph fragment:
# %getitem : [num_users=2] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 0), kwargs = {})
# %getitem_1 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_1 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[262144],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i8', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_1(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 262144
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 32
x1 = (xindex // 32)
x2 = xindex
tmp0 = tl.load(in_ptr0 + ((2*x0) + (128*x1)), None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (2*x0) + (128*x1)), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (64 + (2*x0) + (128*x1)), None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (65 + (2*x0) + (128*x1)), None, eviction_policy='evict_last')
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + (x2), tmp6, None)
tl.store(out_ptr1 + (x2), tmp16, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/xq/cxqz2dr7nh2qabrtemj52pazmhrknj5ltcy32ka252ia6a3jgpqi.py
# Topologically Sorted Source Nodes: [x_4, x_5], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# x_4 => convolution_2
# x_5 => relu_2
# Graph fragment:
# %convolution_2 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%getitem, %primals_6, %primals_7, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_2 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_2,), kwargs = {})
triton_poi_fused_convolution_relu_2 = async_compile.triton('triton_poi_fused_convolution_relu_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[524288],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_2(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 524288
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = (xindex // 1024) % 128
tmp0 = tl.load(in_out_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/pr/cpri5daxkfbmt5ostbhb5o2avircr64a2rmdkxfackaxyjfc7owe.py
# Topologically Sorted Source Nodes: [y2_1], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# y2_1 => getitem_2, getitem_3
# Graph fragment:
# %getitem_2 : [num_users=2] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_1, 0), kwargs = {})
# %getitem_3 : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets_1, 1), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_3 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[131072],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i8', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_3(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 131072
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 16
x1 = (xindex // 16)
x2 = xindex
tmp0 = tl.load(in_ptr0 + ((2*x0) + (64*x1)), None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (2*x0) + (64*x1)), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (32 + (2*x0) + (64*x1)), None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (33 + (2*x0) + (64*x1)), None, eviction_policy='evict_last')
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + (x2), tmp6, None)
tl.store(out_ptr1 + (x2), tmp16, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/of/cof37d5wbqzvtkioj7k4me7wqpvfv55rs62ytonj7gij2o3abnod.py
# Topologically Sorted Source Nodes: [x_8, x_9], Original ATen: [aten.convolution, aten.relu]
# Source node to ATen node mapping:
# x_8 => convolution_4
# x_9 => relu_4
# Graph fragment:
# %convolution_4 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%getitem_2, %primals_10, %primals_11, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_4 : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_4,), kwargs = {})
triton_poi_fused_convolution_relu_4 = async_compile.triton('triton_poi_fused_convolution_relu_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[262144],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_4', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_4(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 262144
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = (xindex // 256) % 256
tmp0 = tl.load(in_out_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x3), tmp4, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/4y/c4yi677k2khitgtsgyfed4k33ti5roqfeskmbxbx6lxxqvdbx2bm.py
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten._to_copy]
# Source node to ATen node mapping:
# x2 => convert_element_type_1
# Graph fragment:
# %convert_element_type_1 : [num_users=5] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%view, torch.int64), kwargs = {})
triton_poi_fused__to_copy_5 = async_compile.triton('triton_poi_fused__to_copy_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_5(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tl.store(out_ptr0 + (x0), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/md/cmdzb3g7ta6wjq6r3fstpbzeaul3ol5l3xra37jbj2vydlmdmvsb.py
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten.add, aten.clamp]
# Source node to ATen node mapping:
# x2 => add_1, clamp_max
# Graph fragment:
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convert_element_type_1, 1), kwargs = {})
# %clamp_max : [num_users=3] = call_function[target=torch.ops.aten.clamp_max.default](args = (%add_1, 15), kwargs = {})
triton_poi_fused_add_clamp_6 = async_compile.triton('triton_poi_fused_add_clamp_6', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_6', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_6(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.full([1], 1, tl.int64)
tmp10 = tmp8 + tmp9
tmp11 = tl.full([1], 15, tl.int64)
tmp12 = triton_helpers.minimum(tmp10, tmp11)
tl.store(out_ptr0 + (x0), tmp12, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/bo/cbopvl3gqicrov7xowxovrkp6amyvpvxr7d7id6h2qyxy5qpmrr2.py
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
# Source node to ATen node mapping:
# x2 => add, clamp_max_2, clamp_min, clamp_min_2, convert_element_type, iota, mul, sub, sub_2
# Graph fragment:
# %iota : [num_users=1] = call_function[target=torch.ops.prims.iota.default](args = (32,), kwargs = {start: 0, step: 1, dtype: torch.int64, device: cuda:0, requires_grad: False})
# %convert_element_type : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%iota, torch.float32), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convert_element_type, 0.5), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add, 0.5), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul, 0.5), kwargs = {})
# %clamp_min : [num_users=3] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub, 0.0), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%clamp_min, %convert_element_type_3), kwargs = {})
# %clamp_min_2 : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub_2, 0.0), kwargs = {})
# %clamp_max_2 : [num_users=3] = call_function[target=torch.ops.aten.clamp_max.default](args = (%clamp_min_2, 1.0), kwargs = {})
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7 = async_compile.triton('triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[32],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tmp8.to(tl.float32)
tmp10 = tmp7 - tmp9
tmp11 = triton_helpers.maximum(tmp10, tmp6)
tmp12 = 1.0
tmp13 = triton_helpers.minimum(tmp11, tmp12)
tl.store(out_ptr0 + (x0), tmp13, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/dp/cdpaggpdddpnqjqbuombdevdldilxunrofiu4bqqmupcln7ot7fy.py
# Topologically Sorted Source Nodes: [x_10, x_11, x2], Original ATen: [aten.convolution, aten.relu, aten._unsafe_index, aten.sub, aten.mul, aten.add]
# Source node to ATen node mapping:
# x2 => _unsafe_index, _unsafe_index_1, _unsafe_index_2, _unsafe_index_3, add_4, add_5, mul_2, mul_3, mul_4, sub_3, sub_4, sub_6
# x_10 => convolution_5
# x_11 => relu_5
# Graph fragment:
# %convolution_5 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_4, %primals_12, %primals_13, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_5 : [num_users=5] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_5,), kwargs = {})
# %_unsafe_index : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_5, [None, None, %convert_element_type_1, %convert_element_type_3]), kwargs = {})
# %_unsafe_index_1 : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_5, [None, None, %convert_element_type_1, %clamp_max_1]), kwargs = {})
# %_unsafe_index_2 : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_5, [None, None, %clamp_max, %convert_element_type_3]), kwargs = {})
# %_unsafe_index_3 : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_5, [None, None, %clamp_max, %clamp_max_1]), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%_unsafe_index_1, %_unsafe_index), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_3, %clamp_max_2), kwargs = {})
# %add_4 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%_unsafe_index, %mul_2), kwargs = {})
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%_unsafe_index_3, %_unsafe_index_2), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_4, %clamp_max_2), kwargs = {})
# %add_5 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%_unsafe_index_2, %mul_3), kwargs = {})
# %sub_6 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_5, %add_4), kwargs = {})
# %mul_4 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_6, %clamp_max_3), kwargs = {})
triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_8 = async_compile.triton('triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_8', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1048576],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: '*i64', 2: '*fp32', 3: '*fp32', 4: '*i64', 5: '*fp32', 6: '*i64', 7: '*fp32', 8: '*fp32', 9: '*fp32', 10: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_8', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 7, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_8(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 1048576
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x1 = (xindex // 32) % 32
x0 = xindex % 32
x5 = (xindex // 1024)
x2 = (xindex // 1024) % 256
x6 = xindex
tmp0 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr1 + (x0), None, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr3 + (x2), None, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr4 + (x0), None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr5 + (x0), None, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr6 + (x1), None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr7 + (x1), None, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 16, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tmp6 = tmp5 + tmp1
tmp7 = tmp5 < 0
tmp8 = tl.where(tmp7, tmp6, tmp5)
tmp9 = tl.load(in_ptr2 + (tmp8 + (16*tmp4) + (256*x5)), None, eviction_policy='evict_last')
tmp11 = tmp9 + tmp10
tmp12 = tl.full([1], 0, tl.int32)
tmp13 = triton_helpers.maximum(tmp12, tmp11)
tmp15 = tmp14 + tmp1
tmp16 = tmp14 < 0
tmp17 = tl.where(tmp16, tmp15, tmp14)
tmp18 = tl.load(in_ptr2 + (tmp17 + (16*tmp4) + (256*x5)), None, eviction_policy='evict_last')
tmp19 = tmp18 + tmp10
tmp20 = triton_helpers.maximum(tmp12, tmp19)
tmp21 = tmp20 - tmp13
tmp23 = tmp21 * tmp22
tmp24 = tmp13 + tmp23
tmp26 = tmp25 + tmp1
tmp27 = tmp25 < 0
tmp28 = tl.where(tmp27, tmp26, tmp25)
tmp29 = tl.load(in_ptr2 + (tmp8 + (16*tmp28) + (256*x5)), None, eviction_policy='evict_last')
tmp30 = tmp29 + tmp10
tmp31 = triton_helpers.maximum(tmp12, tmp30)
tmp32 = tl.load(in_ptr2 + (tmp17 + (16*tmp28) + (256*x5)), None, eviction_policy='evict_last')
tmp33 = tmp32 + tmp10
tmp34 = triton_helpers.maximum(tmp12, tmp33)
tmp35 = tmp34 - tmp31
tmp36 = tmp35 * tmp22
tmp37 = tmp31 + tmp36
tmp38 = tmp37 - tmp24
tmp40 = tmp38 * tmp39
tl.store(out_ptr0 + (x6), tmp24, None)
tl.store(out_ptr1 + (x6), tmp40, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/yd/cydevr7tg7eqnke2m5at7mbt2q2ln7fsm4jiuhcfnhfuhusbiqdn.py
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat => cat
# Graph fragment:
# %cat : [num_users=2] = call_function[target=torch.ops.aten.cat.default](args = ([%relu_3, %add_6], 1), kwargs = {})
triton_poi_fused_cat_9 = async_compile.triton('triton_poi_fused_cat_9', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2097152],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_9', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_9(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1572864
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x1 = (xindex // 1024) % 384
x0 = xindex % 1024
x2 = (xindex // 393216)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 128, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (1024*x1) + (131072*x2)), tmp4, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 384, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr1 + (x0 + (1024*((-128) + x1)) + (262144*x2)), tmp6, other=0.0)
tmp10 = tl.load(in_ptr2 + (x0 + (1024*((-128) + x1)) + (262144*x2)), tmp6, other=0.0)
tmp11 = tmp9 + tmp10
tmp12 = tl.full(tmp11.shape, 0.0, tmp11.dtype)
tmp13 = tl.where(tmp6, tmp11, tmp12)
tmp14 = tl.where(tmp4, tmp5, tmp13)
tl.store(out_ptr0 + (x3), tmp14, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ni/cniq2f6daq5t46e6qe2fkvxb264u2jsxwi4ve5g4nxjnnvswffw3.py
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten._to_copy]
# Source node to ATen node mapping:
# x2_1 => convert_element_type_5
# Graph fragment:
# %convert_element_type_5 : [num_users=5] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%view_2, torch.int64), kwargs = {})
triton_poi_fused__to_copy_10 = async_compile.triton('triton_poi_fused__to_copy_10', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_10', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_10(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tl.store(out_ptr0 + (x0), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/pv/cpv5mon3db2275vzj6vzbbzn5bnpvw33ca7brydombeog7ukrdkt.py
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten.add, aten.clamp]
# Source node to ATen node mapping:
# x2_1 => add_8, clamp_max_4
# Graph fragment:
# %add_8 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convert_element_type_5, 1), kwargs = {})
# %clamp_max_4 : [num_users=3] = call_function[target=torch.ops.aten.clamp_max.default](args = (%add_8, 31), kwargs = {})
triton_poi_fused_add_clamp_11 = async_compile.triton('triton_poi_fused_add_clamp_11', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_11', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_11(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.full([1], 1, tl.int64)
tmp10 = tmp8 + tmp9
tmp11 = tl.full([1], 31, tl.int64)
tmp12 = triton_helpers.minimum(tmp10, tmp11)
tl.store(out_ptr0 + (x0), tmp12, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/nx/cnxxrpmizvmn7bc4n46jide7fktikqdg5ugsihibsemngrexcrrc.py
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
# Source node to ATen node mapping:
# x2_1 => add_7, clamp_max_6, clamp_min_4, clamp_min_6, convert_element_type_4, iota_2, mul_5, sub_7, sub_9
# Graph fragment:
# %iota_2 : [num_users=1] = call_function[target=torch.ops.prims.iota.default](args = (64,), kwargs = {start: 0, step: 1, dtype: torch.int64, device: cuda:0, requires_grad: False})
# %convert_element_type_4 : [num_users=1] = call_function[target=torch.ops.prims.convert_element_type.default](args = (%iota_2, torch.float32), kwargs = {})
# %add_7 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convert_element_type_4, 0.5), kwargs = {})
# %mul_5 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%add_7, 0.5), kwargs = {})
# %sub_7 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul_5, 0.5), kwargs = {})
# %clamp_min_4 : [num_users=3] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub_7, 0.0), kwargs = {})
# %sub_9 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%clamp_min_4, %convert_element_type_7), kwargs = {})
# %clamp_min_6 : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%sub_9, 0.0), kwargs = {})
# %clamp_max_6 : [num_users=3] = call_function[target=torch.ops.aten.clamp_max.default](args = (%clamp_min_6, 1.0), kwargs = {})
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12 = async_compile.triton('triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 0, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12(out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tmp8.to(tl.float32)
tmp10 = tmp7 - tmp9
tmp11 = triton_helpers.maximum(tmp10, tmp6)
tmp12 = 1.0
tmp13 = triton_helpers.minimum(tmp11, tmp12)
tl.store(out_ptr0 + (x0), tmp13, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/wv/cwv7kfzc3llfmzv7nv5r5gayvjdtpwlbwtvpwkn2dmgomldzye3f.py
# Topologically Sorted Source Nodes: [x_14, x_15, x2_1], Original ATen: [aten.convolution, aten.relu, aten._unsafe_index, aten.sub, aten.mul, aten.add]
# Source node to ATen node mapping:
# x2_1 => _unsafe_index_4, _unsafe_index_5, _unsafe_index_6, _unsafe_index_7, add_11, add_12, mul_7, mul_8, mul_9, sub_10, sub_11, sub_13
# x_14 => convolution_7
# x_15 => relu_7
# Graph fragment:
# %convolution_7 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_6, %primals_16, %primals_17, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_7 : [num_users=5] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_7,), kwargs = {})
# %_unsafe_index_4 : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_7, [None, None, %convert_element_type_5, %convert_element_type_7]), kwargs = {})
# %_unsafe_index_5 : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_7, [None, None, %convert_element_type_5, %clamp_max_5]), kwargs = {})
# %_unsafe_index_6 : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_7, [None, None, %clamp_max_4, %convert_element_type_7]), kwargs = {})
# %_unsafe_index_7 : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%relu_7, [None, None, %clamp_max_4, %clamp_max_5]), kwargs = {})
# %sub_10 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%_unsafe_index_5, %_unsafe_index_4), kwargs = {})
# %mul_7 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_10, %clamp_max_6), kwargs = {})
# %add_11 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%_unsafe_index_4, %mul_7), kwargs = {})
# %sub_11 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%_unsafe_index_7, %_unsafe_index_6), kwargs = {})
# %mul_8 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_11, %clamp_max_6), kwargs = {})
# %add_12 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%_unsafe_index_6, %mul_8), kwargs = {})
# %sub_13 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_12, %add_11), kwargs = {})
# %mul_9 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_13, %clamp_max_7), kwargs = {})
triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_13 = async_compile.triton('triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_13', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2097152],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: '*i64', 2: '*fp32', 3: '*fp32', 4: '*i64', 5: '*fp32', 6: '*i64', 7: '*fp32', 8: '*fp32', 9: '*fp32', 10: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_13', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 7, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_13(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 2097152
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x1 = (xindex // 64) % 64
x0 = xindex % 64
x5 = (xindex // 4096)
x2 = (xindex // 4096) % 128
x6 = xindex
tmp0 = tl.load(in_ptr0 + (x1), None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr1 + (x0), None, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr3 + (x2), None, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr4 + (x0), None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr5 + (x0), None, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr6 + (x1), None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr7 + (x1), None, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 32, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tmp6 = tmp5 + tmp1
tmp7 = tmp5 < 0
tmp8 = tl.where(tmp7, tmp6, tmp5)
tmp9 = tl.load(in_ptr2 + (tmp8 + (32*tmp4) + (1024*x5)), None, eviction_policy='evict_last')
tmp11 = tmp9 + tmp10
tmp12 = tl.full([1], 0, tl.int32)
tmp13 = triton_helpers.maximum(tmp12, tmp11)
tmp15 = tmp14 + tmp1
tmp16 = tmp14 < 0
tmp17 = tl.where(tmp16, tmp15, tmp14)
tmp18 = tl.load(in_ptr2 + (tmp17 + (32*tmp4) + (1024*x5)), None, eviction_policy='evict_last')
tmp19 = tmp18 + tmp10
tmp20 = triton_helpers.maximum(tmp12, tmp19)
tmp21 = tmp20 - tmp13
tmp23 = tmp21 * tmp22
tmp24 = tmp13 + tmp23
tmp26 = tmp25 + tmp1
tmp27 = tmp25 < 0
tmp28 = tl.where(tmp27, tmp26, tmp25)
tmp29 = tl.load(in_ptr2 + (tmp8 + (32*tmp28) + (1024*x5)), None, eviction_policy='evict_last')
tmp30 = tmp29 + tmp10
tmp31 = triton_helpers.maximum(tmp12, tmp30)
tmp32 = tl.load(in_ptr2 + (tmp17 + (32*tmp28) + (1024*x5)), None, eviction_policy='evict_last')
tmp33 = tmp32 + tmp10
tmp34 = triton_helpers.maximum(tmp12, tmp33)
tmp35 = tmp34 - tmp31
tmp36 = tmp35 * tmp22
tmp37 = tmp31 + tmp36
tmp38 = tmp37 - tmp24
tmp40 = tmp38 * tmp39
tl.store(out_ptr0 + (x6), tmp24, None)
tl.store(out_ptr1 + (x6), tmp40, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/o2/co2wtxqpyglobhahytbmlui2bz54bijqrhf23q5eqmfyydc4xwjz.py
# Topologically Sorted Source Nodes: [cat_1], Original ATen: [aten.cat]
# Source node to ATen node mapping:
# cat_1 => cat_1
# Graph fragment:
# %cat_1 : [num_users=2] = call_function[target=torch.ops.aten.cat.default](args = ([%relu_1, %add_13], 1), kwargs = {})
triton_poi_fused_cat_14 = async_compile.triton('triton_poi_fused_cat_14', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4194304],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_14', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_cat_14(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 3145728
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x1 = (xindex // 4096) % 192
x0 = xindex % 4096
x2 = (xindex // 786432)
x3 = xindex
tmp0 = x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 64, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + (4096*x1) + (262144*x2)), tmp4, other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1], 192, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr1 + (x0 + (4096*((-64) + x1)) + (524288*x2)), tmp6, other=0.0)
tmp10 = tl.load(in_ptr2 + (x0 + (4096*((-64) + x1)) + (524288*x2)), tmp6, other=0.0)
tmp11 = tmp9 + tmp10
tmp12 = tl.full(tmp11.shape, 0.0, tmp11.dtype)
tmp13 = tl.where(tmp6, tmp11, tmp12)
tmp14 = tl.where(tmp4, tmp5, tmp13)
tl.store(out_ptr0 + (x3), tmp14, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/bq/cbqhd5jnhupbgc2iqkdvblj2dhf7oqvkurq7tlc3hcaievsm4jdp.py
# Topologically Sorted Source Nodes: [x_20], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x_20 => convolution_10
# Graph fragment:
# %convolution_10 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_9, %primals_22, %primals_23, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_15 = async_compile.triton('triton_poi_fused_convolution_15', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16384],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_15', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_15(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16384
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), None)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tl.store(in_out_ptr0 + (x0), tmp3, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/hh/chh3c6ssaz7eobl7f4qcmbxg5zyvo4pizyqmdpko6a2s7pqtslff.py
# Topologically Sorted Source Nodes: [x_14, x_15], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x_14 => convolution_7
# x_15 => relu_7
# Graph fragment:
# %convolution_7 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_6, %primals_16, %primals_17, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_7 : [num_users=5] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_7,), kwargs = {})
# %le_2 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_7, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_16 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_16', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[524288],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_16', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_16(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 524288
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = (xindex // 1024) % 128
tmp0 = tl.load(in_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr1 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x3), tmp6, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/5h/c5hl75ncttq3lk7mlvyxqmzfiikhcuuhomaciav64cpk7egkajir.py
# Topologically Sorted Source Nodes: [x_10, x_11], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x_10 => convolution_5
# x_11 => relu_5
# Graph fragment:
# %convolution_5 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%relu_4, %primals_12, %primals_13, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
# %relu_5 : [num_users=5] = call_function[target=torch.ops.aten.relu.default](args = (%convolution_5,), kwargs = {})
# %le_4 : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu_5, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_17 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_17', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[262144],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_17', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_17(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 262144
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = (xindex // 256) % 256
tmp0 = tl.load(in_ptr0 + (x3), None)
tmp1 = tl.load(in_ptr1 + (x1), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x3), tmp6, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14, primals_15, primals_16, primals_17, primals_18, primals_19, primals_20, primals_21, primals_22, primals_23 = args
args.clear()
assert_size_stride(primals_1, (64, 3, 3, 3), (27, 9, 3, 1))
assert_size_stride(primals_2, (64, ), (1, ))
assert_size_stride(primals_3, (4, 3, 64, 64), (12288, 4096, 64, 1))
assert_size_stride(primals_4, (64, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_5, (64, ), (1, ))
assert_size_stride(primals_6, (128, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_7, (128, ), (1, ))
assert_size_stride(primals_8, (128, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_9, (128, ), (1, ))
assert_size_stride(primals_10, (256, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_11, (256, ), (1, ))
assert_size_stride(primals_12, (256, 256, 3, 3), (2304, 9, 3, 1))
assert_size_stride(primals_13, (256, ), (1, ))
assert_size_stride(primals_14, (128, 384, 3, 3), (3456, 9, 3, 1))
assert_size_stride(primals_15, (128, ), (1, ))
assert_size_stride(primals_16, (128, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_17, (128, ), (1, ))
assert_size_stride(primals_18, (64, 192, 3, 3), (1728, 9, 3, 1))
assert_size_stride(primals_19, (64, ), (1, ))
assert_size_stride(primals_20, (64, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_21, (64, ), (1, ))
assert_size_stride(primals_22, (1, 64, 1, 1), (64, 1, 1, 1))
assert_size_stride(primals_23, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 64, 64, 64), (262144, 4096, 64, 1))
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.convolution, aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_relu_0.run(buf1, primals_2, 1048576, grid=grid(1048576), stream=stream0)
del primals_2
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 64, 64, 64), (262144, 4096, 64, 1))
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [x_2, x_3], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_0.run(buf3, primals_5, 1048576, grid=grid(1048576), stream=stream0)
del primals_5
buf4 = empty_strided_cuda((4, 64, 32, 32), (65536, 1024, 32, 1), torch.float32)
buf5 = empty_strided_cuda((4, 64, 32, 32), (65536, 1024, 32, 1), torch.int8)
# Topologically Sorted Source Nodes: [y2], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_1.run(buf3, buf4, buf5, 262144, grid=grid(262144), stream=stream0)
# Topologically Sorted Source Nodes: [x_4], Original ATen: [aten.convolution]
buf6 = extern_kernels.convolution(buf4, primals_6, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf6, (4, 128, 32, 32), (131072, 1024, 32, 1))
buf7 = buf6; del buf6 # reuse
# Topologically Sorted Source Nodes: [x_4, x_5], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_2.run(buf7, primals_7, 524288, grid=grid(524288), stream=stream0)
del primals_7
# Topologically Sorted Source Nodes: [x_6], Original ATen: [aten.convolution]
buf8 = extern_kernels.convolution(buf7, primals_8, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf8, (4, 128, 32, 32), (131072, 1024, 32, 1))
buf9 = buf8; del buf8 # reuse
# Topologically Sorted Source Nodes: [x_6, x_7], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_2.run(buf9, primals_9, 524288, grid=grid(524288), stream=stream0)
del primals_9
buf10 = empty_strided_cuda((4, 128, 16, 16), (32768, 256, 16, 1), torch.float32)
buf11 = empty_strided_cuda((4, 128, 16, 16), (32768, 256, 16, 1), torch.int8)
# Topologically Sorted Source Nodes: [y2_1], Original ATen: [aten.max_pool2d_with_indices]
triton_poi_fused_max_pool2d_with_indices_3.run(buf9, buf10, buf11, 131072, grid=grid(131072), stream=stream0)
# Topologically Sorted Source Nodes: [x_8], Original ATen: [aten.convolution]
buf12 = extern_kernels.convolution(buf10, primals_10, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf12, (4, 256, 16, 16), (65536, 256, 16, 1))
buf13 = buf12; del buf12 # reuse
# Topologically Sorted Source Nodes: [x_8, x_9], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_4.run(buf13, primals_11, 262144, grid=grid(262144), stream=stream0)
del primals_11
# Topologically Sorted Source Nodes: [x_10], Original ATen: [aten.convolution]
buf14 = extern_kernels.convolution(buf13, primals_12, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf14, (4, 256, 16, 16), (65536, 256, 16, 1))
buf15 = empty_strided_cuda((32, 1), (1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten._to_copy]
triton_poi_fused__to_copy_5.run(buf15, 32, grid=grid(32), stream=stream0)
buf16 = empty_strided_cuda((32, 1), (1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten.add, aten.clamp]
triton_poi_fused_add_clamp_6.run(buf16, 32, grid=grid(32), stream=stream0)
buf17 = empty_strided_cuda((32, ), (1, ), torch.int64)
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
triton_poi_fused__to_copy_5.run(buf17, 32, grid=grid(32), stream=stream0)
buf18 = empty_strided_cuda((32, ), (1, ), torch.int64)
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten.add, aten.clamp]
triton_poi_fused_add_clamp_6.run(buf18, 32, grid=grid(32), stream=stream0)
buf19 = empty_strided_cuda((32, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7.run(buf19, 32, grid=grid(32), stream=stream0)
buf21 = empty_strided_cuda((32, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [x2], Original ATen: [aten.sub, aten.clamp]
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7.run(buf21, 32, grid=grid(32), stream=stream0)
buf20 = empty_strided_cuda((4, 256, 32, 32), (262144, 1024, 32, 1), torch.float32)
buf22 = empty_strided_cuda((4, 256, 32, 32), (262144, 1024, 32, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_10, x_11, x2], Original ATen: [aten.convolution, aten.relu, aten._unsafe_index, aten.sub, aten.mul, aten.add]
triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_8.run(buf15, buf17, buf14, primals_13, buf18, buf19, buf16, buf21, buf20, buf22, 1048576, grid=grid(1048576), stream=stream0)
buf23 = empty_strided_cuda((4, 384, 32, 32), (393216, 1024, 32, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat], Original ATen: [aten.cat]
triton_poi_fused_cat_9.run(buf9, buf20, buf22, buf23, 1572864, grid=grid(1572864), stream=stream0)
del buf20
del buf22
# Topologically Sorted Source Nodes: [x_12], Original ATen: [aten.convolution]
buf24 = extern_kernels.convolution(buf23, primals_14, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf24, (4, 128, 32, 32), (131072, 1024, 32, 1))
buf25 = buf24; del buf24 # reuse
# Topologically Sorted Source Nodes: [x_12, x_13], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_2.run(buf25, primals_15, 524288, grid=grid(524288), stream=stream0)
del primals_15
# Topologically Sorted Source Nodes: [x_14], Original ATen: [aten.convolution]
buf26 = extern_kernels.convolution(buf25, primals_16, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf26, (4, 128, 32, 32), (131072, 1024, 32, 1))
buf27 = empty_strided_cuda((64, 1), (1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten._to_copy]
triton_poi_fused__to_copy_10.run(buf27, 64, grid=grid(64), stream=stream0)
buf28 = empty_strided_cuda((64, 1), (1, 1), torch.int64)
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten.add, aten.clamp]
triton_poi_fused_add_clamp_11.run(buf28, 64, grid=grid(64), stream=stream0)
buf29 = empty_strided_cuda((64, ), (1, ), torch.int64)
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
triton_poi_fused__to_copy_10.run(buf29, 64, grid=grid(64), stream=stream0)
buf30 = empty_strided_cuda((64, ), (1, ), torch.int64)
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten.add, aten.clamp]
triton_poi_fused_add_clamp_11.run(buf30, 64, grid=grid(64), stream=stream0)
buf31 = empty_strided_cuda((64, ), (1, ), torch.float32)
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten.arange, aten._to_copy, aten.add, aten.mul, aten.sub, aten.clamp]
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12.run(buf31, 64, grid=grid(64), stream=stream0)
buf33 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [x2_1], Original ATen: [aten.sub, aten.clamp]
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12.run(buf33, 64, grid=grid(64), stream=stream0)
buf32 = empty_strided_cuda((4, 128, 64, 64), (524288, 4096, 64, 1), torch.float32)
buf34 = empty_strided_cuda((4, 128, 64, 64), (524288, 4096, 64, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_14, x_15, x2_1], Original ATen: [aten.convolution, aten.relu, aten._unsafe_index, aten.sub, aten.mul, aten.add]
triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_13.run(buf27, buf29, buf26, primals_17, buf30, buf31, buf28, buf33, buf32, buf34, 2097152, grid=grid(2097152), stream=stream0)
buf35 = empty_strided_cuda((4, 192, 64, 64), (786432, 4096, 64, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat_1], Original ATen: [aten.cat]
triton_poi_fused_cat_14.run(buf3, buf32, buf34, buf35, 3145728, grid=grid(3145728), stream=stream0)
del buf32
del buf34
# Topologically Sorted Source Nodes: [x_16], Original ATen: [aten.convolution]
buf36 = extern_kernels.convolution(buf35, primals_18, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf36, (4, 64, 64, 64), (262144, 4096, 64, 1))
buf37 = buf36; del buf36 # reuse
# Topologically Sorted Source Nodes: [x_16, x_17], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_0.run(buf37, primals_19, 1048576, grid=grid(1048576), stream=stream0)
del primals_19
# Topologically Sorted Source Nodes: [x_18], Original ATen: [aten.convolution]
buf38 = extern_kernels.convolution(buf37, primals_20, stride=(1, 1), padding=(1, 1), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf38, (4, 64, 64, 64), (262144, 4096, 64, 1))
buf39 = buf38; del buf38 # reuse
# Topologically Sorted Source Nodes: [x_18, x_19], Original ATen: [aten.convolution, aten.relu]
triton_poi_fused_convolution_relu_0.run(buf39, primals_21, 1048576, grid=grid(1048576), stream=stream0)
del primals_21
# Topologically Sorted Source Nodes: [x_20], Original ATen: [aten.convolution]
buf40 = extern_kernels.convolution(buf39, primals_22, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf40, (4, 1, 64, 64), (4096, 4096, 64, 1))
buf41 = buf40; del buf40 # reuse
# Topologically Sorted Source Nodes: [x_20], Original ATen: [aten.convolution]
triton_poi_fused_convolution_15.run(buf41, primals_23, 16384, grid=grid(16384), stream=stream0)
del primals_23
buf42 = empty_strided_cuda((4, 128, 32, 32), (131072, 1024, 32, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_14, x_15], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_16.run(buf26, primals_17, buf42, 524288, grid=grid(524288), stream=stream0)
del buf26
del primals_17
buf43 = empty_strided_cuda((4, 256, 16, 16), (65536, 256, 16, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_10, x_11], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_17.run(buf14, primals_13, buf43, 262144, grid=grid(262144), stream=stream0)
del buf14
del primals_13
return (buf41, primals_1, primals_3, primals_4, primals_6, primals_8, primals_10, primals_12, primals_14, primals_16, primals_18, primals_20, primals_22, buf1, buf3, buf4, buf5, buf7, buf9, buf10, buf11, buf13, buf15, buf16, buf17, buf18, buf19, buf21, buf23, buf25, buf27, buf28, buf29, buf30, buf31, buf33, buf35, buf37, buf39, buf42, buf43, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((64, 3, 3, 3), (27, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 3, 64, 64), (12288, 4096, 64, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((64, 64, 3, 3), (576, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((128, 64, 3, 3), (576, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((128, 128, 3, 3), (1152, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((256, 128, 3, 3), (1152, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((256, 256, 3, 3), (2304, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((256, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_14 = rand_strided((128, 384, 3, 3), (3456, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_15 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_16 = rand_strided((128, 128, 3, 3), (1152, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_17 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_18 = rand_strided((64, 192, 3, 3), (1728, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_19 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_20 = rand_strided((64, 64, 3, 3), (576, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_21 = rand_strided((64, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_22 = rand_strided((1, 64, 1, 1), (64, 1, 1, 1), device='cuda:0', dtype=torch.float32)
primals_23 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13, primals_14, primals_15, primals_16, primals_17, primals_18, primals_19, primals_20, primals_21, primals_22, primals_23])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.nn.functional
def conv3x3(in_, out):
return nn.Conv2d(in_, out, 3, padding=1)
class ConvRelu(nn.Module):
def __init__(self, in_, out):
super().__init__()
self.conv = conv3x3(in_, out)
self.activation = nn.ReLU(inplace=True)
def forward(self, x):
x = self.conv(x)
x = self.activation(x)
return x
class ConvRelu2(nn.Module):
def __init__(self, _in, _out):
super(ConvRelu2, self).__init__()
self.cr1 = ConvRelu(_in, _out)
self.cr2 = ConvRelu(_out, _out)
def forward(self, x):
x = self.cr1(x)
x = self.cr2(x)
return x
class Coder(nn.Module):
def __init__(self, in_size, out_size):
super(Coder, self).__init__()
self.conv = ConvRelu2(in_size, out_size)
self.down = nn.MaxPool2d(2, 2)
def forward(self, x):
y1 = self.conv(x)
y2 = self.down(y1)
return y2, y1
class Decoder(nn.Module):
def __init__(self, in_size, out_size):
super(Decoder, self).__init__()
self.conv = ConvRelu2(in_size, out_size)
self.up = F.interpolate
def forward(self, x1, x2):
x2 = self.up(x2, scale_factor=2, mode='bilinear', align_corners=False)
return self.conv(torch.cat([x1, x2], 1))
class AttentionNet(nn.Module):
def __init__(self, in_channels=3, out_channels=1):
super(AttentionNet, self).__init__()
self.in_channels = in_channels
self.out_channels = out_channels
filters = [64, 128, 256]
self.down1 = Coder(in_channels, filters[0])
self.down2 = Coder(filters[0], filters[1])
self.center = ConvRelu2(filters[1], filters[2])
self.up2 = Decoder(filters[2] + filters[1], filters[1])
self.up1 = Decoder(filters[1] + filters[0], filters[0])
self.final = nn.Conv2d(filters[0], out_channels, 1)
def forward(self, x):
x, befdown1 = self.down1(x)
x, befdown2 = self.down2(x)
x = self.center(x)
x = self.up2(befdown2, x)
x = self.up1(befdown1, x)
x = self.final(x)
return x
def get_inputs():
return [torch.rand([4, 3, 64, 64])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.nn.functional as F
import torch.nn.functional
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_convolution_relu_0(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = xindex // 4096 % 64
tmp0 = tl.load(in_out_ptr0 + x3, None)
tmp1 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, None)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_1(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 32
x1 = xindex // 32
x2 = xindex
tmp0 = tl.load(in_ptr0 + (2 * x0 + 128 * x1), None, eviction_policy=
'evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 2 * x0 + 128 * x1), None, eviction_policy
='evict_last')
tmp3 = tl.load(in_ptr0 + (64 + 2 * x0 + 128 * x1), None,
eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (65 + 2 * x0 + 128 * x1), None,
eviction_policy='evict_last')
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + x2, tmp6, None)
tl.store(out_ptr1 + x2, tmp16, None)
@triton.jit
def triton_poi_fused_convolution_relu_2(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = xindex // 1024 % 128
tmp0 = tl.load(in_out_ptr0 + x3, None)
tmp1 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, None)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_3(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x0 = xindex % 16
x1 = xindex // 16
x2 = xindex
tmp0 = tl.load(in_ptr0 + (2 * x0 + 64 * x1), None, eviction_policy=
'evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 2 * x0 + 64 * x1), None, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (32 + 2 * x0 + 64 * x1), None, eviction_policy
='evict_last')
tmp5 = tl.load(in_ptr0 + (33 + 2 * x0 + 64 * x1), None, eviction_policy
='evict_last')
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp6 = triton_helpers.maximum(tmp5, tmp4)
tmp7 = tmp1 > tmp0
tmp8 = tl.full([1], 1, tl.int8)
tmp9 = tl.full([1], 0, tl.int8)
tmp10 = tl.where(tmp7, tmp8, tmp9)
tmp11 = tmp3 > tmp2
tmp12 = tl.full([1], 2, tl.int8)
tmp13 = tl.where(tmp11, tmp12, tmp10)
tmp14 = tmp5 > tmp4
tmp15 = tl.full([1], 3, tl.int8)
tmp16 = tl.where(tmp14, tmp15, tmp13)
tl.store(out_ptr0 + x2, tmp6, None)
tl.store(out_ptr1 + x2, tmp16, None)
@triton.jit
def triton_poi_fused_convolution_relu_4(in_out_ptr0, in_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = xindex // 256 % 256
tmp0 = tl.load(in_out_ptr0 + x3, None)
tmp1 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x3, tmp4, None)
@triton.jit
def triton_poi_fused__to_copy_5(out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tl.store(out_ptr0 + x0, tmp8, xmask)
@triton.jit
def triton_poi_fused_add_clamp_6(out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.full([1], 1, tl.int64)
tmp10 = tmp8 + tmp9
tmp11 = tl.full([1], 15, tl.int64)
tmp12 = triton_helpers.minimum(tmp10, tmp11)
tl.store(out_ptr0 + x0, tmp12, xmask)
@triton.jit
def triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7(out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 32
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tmp8.to(tl.float32)
tmp10 = tmp7 - tmp9
tmp11 = triton_helpers.maximum(tmp10, tmp6)
tmp12 = 1.0
tmp13 = triton_helpers.minimum(tmp11, tmp12)
tl.store(out_ptr0 + x0, tmp13, xmask)
@triton.jit
def triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_8(in_ptr0,
in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x1 = xindex // 32 % 32
x0 = xindex % 32
x5 = xindex // 1024
x2 = xindex // 1024 % 256
x6 = xindex
tmp0 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr1 + x0, None, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr3 + x2, None, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr4 + x0, None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr5 + x0, None, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr6 + x1, None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr7 + x1, None, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 16, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tmp6 = tmp5 + tmp1
tmp7 = tmp5 < 0
tmp8 = tl.where(tmp7, tmp6, tmp5)
tmp9 = tl.load(in_ptr2 + (tmp8 + 16 * tmp4 + 256 * x5), None,
eviction_policy='evict_last')
tmp11 = tmp9 + tmp10
tmp12 = tl.full([1], 0, tl.int32)
tmp13 = triton_helpers.maximum(tmp12, tmp11)
tmp15 = tmp14 + tmp1
tmp16 = tmp14 < 0
tmp17 = tl.where(tmp16, tmp15, tmp14)
tmp18 = tl.load(in_ptr2 + (tmp17 + 16 * tmp4 + 256 * x5), None,
eviction_policy='evict_last')
tmp19 = tmp18 + tmp10
tmp20 = triton_helpers.maximum(tmp12, tmp19)
tmp21 = tmp20 - tmp13
tmp23 = tmp21 * tmp22
tmp24 = tmp13 + tmp23
tmp26 = tmp25 + tmp1
tmp27 = tmp25 < 0
tmp28 = tl.where(tmp27, tmp26, tmp25)
tmp29 = tl.load(in_ptr2 + (tmp8 + 16 * tmp28 + 256 * x5), None,
eviction_policy='evict_last')
tmp30 = tmp29 + tmp10
tmp31 = triton_helpers.maximum(tmp12, tmp30)
tmp32 = tl.load(in_ptr2 + (tmp17 + 16 * tmp28 + 256 * x5), None,
eviction_policy='evict_last')
tmp33 = tmp32 + tmp10
tmp34 = triton_helpers.maximum(tmp12, tmp33)
tmp35 = tmp34 - tmp31
tmp36 = tmp35 * tmp22
tmp37 = tmp31 + tmp36
tmp38 = tmp37 - tmp24
tmp40 = tmp38 * tmp39
tl.store(out_ptr0 + x6, tmp24, None)
tl.store(out_ptr1 + x6, tmp40, None)
@triton.jit
def triton_poi_fused_cat_9(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x1 = xindex // 1024 % 384
x0 = xindex % 1024
x2 = xindex // 393216
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 128, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 1024 * x1 + 131072 * x2), tmp4, other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1], 384, tl.int64)
tmp9 = tl.load(in_ptr1 + (x0 + 1024 * (-128 + x1) + 262144 * x2), tmp6,
other=0.0)
tmp10 = tl.load(in_ptr2 + (x0 + 1024 * (-128 + x1) + 262144 * x2), tmp6,
other=0.0)
tmp11 = tmp9 + tmp10
tmp12 = tl.full(tmp11.shape, 0.0, tmp11.dtype)
tmp13 = tl.where(tmp6, tmp11, tmp12)
tmp14 = tl.where(tmp4, tmp5, tmp13)
tl.store(out_ptr0 + x3, tmp14, None)
@triton.jit
def triton_poi_fused__to_copy_10(out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tl.store(out_ptr0 + x0, tmp8, xmask)
@triton.jit
def triton_poi_fused_add_clamp_11(out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tl.full([1], 1, tl.int64)
tmp10 = tmp8 + tmp9
tmp11 = tl.full([1], 31, tl.int64)
tmp12 = triton_helpers.minimum(tmp10, tmp11)
tl.store(out_ptr0 + x0, tmp12, xmask)
@triton.jit
def triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12(out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = x0
tmp1 = tmp0.to(tl.float32)
tmp2 = 0.5
tmp3 = tmp1 + tmp2
tmp4 = tmp3 * tmp2
tmp5 = tmp4 - tmp2
tmp6 = 0.0
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp7.to(tl.int32)
tmp9 = tmp8.to(tl.float32)
tmp10 = tmp7 - tmp9
tmp11 = triton_helpers.maximum(tmp10, tmp6)
tmp12 = 1.0
tmp13 = triton_helpers.minimum(tmp11, tmp12)
tl.store(out_ptr0 + x0, tmp13, xmask)
@triton.jit
def triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_13(in_ptr0,
in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x1 = xindex // 64 % 64
x0 = xindex % 64
x5 = xindex // 4096
x2 = xindex // 4096 % 128
x6 = xindex
tmp0 = tl.load(in_ptr0 + x1, None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr1 + x0, None, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr3 + x2, None, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr4 + x0, None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr5 + x0, None, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr6 + x1, None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr7 + x1, None, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 32, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tmp6 = tmp5 + tmp1
tmp7 = tmp5 < 0
tmp8 = tl.where(tmp7, tmp6, tmp5)
tmp9 = tl.load(in_ptr2 + (tmp8 + 32 * tmp4 + 1024 * x5), None,
eviction_policy='evict_last')
tmp11 = tmp9 + tmp10
tmp12 = tl.full([1], 0, tl.int32)
tmp13 = triton_helpers.maximum(tmp12, tmp11)
tmp15 = tmp14 + tmp1
tmp16 = tmp14 < 0
tmp17 = tl.where(tmp16, tmp15, tmp14)
tmp18 = tl.load(in_ptr2 + (tmp17 + 32 * tmp4 + 1024 * x5), None,
eviction_policy='evict_last')
tmp19 = tmp18 + tmp10
tmp20 = triton_helpers.maximum(tmp12, tmp19)
tmp21 = tmp20 - tmp13
tmp23 = tmp21 * tmp22
tmp24 = tmp13 + tmp23
tmp26 = tmp25 + tmp1
tmp27 = tmp25 < 0
tmp28 = tl.where(tmp27, tmp26, tmp25)
tmp29 = tl.load(in_ptr2 + (tmp8 + 32 * tmp28 + 1024 * x5), None,
eviction_policy='evict_last')
tmp30 = tmp29 + tmp10
tmp31 = triton_helpers.maximum(tmp12, tmp30)
tmp32 = tl.load(in_ptr2 + (tmp17 + 32 * tmp28 + 1024 * x5), None,
eviction_policy='evict_last')
tmp33 = tmp32 + tmp10
tmp34 = triton_helpers.maximum(tmp12, tmp33)
tmp35 = tmp34 - tmp31
tmp36 = tmp35 * tmp22
tmp37 = tmp31 + tmp36
tmp38 = tmp37 - tmp24
tmp40 = tmp38 * tmp39
tl.store(out_ptr0 + x6, tmp24, None)
tl.store(out_ptr1 + x6, tmp40, None)
@triton.jit
def triton_poi_fused_cat_14(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x1 = xindex // 4096 % 192
x0 = xindex % 4096
x2 = xindex // 786432
x3 = xindex
tmp0 = x1
tl.full([1], 0, tl.int64)
tmp3 = tl.full([1], 64, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (x0 + 4096 * x1 + 262144 * x2), tmp4, other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1], 192, tl.int64)
tmp9 = tl.load(in_ptr1 + (x0 + 4096 * (-64 + x1) + 524288 * x2), tmp6,
other=0.0)
tmp10 = tl.load(in_ptr2 + (x0 + 4096 * (-64 + x1) + 524288 * x2), tmp6,
other=0.0)
tmp11 = tmp9 + tmp10
tmp12 = tl.full(tmp11.shape, 0.0, tmp11.dtype)
tmp13 = tl.where(tmp6, tmp11, tmp12)
tmp14 = tl.where(tmp4, tmp5, tmp13)
tl.store(out_ptr0 + x3, tmp14, None)
@triton.jit
def triton_poi_fused_convolution_15(in_out_ptr0, in_ptr0, xnumel, XBLOCK:
tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, None)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tl.store(in_out_ptr0 + x0, tmp3, None)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_16(in_ptr0,
in_ptr1, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = xindex // 1024 % 128
tmp0 = tl.load(in_ptr0 + x3, None)
tmp1 = tl.load(in_ptr1 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + x3, tmp6, None)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_17(in_ptr0,
in_ptr1, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x3 = xindex
x1 = xindex // 256 % 256
tmp0 = tl.load(in_ptr0 + x3, None)
tmp1 = tl.load(in_ptr1 + x1, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + x3, tmp6, None)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13, primals_14, primals_15, primals_16, primals_17,
primals_18, primals_19, primals_20, primals_21, primals_22, primals_23
) = args
args.clear()
assert_size_stride(primals_1, (64, 3, 3, 3), (27, 9, 3, 1))
assert_size_stride(primals_2, (64,), (1,))
assert_size_stride(primals_3, (4, 3, 64, 64), (12288, 4096, 64, 1))
assert_size_stride(primals_4, (64, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_5, (64,), (1,))
assert_size_stride(primals_6, (128, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_7, (128,), (1,))
assert_size_stride(primals_8, (128, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_9, (128,), (1,))
assert_size_stride(primals_10, (256, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_11, (256,), (1,))
assert_size_stride(primals_12, (256, 256, 3, 3), (2304, 9, 3, 1))
assert_size_stride(primals_13, (256,), (1,))
assert_size_stride(primals_14, (128, 384, 3, 3), (3456, 9, 3, 1))
assert_size_stride(primals_15, (128,), (1,))
assert_size_stride(primals_16, (128, 128, 3, 3), (1152, 9, 3, 1))
assert_size_stride(primals_17, (128,), (1,))
assert_size_stride(primals_18, (64, 192, 3, 3), (1728, 9, 3, 1))
assert_size_stride(primals_19, (64,), (1,))
assert_size_stride(primals_20, (64, 64, 3, 3), (576, 9, 3, 1))
assert_size_stride(primals_21, (64,), (1,))
assert_size_stride(primals_22, (1, 64, 1, 1), (64, 1, 1, 1))
assert_size_stride(primals_23, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,
1), padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 64, 64, 64), (262144, 4096, 64, 1))
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_relu_0[grid(1048576)](buf1, primals_2,
1048576, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_2
buf2 = extern_kernels.convolution(buf1, primals_4, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf2, (4, 64, 64, 64), (262144, 4096, 64, 1))
buf3 = buf2
del buf2
triton_poi_fused_convolution_relu_0[grid(1048576)](buf3, primals_5,
1048576, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_5
buf4 = empty_strided_cuda((4, 64, 32, 32), (65536, 1024, 32, 1),
torch.float32)
buf5 = empty_strided_cuda((4, 64, 32, 32), (65536, 1024, 32, 1),
torch.int8)
triton_poi_fused_max_pool2d_with_indices_1[grid(262144)](buf3, buf4,
buf5, 262144, XBLOCK=512, num_warps=8, num_stages=1)
buf6 = extern_kernels.convolution(buf4, primals_6, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf6, (4, 128, 32, 32), (131072, 1024, 32, 1))
buf7 = buf6
del buf6
triton_poi_fused_convolution_relu_2[grid(524288)](buf7, primals_7,
524288, XBLOCK=512, num_warps=8, num_stages=1)
del primals_7
buf8 = extern_kernels.convolution(buf7, primals_8, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf8, (4, 128, 32, 32), (131072, 1024, 32, 1))
buf9 = buf8
del buf8
triton_poi_fused_convolution_relu_2[grid(524288)](buf9, primals_9,
524288, XBLOCK=512, num_warps=8, num_stages=1)
del primals_9
buf10 = empty_strided_cuda((4, 128, 16, 16), (32768, 256, 16, 1),
torch.float32)
buf11 = empty_strided_cuda((4, 128, 16, 16), (32768, 256, 16, 1),
torch.int8)
triton_poi_fused_max_pool2d_with_indices_3[grid(131072)](buf9,
buf10, buf11, 131072, XBLOCK=512, num_warps=8, num_stages=1)
buf12 = extern_kernels.convolution(buf10, primals_10, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf12, (4, 256, 16, 16), (65536, 256, 16, 1))
buf13 = buf12
del buf12
triton_poi_fused_convolution_relu_4[grid(262144)](buf13, primals_11,
262144, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_11
buf14 = extern_kernels.convolution(buf13, primals_12, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf14, (4, 256, 16, 16), (65536, 256, 16, 1))
buf15 = empty_strided_cuda((32, 1), (1, 1), torch.int64)
triton_poi_fused__to_copy_5[grid(32)](buf15, 32, XBLOCK=32,
num_warps=1, num_stages=1)
buf16 = empty_strided_cuda((32, 1), (1, 1), torch.int64)
triton_poi_fused_add_clamp_6[grid(32)](buf16, 32, XBLOCK=32,
num_warps=1, num_stages=1)
buf17 = empty_strided_cuda((32,), (1,), torch.int64)
triton_poi_fused__to_copy_5[grid(32)](buf17, 32, XBLOCK=32,
num_warps=1, num_stages=1)
buf18 = empty_strided_cuda((32,), (1,), torch.int64)
triton_poi_fused_add_clamp_6[grid(32)](buf18, 32, XBLOCK=32,
num_warps=1, num_stages=1)
buf19 = empty_strided_cuda((32,), (1,), torch.float32)
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7[grid(32)](buf19,
32, XBLOCK=32, num_warps=1, num_stages=1)
buf21 = empty_strided_cuda((32, 1), (1, 1), torch.float32)
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_7[grid(32)](buf21,
32, XBLOCK=32, num_warps=1, num_stages=1)
buf20 = empty_strided_cuda((4, 256, 32, 32), (262144, 1024, 32, 1),
torch.float32)
buf22 = empty_strided_cuda((4, 256, 32, 32), (262144, 1024, 32, 1),
torch.float32)
triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_8[grid(
1048576)](buf15, buf17, buf14, primals_13, buf18, buf19, buf16,
buf21, buf20, buf22, 1048576, XBLOCK=1024, num_warps=4,
num_stages=1)
buf23 = empty_strided_cuda((4, 384, 32, 32), (393216, 1024, 32, 1),
torch.float32)
triton_poi_fused_cat_9[grid(1572864)](buf9, buf20, buf22, buf23,
1572864, XBLOCK=1024, num_warps=4, num_stages=1)
del buf20
del buf22
buf24 = extern_kernels.convolution(buf23, primals_14, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf24, (4, 128, 32, 32), (131072, 1024, 32, 1))
buf25 = buf24
del buf24
triton_poi_fused_convolution_relu_2[grid(524288)](buf25, primals_15,
524288, XBLOCK=512, num_warps=8, num_stages=1)
del primals_15
buf26 = extern_kernels.convolution(buf25, primals_16, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf26, (4, 128, 32, 32), (131072, 1024, 32, 1))
buf27 = empty_strided_cuda((64, 1), (1, 1), torch.int64)
triton_poi_fused__to_copy_10[grid(64)](buf27, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf28 = empty_strided_cuda((64, 1), (1, 1), torch.int64)
triton_poi_fused_add_clamp_11[grid(64)](buf28, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf29 = empty_strided_cuda((64,), (1,), torch.int64)
triton_poi_fused__to_copy_10[grid(64)](buf29, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf30 = empty_strided_cuda((64,), (1,), torch.int64)
triton_poi_fused_add_clamp_11[grid(64)](buf30, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf31 = empty_strided_cuda((64,), (1,), torch.float32)
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12[grid(64)](buf31,
64, XBLOCK=64, num_warps=1, num_stages=1)
buf33 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
triton_poi_fused__to_copy_add_arange_clamp_mul_sub_12[grid(64)](buf33,
64, XBLOCK=64, num_warps=1, num_stages=1)
buf32 = empty_strided_cuda((4, 128, 64, 64), (524288, 4096, 64, 1),
torch.float32)
buf34 = empty_strided_cuda((4, 128, 64, 64), (524288, 4096, 64, 1),
torch.float32)
triton_poi_fused__unsafe_index_add_convolution_mul_relu_sub_13[grid
(2097152)](buf27, buf29, buf26, primals_17, buf30, buf31, buf28,
buf33, buf32, buf34, 2097152, XBLOCK=512, num_warps=8, num_stages=1
)
buf35 = empty_strided_cuda((4, 192, 64, 64), (786432, 4096, 64, 1),
torch.float32)
triton_poi_fused_cat_14[grid(3145728)](buf3, buf32, buf34, buf35,
3145728, XBLOCK=1024, num_warps=4, num_stages=1)
del buf32
del buf34
buf36 = extern_kernels.convolution(buf35, primals_18, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf36, (4, 64, 64, 64), (262144, 4096, 64, 1))
buf37 = buf36
del buf36
triton_poi_fused_convolution_relu_0[grid(1048576)](buf37,
primals_19, 1048576, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_19
buf38 = extern_kernels.convolution(buf37, primals_20, stride=(1, 1),
padding=(1, 1), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf38, (4, 64, 64, 64), (262144, 4096, 64, 1))
buf39 = buf38
del buf38
triton_poi_fused_convolution_relu_0[grid(1048576)](buf39,
primals_21, 1048576, XBLOCK=1024, num_warps=4, num_stages=1)
del primals_21
buf40 = extern_kernels.convolution(buf39, primals_22, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf40, (4, 1, 64, 64), (4096, 4096, 64, 1))
buf41 = buf40
del buf40
triton_poi_fused_convolution_15[grid(16384)](buf41, primals_23,
16384, XBLOCK=256, num_warps=4, num_stages=1)
del primals_23
buf42 = empty_strided_cuda((4, 128, 32, 32), (131072, 1024, 32, 1),
torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_16[grid(524288)](
buf26, primals_17, buf42, 524288, XBLOCK=512, num_warps=8,
num_stages=1)
del buf26
del primals_17
buf43 = empty_strided_cuda((4, 256, 16, 16), (65536, 256, 16, 1),
torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_17[grid(262144)](
buf14, primals_13, buf43, 262144, XBLOCK=1024, num_warps=4,
num_stages=1)
del buf14
del primals_13
return (buf41, primals_1, primals_3, primals_4, primals_6, primals_8,
primals_10, primals_12, primals_14, primals_16, primals_18,
primals_20, primals_22, buf1, buf3, buf4, buf5, buf7, buf9, buf10,
buf11, buf13, buf15, buf16, buf17, buf18, buf19, buf21, buf23,
buf25, buf27, buf28, buf29, buf30, buf31, buf33, buf35, buf37,
buf39, buf42, buf43)
def conv3x3(in_, out):
return nn.Conv2d(in_, out, 3, padding=1)
class ConvRelu(nn.Module):
def __init__(self, in_, out):
super().__init__()
self.conv = conv3x3(in_, out)
self.activation = nn.ReLU(inplace=True)
def forward(self, x):
x = self.conv(x)
x = self.activation(x)
return x
class ConvRelu2(nn.Module):
def __init__(self, _in, _out):
super(ConvRelu2, self).__init__()
self.cr1 = ConvRelu(_in, _out)
self.cr2 = ConvRelu(_out, _out)
def forward(self, x):
x = self.cr1(x)
x = self.cr2(x)
return x
class Coder(nn.Module):
def __init__(self, in_size, out_size):
super(Coder, self).__init__()
self.conv = ConvRelu2(in_size, out_size)
self.down = nn.MaxPool2d(2, 2)
def forward(self, x):
y1 = self.conv(x)
y2 = self.down(y1)
return y2, y1
class Decoder(nn.Module):
def __init__(self, in_size, out_size):
super(Decoder, self).__init__()
self.conv = ConvRelu2(in_size, out_size)
self.up = F.interpolate
def forward(self, x1, x2):
x2 = self.up(x2, scale_factor=2, mode='bilinear', align_corners=False)
return self.conv(torch.cat([x1, x2], 1))
class AttentionNetNew(nn.Module):
def __init__(self, in_channels=3, out_channels=1):
super(AttentionNetNew, self).__init__()
self.in_channels = in_channels
self.out_channels = out_channels
filters = [64, 128, 256]
self.down1 = Coder(in_channels, filters[0])
self.down2 = Coder(filters[0], filters[1])
self.center = ConvRelu2(filters[1], filters[2])
self.up2 = Decoder(filters[2] + filters[1], filters[1])
self.up1 = Decoder(filters[1] + filters[0], filters[0])
self.final = nn.Conv2d(filters[0], out_channels, 1)
def forward(self, input_0):
primals_1 = self.down1.conv.cr1.conv.weight
primals_2 = self.down1.conv.cr1.conv.bias
primals_4 = self.down1.conv.cr2.conv.weight
primals_5 = self.down1.conv.cr2.conv.bias
primals_6 = self.down2.conv.cr1.conv.weight
primals_7 = self.down2.conv.cr1.conv.bias
primals_8 = self.down2.conv.cr2.conv.weight
primals_9 = self.down2.conv.cr2.conv.bias
primals_10 = self.center.cr1.conv.weight
primals_11 = self.center.cr1.conv.bias
primals_12 = self.center.cr2.conv.weight
primals_13 = self.center.cr2.conv.bias
primals_14 = self.up2.conv.cr1.conv.weight
primals_15 = self.up2.conv.cr1.conv.bias
primals_16 = self.up2.conv.cr2.conv.weight
primals_17 = self.up2.conv.cr2.conv.bias
primals_18 = self.up1.conv.cr1.conv.weight
primals_19 = self.up1.conv.cr1.conv.bias
primals_20 = self.up1.conv.cr2.conv.weight
primals_21 = self.up1.conv.cr2.conv.bias
primals_22 = self.final.weight
primals_23 = self.final.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13, primals_14,
primals_15, primals_16, primals_17, primals_18, primals_19,
primals_20, primals_21, primals_22, primals_23])
return output[0]
| lvxiuwang/ferattention | AttentionNet | false | 7,188 | [
"MIT"
] | 1 | 02e97df4a12129ed6706bddf0d2109650eae8765 | https://github.com/lvxiuwang/ferattention/tree/02e97df4a12129ed6706bddf0d2109650eae8765 | import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.nn.functional
def conv3x3(in_, out):
return nn.Conv2d(in_, out, 3, padding=1)
class ConvRelu(nn.Module):
def __init__(self, in_, out):
super().__init__()
self.conv = conv3x3(in_, out)
self.activation = nn.ReLU(inplace=True)
def forward(self, x):
x = self.conv(x)
x = self.activation(x)
return x
class ConvRelu2(nn.Module):
def __init__(self, _in, _out):
super().__init__()
self.cr1 = ConvRelu(_in, _out)
self.cr2 = ConvRelu(_out, _out)
def forward(self, x):
x = self.cr1(x)
x = self.cr2(x)
return x
class Coder(nn.Module):
def __init__(self, in_size, out_size):
super().__init__()
self.conv = ConvRelu2(in_size, out_size)
self.down = nn.MaxPool2d(2, 2)
def forward(self, x):
y1 = self.conv(x)
y2 = self.down(y1)
return y2, y1
class Decoder(nn.Module):
def __init__(self, in_size, out_size):
super().__init__()
self.conv = ConvRelu2(in_size, out_size)
self.up = F.interpolate
def forward(self, x1, x2):
x2 = self.up(x2, scale_factor=2, mode='bilinear', align_corners=False)
return self.conv(torch.cat([x1, x2], 1))
class Model(nn.Module):
def __init__(self, in_channels=3, out_channels=1):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
filters = [64, 128, 256]
self.down1 = Coder(in_channels, filters[0])
self.down2 = Coder(filters[0], filters[1])
self.center = ConvRelu2(filters[1], filters[2])
self.up2 = Decoder(filters[2] + filters[1], filters[1])
self.up1 = Decoder(filters[1] + filters[0], filters[0])
self.final = nn.Conv2d(filters[0], out_channels, 1)
def forward(self, x):
x, befdown1 = self.down1(x)
x, befdown2 = self.down2(x)
x = self.center(x)
x = self.up2(befdown2, x)
x = self.up1(befdown1, x)
x = self.final(x)
return x
def get_inputs():
return [torch.rand([4, 3, 64, 64])]
def get_init_inputs():
return []
|
MaxPool3x3 | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/sf/csfwn4jjx3wjja53qogk34jyei2gmukxbwonjk7dxkt253ety24o.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.max_pool2d_with_indices]
# Source node to ATen node mapping:
# x => getitem
# Graph fragment:
# %getitem : [num_users=1] = call_function[target=operator.getitem](args = (%_low_memory_max_pool2d_with_offsets, 0), kwargs = {})
triton_poi_fused_max_pool2d_with_indices_0 = async_compile.triton('triton_poi_fused_max_pool2d_with_indices_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_max_pool2d_with_indices_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 9, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x0 = xindex % 4
x4 = xindex
tmp0 = (-1) + x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = (-1) + x0
tmp7 = tmp6 >= tmp1
tmp8 = tmp6 < tmp3
tmp9 = tmp7 & tmp8
tmp10 = tmp5 & tmp9
tmp11 = tl.load(in_ptr0 + ((-5) + x4), tmp10 & xmask, other=float("-inf"))
tmp12 = x0
tmp13 = tmp12 >= tmp1
tmp14 = tmp12 < tmp3
tmp15 = tmp13 & tmp14
tmp16 = tmp5 & tmp15
tmp17 = tl.load(in_ptr0 + ((-4) + x4), tmp16 & xmask, other=float("-inf"))
tmp18 = triton_helpers.maximum(tmp17, tmp11)
tmp19 = 1 + x0
tmp20 = tmp19 >= tmp1
tmp21 = tmp19 < tmp3
tmp22 = tmp20 & tmp21
tmp23 = tmp5 & tmp22
tmp24 = tl.load(in_ptr0 + ((-3) + x4), tmp23 & xmask, other=float("-inf"))
tmp25 = triton_helpers.maximum(tmp24, tmp18)
tmp26 = x1
tmp27 = tmp26 >= tmp1
tmp28 = tmp26 < tmp3
tmp29 = tmp27 & tmp28
tmp30 = tmp29 & tmp9
tmp31 = tl.load(in_ptr0 + ((-1) + x4), tmp30 & xmask, other=float("-inf"))
tmp32 = triton_helpers.maximum(tmp31, tmp25)
tmp33 = tmp29 & tmp15
tmp34 = tl.load(in_ptr0 + (x4), tmp33 & xmask, other=float("-inf"))
tmp35 = triton_helpers.maximum(tmp34, tmp32)
tmp36 = tmp29 & tmp22
tmp37 = tl.load(in_ptr0 + (1 + x4), tmp36 & xmask, other=float("-inf"))
tmp38 = triton_helpers.maximum(tmp37, tmp35)
tmp39 = 1 + x1
tmp40 = tmp39 >= tmp1
tmp41 = tmp39 < tmp3
tmp42 = tmp40 & tmp41
tmp43 = tmp42 & tmp9
tmp44 = tl.load(in_ptr0 + (3 + x4), tmp43 & xmask, other=float("-inf"))
tmp45 = triton_helpers.maximum(tmp44, tmp38)
tmp46 = tmp42 & tmp15
tmp47 = tl.load(in_ptr0 + (4 + x4), tmp46 & xmask, other=float("-inf"))
tmp48 = triton_helpers.maximum(tmp47, tmp45)
tmp49 = tmp42 & tmp22
tmp50 = tl.load(in_ptr0 + (5 + x4), tmp49 & xmask, other=float("-inf"))
tmp51 = triton_helpers.maximum(tmp50, tmp48)
tl.store(out_ptr0 + (x4), tmp51, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.max_pool2d_with_indices]
stream0 = get_raw_stream(0)
triton_poi_fused_max_pool2d_with_indices_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.utils.data
class MaxPool3x3(nn.Module):
"""3x3 max pool with no subsampling."""
def __init__(self, in_channels, out_channels, kernel_size=3, stride=1,
padding=1):
super(MaxPool3x3, self).__init__()
self.maxpool = nn.MaxPool2d(kernel_size, stride, padding)
def forward(self, x):
x = self.maxpool(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_max_pool2d_with_indices_0(in_ptr0, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x0 = xindex % 4
x4 = xindex
tmp0 = -1 + x1
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = -1 + x0
tmp7 = tmp6 >= tmp1
tmp8 = tmp6 < tmp3
tmp9 = tmp7 & tmp8
tmp10 = tmp5 & tmp9
tmp11 = tl.load(in_ptr0 + (-5 + x4), tmp10 & xmask, other=float('-inf'))
tmp12 = x0
tmp13 = tmp12 >= tmp1
tmp14 = tmp12 < tmp3
tmp15 = tmp13 & tmp14
tmp16 = tmp5 & tmp15
tmp17 = tl.load(in_ptr0 + (-4 + x4), tmp16 & xmask, other=float('-inf'))
tmp18 = triton_helpers.maximum(tmp17, tmp11)
tmp19 = 1 + x0
tmp20 = tmp19 >= tmp1
tmp21 = tmp19 < tmp3
tmp22 = tmp20 & tmp21
tmp23 = tmp5 & tmp22
tmp24 = tl.load(in_ptr0 + (-3 + x4), tmp23 & xmask, other=float('-inf'))
tmp25 = triton_helpers.maximum(tmp24, tmp18)
tmp26 = x1
tmp27 = tmp26 >= tmp1
tmp28 = tmp26 < tmp3
tmp29 = tmp27 & tmp28
tmp30 = tmp29 & tmp9
tmp31 = tl.load(in_ptr0 + (-1 + x4), tmp30 & xmask, other=float('-inf'))
tmp32 = triton_helpers.maximum(tmp31, tmp25)
tmp33 = tmp29 & tmp15
tmp34 = tl.load(in_ptr0 + x4, tmp33 & xmask, other=float('-inf'))
tmp35 = triton_helpers.maximum(tmp34, tmp32)
tmp36 = tmp29 & tmp22
tmp37 = tl.load(in_ptr0 + (1 + x4), tmp36 & xmask, other=float('-inf'))
tmp38 = triton_helpers.maximum(tmp37, tmp35)
tmp39 = 1 + x1
tmp40 = tmp39 >= tmp1
tmp41 = tmp39 < tmp3
tmp42 = tmp40 & tmp41
tmp43 = tmp42 & tmp9
tmp44 = tl.load(in_ptr0 + (3 + x4), tmp43 & xmask, other=float('-inf'))
tmp45 = triton_helpers.maximum(tmp44, tmp38)
tmp46 = tmp42 & tmp15
tmp47 = tl.load(in_ptr0 + (4 + x4), tmp46 & xmask, other=float('-inf'))
tmp48 = triton_helpers.maximum(tmp47, tmp45)
tmp49 = tmp42 & tmp22
tmp50 = tl.load(in_ptr0 + (5 + x4), tmp49 & xmask, other=float('-inf'))
tmp51 = triton_helpers.maximum(tmp50, tmp48)
tl.store(out_ptr0 + x4, tmp51, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_max_pool2d_with_indices_0[grid(256)](arg0_1, buf0,
256, XBLOCK=256, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class MaxPool3x3New(nn.Module):
"""3x3 max pool with no subsampling."""
def __init__(self, in_channels, out_channels, kernel_size=3, stride=1,
padding=1):
super(MaxPool3x3New, self).__init__()
self.maxpool = nn.MaxPool2d(kernel_size, stride, padding)
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mc-nya/unnas | MaxPool3x3 | false | 7,189 | [
"MIT"
] | 1 | f778bb743144cf56ce2a48ccca20e9f3a97a7b84 | https://github.com/mc-nya/unnas/tree/f778bb743144cf56ce2a48ccca20e9f3a97a7b84 | import torch
import torch.nn as nn
import torch.utils.data
class Model(nn.Module):
"""3x3 max pool with no subsampling."""
def __init__(self, in_channels, out_channels, kernel_size=3, stride=1,
padding=1):
super().__init__()
self.maxpool = nn.MaxPool2d(kernel_size, stride, padding)
def forward(self, x):
x = self.maxpool(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
MultiHeadAttention | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/v2/cv2ha4mes2yy34yfp3rs7w4jsqgoyz7khozf7gmpk6lmpcslkegp.py
# Topologically Sorted Source Nodes: [truediv], Original ATen: [aten.div]
# Source node to ATen node mapping:
# truediv => div
# Graph fragment:
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%permute, 1.0), kwargs = {})
triton_poi_fused_div_0 = async_compile.triton('triton_poi_fused_div_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_div_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_div_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 1.0
tmp4 = tmp2 * tmp3
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/au/cau4pihcaptiev5y2ewn2o2nvrwhk7hogc72cofmmtbyv4rxc2oy.py
# Topologically Sorted Source Nodes: [k], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# k => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_6, %primals_4, %primals_5, [1], [0], [1], False, [0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/tt/cttmvktt3m2x2nl56afa7l3abaxt7wlehowakdzngkhgs35f3n7u.py
# Topologically Sorted Source Nodes: [p_attn], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# p_attn => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%view_5, [-1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view_5, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_2 = async_compile.triton('triton_poi_fused__softmax_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ry/cryn7ntc2gpkbfzbre3xh7lffx7zkbskw6oihbzsekkgajmdbki6.py
# Topologically Sorted Source Nodes: [p_attn], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# p_attn => div_1, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [-1], True), kwargs = {})
# %div_1 : [num_users=2] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_3 = async_compile.triton('triton_poi_fused__softmax_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_7, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_8, (4, ), (1, ))
assert_size_stride(primals_9, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_10, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [q], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4), (16, 4, 1))
# Topologically Sorted Source Nodes: [k], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(primals_6, primals_4, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4), (16, 4, 1))
# Topologically Sorted Source Nodes: [v], Original ATen: [aten.convolution]
buf2 = extern_kernels.convolution(primals_6, primals_7, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4), (16, 4, 1))
buf3 = reinterpret_tensor(buf0, (4, 4, 4, 1), (16, 4, 1, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [truediv], Original ATen: [aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_div_0.run(buf3, primals_2, 64, grid=grid(64), stream=stream0)
del primals_2
buf4 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [k], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf4, primals_5, 64, grid=grid(64), stream=stream0)
del primals_5
buf5 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [scores], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf4, (16, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [p_attn], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf5, buf6, 256, grid=grid(256), stream=stream0)
buf7 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf5 # reuse
# Topologically Sorted Source Nodes: [p_attn], Original ATen: [aten._softmax]
triton_poi_fused__softmax_3.run(buf6, buf7, 256, grid=grid(256), stream=stream0)
del buf6
buf8 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [v], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf8, primals_8, 64, grid=grid(64), stream=stream0)
del primals_8
buf9 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [output], Original ATen: [aten.bmm]
extern_kernels.bmm(reinterpret_tensor(buf7, (16, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf8, (16, 4, 1), (4, 1, 0), 0), out=buf9)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf10 = extern_kernels.convolution(reinterpret_tensor(buf9, (4, 4, 4), (16, 4, 1), 0), primals_9, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf10, (4, 4, 4), (16, 4, 1))
buf11 = buf10; del buf10 # reuse
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf11, primals_10, 64, grid=grid(64), stream=stream0)
del primals_10
return (buf11, buf7, primals_1, primals_3, primals_4, primals_6, primals_7, primals_9, buf7, reinterpret_tensor(buf9, (4, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf8, (16, 1, 4), (4, 4, 1), 0), reinterpret_tensor(buf3, (16, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, 4, 1), (4, 1, 1), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import typing
import torch.multiprocessing
from torch import nn
from torch.nn import functional as F
import torch.optim
import torch.utils.data
import torch.distributed
class MultiHeadAttention(nn.Module):
def __init__(self, channels: 'int', out_channels: 'int', n_heads: 'int',
p_dropout: 'float'=0.0, window_size: 'typing.Optional[int]'=None,
heads_share: 'bool'=True, block_length: 'typing.Optional[int]'=None,
proximal_bias: 'bool'=False, proximal_init: 'bool'=False):
super().__init__()
assert channels % n_heads == 0
self.channels = channels
self.out_channels = out_channels
self.n_heads = n_heads
self.p_dropout = p_dropout
self.window_size = window_size
self.heads_share = heads_share
self.block_length = block_length
self.proximal_bias = proximal_bias
self.proximal_init = proximal_init
self.attn = None
self.k_channels = channels // n_heads
self.conv_q = nn.Conv1d(channels, channels, 1)
self.conv_k = nn.Conv1d(channels, channels, 1)
self.conv_v = nn.Conv1d(channels, channels, 1)
self.conv_o = nn.Conv1d(channels, out_channels, 1)
self.drop = nn.Dropout(p_dropout)
if window_size is not None:
n_heads_rel = 1 if heads_share else n_heads
rel_stddev = self.k_channels ** -0.5
self.emb_rel_k = nn.Parameter(torch.randn(n_heads_rel,
window_size * 2 + 1, self.k_channels) * rel_stddev)
self.emb_rel_v = nn.Parameter(torch.randn(n_heads_rel,
window_size * 2 + 1, self.k_channels) * rel_stddev)
nn.init.xavier_uniform_(self.conv_q.weight)
nn.init.xavier_uniform_(self.conv_k.weight)
nn.init.xavier_uniform_(self.conv_v.weight)
if proximal_init:
with torch.no_grad():
self.conv_k.weight.copy_(self.conv_q.weight)
self.conv_k.bias.copy_(self.conv_q.bias)
def forward(self, x, c, attn_mask=None):
q = self.conv_q(x)
k = self.conv_k(c)
v = self.conv_v(c)
x, self.attn = self.attention(q, k, v, mask=attn_mask)
x = self.conv_o(x)
return x
def attention(self, query, key, value, mask=None):
b, d, t_s, t_t = *key.size(), query.size(2)
query = query.view(b, self.n_heads, self.k_channels, t_t).transpose(
2, 3)
key = key.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
value = value.view(b, self.n_heads, self.k_channels, t_s).transpose(
2, 3)
scores = torch.matmul(query / math.sqrt(self.k_channels), key.
transpose(-2, -1))
if self.window_size is not None:
assert t_s == t_t, 'Relative attention is only available for self-attention.'
key_relative_embeddings = self._get_relative_embeddings(self.
emb_rel_k, t_s)
rel_logits = self._matmul_with_relative_keys(query / math.sqrt(
self.k_channels), key_relative_embeddings)
scores_local = self._relative_position_to_absolute_position(
rel_logits)
scores = scores + scores_local
if self.proximal_bias:
assert t_s == t_t, 'Proximal bias is only available for self-attention.'
scores = scores + self._attention_bias_proximal(t_s).type_as(scores
)
if mask is not None:
scores = scores.masked_fill(mask == 0, -10000.0)
if self.block_length is not None:
assert t_s == t_t, 'Local attention is only available for self-attention.'
block_mask = torch.ones_like(scores).triu(-self.block_length
).tril(self.block_length)
scores = scores.masked_fill(block_mask == 0, -10000.0)
p_attn = F.softmax(scores, dim=-1)
p_attn = self.drop(p_attn)
output = torch.matmul(p_attn, value)
if self.window_size is not None:
relative_weights = self._absolute_position_to_relative_position(
p_attn)
value_relative_embeddings = self._get_relative_embeddings(self.
emb_rel_v, t_s)
output = output + self._matmul_with_relative_values(
relative_weights, value_relative_embeddings)
output = output.transpose(2, 3).contiguous().view(b, d, t_t)
return output, p_attn
def _matmul_with_relative_values(self, x, y):
"""
x: [b, h, l, m]
y: [h or 1, m, d]
ret: [b, h, l, d]
"""
ret = torch.matmul(x, y.unsqueeze(0))
return ret
def _matmul_with_relative_keys(self, x, y):
"""
x: [b, h, l, d]
y: [h or 1, m, d]
ret: [b, h, l, m]
"""
ret = torch.matmul(x, y.unsqueeze(0).transpose(-2, -1))
return ret
def _get_relative_embeddings(self, relative_embeddings, length):
pad_length = max(length - (self.window_size + 1), 0)
slice_start_position = max(self.window_size + 1 - length, 0)
slice_end_position = slice_start_position + 2 * length - 1
if pad_length > 0:
padded_relative_embeddings = F.pad(relative_embeddings, (0, 0,
pad_length, pad_length, 0, 0))
else:
padded_relative_embeddings = relative_embeddings
used_relative_embeddings = padded_relative_embeddings[:,
slice_start_position:slice_end_position]
return used_relative_embeddings
def _relative_position_to_absolute_position(self, x):
"""
x: [b, h, l, 2*l-1]
ret: [b, h, l, l]
"""
batch, heads, length, _ = x.size()
x = F.pad(x, (0, 1, 0, 0, 0, 0, 0, 0))
x_flat = x.view([batch, heads, length * 2 * length])
x_flat = F.pad(x_flat, (0, length - 1, 0, 0, 0, 0))
x_final = x_flat.view([batch, heads, length + 1, 2 * length - 1])[:,
:, :length, length - 1:]
return x_final
def _absolute_position_to_relative_position(self, x):
"""
x: [b, h, l, l]
ret: [b, h, l, 2*l-1]
"""
batch, heads, length, _ = x.size()
x = F.pad(x, (0, length - 1, 0, 0, 0, 0, 0, 0))
x_flat = x.view([batch, heads, length * length + length * (length - 1)]
)
x_flat = F.pad(x_flat, (length, 0, 0, 0, 0, 0))
x_final = x_flat.view([batch, heads, length, 2 * length])[:, :, :, 1:]
return x_final
def _attention_bias_proximal(self, length):
"""Bias for self-attention to encourage attention to close positions.
Args:
length: an integer scalar.
Returns:
a Tensor with shape [1, 1, length, length]
"""
r = torch.arange(length, dtype=torch.float32)
diff = torch.unsqueeze(r, 0) - torch.unsqueeze(r, 1)
return torch.unsqueeze(torch.unsqueeze(-torch.log1p(torch.abs(diff)
), 0), 0)
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'channels': 4, 'out_channels': 4, 'n_heads': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import math
import typing
import torch.multiprocessing
from torch import nn
from torch.nn import functional as F
import torch.optim
import torch.utils.data
import torch.distributed
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_div_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 1.0
tmp4 = tmp2 * tmp3
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_3(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_7, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_8, (4,), (1,))
assert_size_stride(primals_9, (4, 4, 1), (4, 1, 1))
assert_size_stride(primals_10, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 4), (16, 4, 1))
buf1 = extern_kernels.convolution(primals_6, primals_4, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4), (16, 4, 1))
buf2 = extern_kernels.convolution(primals_6, primals_7, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf2, (4, 4, 4), (16, 4, 1))
buf3 = reinterpret_tensor(buf0, (4, 4, 4, 1), (16, 4, 1, 1), 0)
del buf0
get_raw_stream(0)
triton_poi_fused_div_0[grid(64)](buf3, primals_2, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_2
buf4 = buf1
del buf1
triton_poi_fused_convolution_1[grid(64)](buf4, primals_5, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_5
buf5 = empty_strided_cuda((16, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf3, (16, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf4, (16, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_2[grid(256)](buf5, buf6, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf7 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf5
triton_poi_fused__softmax_3[grid(256)](buf6, buf7, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del buf6
buf8 = buf2
del buf2
triton_poi_fused_convolution_1[grid(64)](buf8, primals_8, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_8
buf9 = empty_strided_cuda((16, 4, 1), (4, 1, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf7, (16, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf8, (16, 4, 1), (4, 1, 0), 0), out=buf9)
buf10 = extern_kernels.convolution(reinterpret_tensor(buf9, (4, 4,
4), (16, 4, 1), 0), primals_9, stride=(1,), padding=(0,),
dilation=(1,), transposed=False, output_padding=(0,), groups=1,
bias=None)
assert_size_stride(buf10, (4, 4, 4), (16, 4, 1))
buf11 = buf10
del buf10
triton_poi_fused_convolution_1[grid(64)](buf11, primals_10, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_10
return (buf11, buf7, primals_1, primals_3, primals_4, primals_6,
primals_7, primals_9, buf7, reinterpret_tensor(buf9, (4, 4, 4), (16,
4, 1), 0), reinterpret_tensor(buf8, (16, 1, 4), (4, 4, 1), 0),
reinterpret_tensor(buf3, (16, 1, 4), (4, 1, 1), 0),
reinterpret_tensor(buf4, (16, 4, 1), (4, 1, 4), 0))
class MultiHeadAttentionNew(nn.Module):
def __init__(self, channels: 'int', out_channels: 'int', n_heads: 'int',
p_dropout: 'float'=0.0, window_size: 'typing.Optional[int]'=None,
heads_share: 'bool'=True, block_length: 'typing.Optional[int]'=None,
proximal_bias: 'bool'=False, proximal_init: 'bool'=False):
super().__init__()
assert channels % n_heads == 0
self.channels = channels
self.out_channels = out_channels
self.n_heads = n_heads
self.p_dropout = p_dropout
self.window_size = window_size
self.heads_share = heads_share
self.block_length = block_length
self.proximal_bias = proximal_bias
self.proximal_init = proximal_init
self.attn = None
self.k_channels = channels // n_heads
self.conv_q = nn.Conv1d(channels, channels, 1)
self.conv_k = nn.Conv1d(channels, channels, 1)
self.conv_v = nn.Conv1d(channels, channels, 1)
self.conv_o = nn.Conv1d(channels, out_channels, 1)
self.drop = nn.Dropout(p_dropout)
if window_size is not None:
n_heads_rel = 1 if heads_share else n_heads
rel_stddev = self.k_channels ** -0.5
self.emb_rel_k = nn.Parameter(torch.randn(n_heads_rel,
window_size * 2 + 1, self.k_channels) * rel_stddev)
self.emb_rel_v = nn.Parameter(torch.randn(n_heads_rel,
window_size * 2 + 1, self.k_channels) * rel_stddev)
nn.init.xavier_uniform_(self.conv_q.weight)
nn.init.xavier_uniform_(self.conv_k.weight)
nn.init.xavier_uniform_(self.conv_v.weight)
if proximal_init:
with torch.no_grad():
self.conv_k.weight.copy_(self.conv_q.weight)
self.conv_k.bias.copy_(self.conv_q.bias)
def attention(self, query, key, value, mask=None):
b, d, t_s, t_t = *key.size(), query.size(2)
query = query.view(b, self.n_heads, self.k_channels, t_t).transpose(
2, 3)
key = key.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
value = value.view(b, self.n_heads, self.k_channels, t_s).transpose(
2, 3)
scores = torch.matmul(query / math.sqrt(self.k_channels), key.
transpose(-2, -1))
if self.window_size is not None:
assert t_s == t_t, 'Relative attention is only available for self-attention.'
key_relative_embeddings = self._get_relative_embeddings(self.
emb_rel_k, t_s)
rel_logits = self._matmul_with_relative_keys(query / math.sqrt(
self.k_channels), key_relative_embeddings)
scores_local = self._relative_position_to_absolute_position(
rel_logits)
scores = scores + scores_local
if self.proximal_bias:
assert t_s == t_t, 'Proximal bias is only available for self-attention.'
scores = scores + self._attention_bias_proximal(t_s).type_as(scores
)
if mask is not None:
scores = scores.masked_fill(mask == 0, -10000.0)
if self.block_length is not None:
assert t_s == t_t, 'Local attention is only available for self-attention.'
block_mask = torch.ones_like(scores).triu(-self.block_length
).tril(self.block_length)
scores = scores.masked_fill(block_mask == 0, -10000.0)
p_attn = F.softmax(scores, dim=-1)
p_attn = self.drop(p_attn)
output = torch.matmul(p_attn, value)
if self.window_size is not None:
relative_weights = self._absolute_position_to_relative_position(
p_attn)
value_relative_embeddings = self._get_relative_embeddings(self.
emb_rel_v, t_s)
output = output + self._matmul_with_relative_values(
relative_weights, value_relative_embeddings)
output = output.transpose(2, 3).contiguous().view(b, d, t_t)
return output, p_attn
def _matmul_with_relative_values(self, x, y):
"""
x: [b, h, l, m]
y: [h or 1, m, d]
ret: [b, h, l, d]
"""
ret = torch.matmul(x, y.unsqueeze(0))
return ret
def _matmul_with_relative_keys(self, x, y):
"""
x: [b, h, l, d]
y: [h or 1, m, d]
ret: [b, h, l, m]
"""
ret = torch.matmul(x, y.unsqueeze(0).transpose(-2, -1))
return ret
def _get_relative_embeddings(self, relative_embeddings, length):
pad_length = max(length - (self.window_size + 1), 0)
slice_start_position = max(self.window_size + 1 - length, 0)
slice_end_position = slice_start_position + 2 * length - 1
if pad_length > 0:
padded_relative_embeddings = F.pad(relative_embeddings, (0, 0,
pad_length, pad_length, 0, 0))
else:
padded_relative_embeddings = relative_embeddings
used_relative_embeddings = padded_relative_embeddings[:,
slice_start_position:slice_end_position]
return used_relative_embeddings
def _relative_position_to_absolute_position(self, x):
"""
x: [b, h, l, 2*l-1]
ret: [b, h, l, l]
"""
batch, heads, length, _ = x.size()
x = F.pad(x, (0, 1, 0, 0, 0, 0, 0, 0))
x_flat = x.view([batch, heads, length * 2 * length])
x_flat = F.pad(x_flat, (0, length - 1, 0, 0, 0, 0))
x_final = x_flat.view([batch, heads, length + 1, 2 * length - 1])[:,
:, :length, length - 1:]
return x_final
def _absolute_position_to_relative_position(self, x):
"""
x: [b, h, l, l]
ret: [b, h, l, 2*l-1]
"""
batch, heads, length, _ = x.size()
x = F.pad(x, (0, length - 1, 0, 0, 0, 0, 0, 0))
x_flat = x.view([batch, heads, length * length + length * (length - 1)]
)
x_flat = F.pad(x_flat, (length, 0, 0, 0, 0, 0))
x_final = x_flat.view([batch, heads, length, 2 * length])[:, :, :, 1:]
return x_final
def _attention_bias_proximal(self, length):
"""Bias for self-attention to encourage attention to close positions.
Args:
length: an integer scalar.
Returns:
a Tensor with shape [1, 1, length, length]
"""
r = torch.arange(length, dtype=torch.float32)
diff = torch.unsqueeze(r, 0) - torch.unsqueeze(r, 1)
return torch.unsqueeze(torch.unsqueeze(-torch.log1p(torch.abs(diff)
), 0), 0)
def forward(self, input_0, input_1):
primals_1 = self.conv_q.weight
primals_2 = self.conv_q.bias
primals_4 = self.conv_k.weight
primals_5 = self.conv_k.bias
primals_7 = self.conv_v.weight
primals_8 = self.conv_v.bias
primals_9 = self.conv_o.weight
primals_10 = self.conv_o.bias
primals_3 = input_0
primals_6 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9, primals_10])
return output[0]
| mbarnig/vits-train | MultiHeadAttention | false | 7,190 | [
"MIT"
] | 1 | cfb8a0fc91daad868fe3d062ebf85d62edbd7506 | https://github.com/mbarnig/vits-train/tree/cfb8a0fc91daad868fe3d062ebf85d62edbd7506 | import math
import torch
import typing
import torch.multiprocessing
from torch import nn
from torch.nn import functional as F
import torch.optim
import torch.utils.data
import torch.distributed
class Model(nn.Module):
def __init__(self, channels: 'int', out_channels: 'int', n_heads: 'int',
p_dropout: 'float'=0.0, window_size: 'typing.Optional[int]'=None,
heads_share: 'bool'=True, block_length: 'typing.Optional[int]'=None,
proximal_bias: 'bool'=False, proximal_init: 'bool'=False):
super().__init__()
assert channels % n_heads == 0
self.channels = channels
self.out_channels = out_channels
self.n_heads = n_heads
self.p_dropout = p_dropout
self.window_size = window_size
self.heads_share = heads_share
self.block_length = block_length
self.proximal_bias = proximal_bias
self.proximal_init = proximal_init
self.attn = None
self.k_channels = channels // n_heads
self.conv_q = nn.Conv1d(channels, channels, 1)
self.conv_k = nn.Conv1d(channels, channels, 1)
self.conv_v = nn.Conv1d(channels, channels, 1)
self.conv_o = nn.Conv1d(channels, out_channels, 1)
self.drop = nn.Dropout(p_dropout)
if window_size is not None:
n_heads_rel = 1 if heads_share else n_heads
rel_stddev = self.k_channels ** -0.5
self.emb_rel_k = nn.Parameter(torch.randn(n_heads_rel,
window_size * 2 + 1, self.k_channels) * rel_stddev)
self.emb_rel_v = nn.Parameter(torch.randn(n_heads_rel,
window_size * 2 + 1, self.k_channels) * rel_stddev)
nn.init.xavier_uniform_(self.conv_q.weight)
nn.init.xavier_uniform_(self.conv_k.weight)
nn.init.xavier_uniform_(self.conv_v.weight)
if proximal_init:
with torch.no_grad():
self.conv_k.weight.copy_(self.conv_q.weight)
self.conv_k.bias.copy_(self.conv_q.bias)
def forward(self, x, c, attn_mask=None):
q = self.conv_q(x)
k = self.conv_k(c)
v = self.conv_v(c)
x, self.attn = self.attention(q, k, v, mask=attn_mask)
x = self.conv_o(x)
return x
def attention(self, query, key, value, mask=None):
b, d, t_s, t_t = *key.size(), query.size(2)
query = query.view(b, self.n_heads, self.k_channels, t_t).transpose(
2, 3)
key = key.view(b, self.n_heads, self.k_channels, t_s).transpose(2, 3)
value = value.view(b, self.n_heads, self.k_channels, t_s).transpose(
2, 3)
scores = torch.matmul(query / math.sqrt(self.k_channels), key.
transpose(-2, -1))
if self.window_size is not None:
assert t_s == t_t, 'Relative attention is only available for self-attention.'
key_relative_embeddings = self._get_relative_embeddings(self.
emb_rel_k, t_s)
rel_logits = self._matmul_with_relative_keys(query / math.sqrt(
self.k_channels), key_relative_embeddings)
scores_local = self._relative_position_to_absolute_position(
rel_logits)
scores = scores + scores_local
if self.proximal_bias:
assert t_s == t_t, 'Proximal bias is only available for self-attention.'
scores = scores + self._attention_bias_proximal(t_s).type_as(scores
)
if mask is not None:
scores = scores.masked_fill(mask == 0, -10000.0)
if self.block_length is not None:
assert t_s == t_t, 'Local attention is only available for self-attention.'
block_mask = torch.ones_like(scores).triu(-self.block_length
).tril(self.block_length)
scores = scores.masked_fill(block_mask == 0, -10000.0)
p_attn = F.softmax(scores, dim=-1)
p_attn = self.drop(p_attn)
output = torch.matmul(p_attn, value)
# ... truncated (>4000 chars) for memory efficiency |
ChannelNorm | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/i7/ci7t7iiz7rvr7feeg7u3oqbndzrc2eexgichqwatlcys5unofv7u.py
# Topologically Sorted Source Nodes: [mean, pow_1, mean_x2, pow_2, var, sub_1, add, sqrt, x_norm, mul, x_norm_1], Original ATen: [aten.mean, aten.pow, aten.sub, aten.add, aten.sqrt, aten.div, aten.mul]
# Source node to ATen node mapping:
# add => add
# mean => mean
# mean_x2 => mean_1
# mul => mul
# pow_1 => pow_1
# pow_2 => pow_2
# sqrt => sqrt
# sub_1 => sub_1
# var => sub
# x_norm => div
# x_norm_1 => add_1
# Graph fragment:
# %mean : [num_users=3] = call_function[target=torch.ops.aten.mean.dim](args = (%view, [-1], True), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%view, 2), kwargs = {})
# %mean_1 : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%pow_1, [-1], True), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%mean, 2), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mean_1, %pow_2), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view, %mean), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sub, 1e-05), kwargs = {})
# %sqrt : [num_users=2] = call_function[target=torch.ops.aten.sqrt.default](args = (%add,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sub_1, %sqrt), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_1, %div), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %view_2), kwargs = {})
triton_per_fused_add_div_mean_mul_pow_sqrt_sub_0 = async_compile.triton('triton_per_fused_add_div_mean_mul_pow_sqrt_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[4, 64],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32', 7: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 7), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_div_mean_mul_pow_sqrt_sub_0', 'mutated_arg_names': ['in_out_ptr0', 'in_out_ptr1'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 2, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_div_mean_mul_pow_sqrt_sub_0(in_out_ptr0, in_out_ptr1, in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 4
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + (64*x0)), xmask, other=0.0)
tmp18 = tl.load(in_ptr1 + (0))
tmp19 = tl.broadcast_to(tmp18, [XBLOCK, RBLOCK])
tmp23 = tl.load(in_ptr2 + (0))
tmp24 = tl.broadcast_to(tmp23, [XBLOCK, RBLOCK])
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp5 = tmp0 * tmp0
tmp6 = tl.broadcast_to(tmp5, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = 64.0
tmp11 = tmp4 / tmp10
tmp12 = tmp9 / tmp10
tmp13 = tmp11 * tmp11
tmp14 = tmp12 - tmp13
tmp15 = 1e-05
tmp16 = tmp14 + tmp15
tmp17 = libdevice.sqrt(tmp16)
tmp20 = tmp0 - tmp11
tmp21 = tmp20 / tmp17
tmp22 = tmp19 * tmp21
tmp25 = tmp22 + tmp24
tl.debug_barrier()
tl.store(in_out_ptr0 + (x0), tmp11, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + (x0), tmp17, xmask)
tl.store(out_ptr0 + (r1 + (64*x0)), tmp25, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, ), (1, ))
assert_size_stride(primals_3, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 1, 1), (1, 4, 4), torch.float32)
buf2 = empty_strided_cuda((4, 1, 1), (1, 4, 4), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 1, 1), (1, 1, 1), 0); del buf0 # reuse
buf3 = reinterpret_tensor(buf2, (4, 1, 1), (1, 1, 1), 0); del buf2 # reuse
buf4 = empty_strided_cuda((4, 1, 64), (64, 64, 1), torch.float32)
# Topologically Sorted Source Nodes: [mean, pow_1, mean_x2, pow_2, var, sub_1, add, sqrt, x_norm, mul, x_norm_1], Original ATen: [aten.mean, aten.pow, aten.sub, aten.add, aten.sqrt, aten.div, aten.mul]
stream0 = get_raw_stream(0)
triton_per_fused_add_div_mean_mul_pow_sqrt_sub_0.run(buf1, buf3, primals_1, primals_2, primals_3, buf4, 4, 64, grid=grid(4), stream=stream0)
del primals_2
del primals_3
return (reinterpret_tensor(buf4, (4, 4, 4, 4), (64, 16, 4, 1), 0), primals_1, buf1, buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class ChannelNorm(Module):
"""
## Channel Normalization
This is similar to [Group Normalization](../group_norm/index.html) but affine transform is done group wise.
"""
def __init__(self, channels, groups, eps: 'float'=1e-05, affine: 'bool'
=True):
"""
* `groups` is the number of groups the features are divided into
* `channels` is the number of features in the input
* `eps` is $\\epsilon$, used in $\\sqrt{Var[x^{(k)}] + \\epsilon}$ for numerical stability
* `affine` is whether to scale and shift the normalized value
"""
super().__init__()
self.channels = channels
self.groups = groups
self.eps = eps
self.affine = affine
if self.affine:
self.scale = nn.Parameter(torch.ones(groups))
self.shift = nn.Parameter(torch.zeros(groups))
def forward(self, x: 'torch.Tensor'):
"""
`x` is a tensor of shape `[batch_size, channels, *]`.
`*` denotes any number of (possibly 0) dimensions.
For example, in an image (2D) convolution this will be
`[batch_size, channels, height, width]`
"""
x_shape = x.shape
batch_size = x_shape[0]
assert self.channels == x.shape[1]
x = x.view(batch_size, self.groups, -1)
mean = x.mean(dim=[-1], keepdim=True)
mean_x2 = (x ** 2).mean(dim=[-1], keepdim=True)
var = mean_x2 - mean ** 2
x_norm = (x - mean) / torch.sqrt(var + self.eps)
if self.affine:
x_norm = self.scale.view(1, -1, 1) * x_norm + self.shift.view(1,
-1, 1)
return x_norm.view(x_shape)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'channels': 4, 'groups': 1}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_per_fused_add_div_mean_mul_pow_sqrt_sub_0(in_out_ptr0,
in_out_ptr1, in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, rnumel,
XBLOCK: tl.constexpr):
xnumel = 4
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = tl.load(in_ptr0 + (r1 + 64 * x0), xmask, other=0.0)
tmp18 = tl.load(in_ptr1 + 0)
tmp19 = tl.broadcast_to(tmp18, [XBLOCK, RBLOCK])
tmp23 = tl.load(in_ptr2 + 0)
tmp24 = tl.broadcast_to(tmp23, [XBLOCK, RBLOCK])
tmp1 = tl.broadcast_to(tmp0, [XBLOCK, RBLOCK])
tmp3 = tl.where(xmask, tmp1, 0)
tmp4 = tl.sum(tmp3, 1)[:, None]
tmp5 = tmp0 * tmp0
tmp6 = tl.broadcast_to(tmp5, [XBLOCK, RBLOCK])
tmp8 = tl.where(xmask, tmp6, 0)
tmp9 = tl.sum(tmp8, 1)[:, None]
tmp10 = 64.0
tmp11 = tmp4 / tmp10
tmp12 = tmp9 / tmp10
tmp13 = tmp11 * tmp11
tmp14 = tmp12 - tmp13
tmp15 = 1e-05
tmp16 = tmp14 + tmp15
tmp17 = libdevice.sqrt(tmp16)
tmp20 = tmp0 - tmp11
tmp21 = tmp20 / tmp17
tmp22 = tmp19 * tmp21
tmp25 = tmp22 + tmp24
tl.debug_barrier()
tl.store(in_out_ptr0 + x0, tmp11, xmask)
tl.debug_barrier()
tl.store(in_out_ptr1 + x0, tmp17, xmask)
tl.store(out_ptr0 + (r1 + 64 * x0), tmp25, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1,), (1,))
assert_size_stride(primals_3, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 1, 1), (1, 4, 4), torch.float32)
buf2 = empty_strided_cuda((4, 1, 1), (1, 4, 4), torch.float32)
buf1 = reinterpret_tensor(buf0, (4, 1, 1), (1, 1, 1), 0)
del buf0
buf3 = reinterpret_tensor(buf2, (4, 1, 1), (1, 1, 1), 0)
del buf2
buf4 = empty_strided_cuda((4, 1, 64), (64, 64, 1), torch.float32)
get_raw_stream(0)
triton_per_fused_add_div_mean_mul_pow_sqrt_sub_0[grid(4)](buf1,
buf3, primals_1, primals_2, primals_3, buf4, 4, 64, XBLOCK=1,
num_warps=2, num_stages=1)
del primals_2
del primals_3
return reinterpret_tensor(buf4, (4, 4, 4, 4), (64, 16, 4, 1), 0
), primals_1, buf1, buf3
class ChannelNormNew(Module):
"""
## Channel Normalization
This is similar to [Group Normalization](../group_norm/index.html) but affine transform is done group wise.
"""
def __init__(self, channels, groups, eps: 'float'=1e-05, affine: 'bool'
=True):
"""
* `groups` is the number of groups the features are divided into
* `channels` is the number of features in the input
* `eps` is $\\epsilon$, used in $\\sqrt{Var[x^{(k)}] + \\epsilon}$ for numerical stability
* `affine` is whether to scale and shift the normalized value
"""
super().__init__()
self.channels = channels
self.groups = groups
self.eps = eps
self.affine = affine
if self.affine:
self.scale = nn.Parameter(torch.ones(groups))
self.shift = nn.Parameter(torch.zeros(groups))
def forward(self, input_0):
primals_2 = self.scale
primals_3 = self.shift
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | ChannelNorm | false | 7,191 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Channel Normalization
This is similar to [Group Normalization](../group_norm/index.html) but affine transform is done group wise.
"""
def __init__(self, channels, groups, eps: 'float'=1e-05, affine: 'bool'
=True):
"""
* `groups` is the number of groups the features are divided into
* `channels` is the number of features in the input
* `eps` is $\\epsilon$, used in $\\sqrt{Var[x^{(k)}] + \\epsilon}$ for numerical stability
* `affine` is whether to scale and shift the normalized value
"""
super().__init__()
self.channels = channels
self.groups = groups
self.eps = eps
self.affine = affine
if self.affine:
self.scale = nn.Parameter(torch.ones(groups))
self.shift = nn.Parameter(torch.zeros(groups))
def forward(self, x: 'torch.Tensor'):
"""
`x` is a tensor of shape `[batch_size, channels, *]`.
`*` denotes any number of (possibly 0) dimensions.
For example, in an image (2D) convolution this will be
`[batch_size, channels, height, width]`
"""
x_shape = x.shape
batch_size = x_shape[0]
assert self.channels == x.shape[1]
x = x.view(batch_size, self.groups, -1)
mean = x.mean(dim=[-1], keepdim=True)
mean_x2 = (x ** 2).mean(dim=[-1], keepdim=True)
var = mean_x2 - mean ** 2
x_norm = (x - mean) / torch.sqrt(var + self.eps)
if self.affine:
x_norm = self.scale.view(1, -1, 1) * x_norm + self.shift.view(1,
-1, 1)
return x_norm.view(x_shape)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 1]
|
NodeNetwork | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/gd/cgdkwbgll3qv33h27jvqqssey37x43ic2aqfujlpcwlrzderhg35.py
# Topologically Sorted Source Nodes: [output, x], Original ATen: [aten.add, aten.clamp, aten.ge]
# Source node to ATen node mapping:
# output => add_1
# x => clamp_min
# Graph fragment:
# %add_1 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_plus_mm_1, %primals_5), kwargs = {})
# %clamp_min : [num_users=3] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add_1, 0), kwargs = {})
# %ge_1 : [num_users=1] = call_function[target=torch.ops.aten.ge.Scalar](args = (%add_1, 0), kwargs = {})
triton_poi_fused_add_clamp_ge_0 = async_compile.triton('triton_poi_fused_add_clamp_ge_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*i1', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_ge_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_ge_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 32
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp5 = tmp2 >= tmp3
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp5, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/qf/cqfba74b3yhmr4smax2cvjl7whxhynb3lorxdgir7eilewsighvu.py
# Topologically Sorted Source Nodes: [add_2, output_1, x_1], Original ATen: [aten.add, aten.clamp, aten.ge]
# Source node to ATen node mapping:
# add_2 => mm_plus_mm
# output_1 => add_3
# x_1 => clamp_min_1
# Graph fragment:
# %mm_plus_mm : [num_users=1] = call_function[target=torch._inductor.fx_passes.post_grad.mm_plus_mm](args = (), kwargs = {mat1: %primals_3, mat2: %mm_3, mat3: %clamp_min, mat4: %primals_7})
# %add_3 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_plus_mm, %primals_8), kwargs = {})
# %clamp_min_1 : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.default](args = (%add_3, 0), kwargs = {})
# %ge : [num_users=1] = call_function[target=torch.ops.aten.ge.Scalar](args = (%add_3, 0), kwargs = {})
triton_poi_fused_add_clamp_ge_1 = async_compile.triton('triton_poi_fused_add_clamp_ge_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*i1', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_clamp_ge_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_clamp_ge_1(in_ptr0, in_ptr1, in_ptr2, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 32
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp3 = tl.load(in_ptr2 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp5 = 0.0
tmp6 = triton_helpers.maximum(tmp4, tmp5)
tmp7 = tmp4 >= tmp5
tl.store(out_ptr0 + (x2), tmp6, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/w4/cw4mbfaglppoilurhnlij4aivvpd6du5dgvm56bnk7ylgctpb4hw.py
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# softmax => amax, div, exp, sub, sum_1
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%addmm_1, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%addmm_1, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_2 = async_compile.triton('triton_poi_fused__softmax_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 2)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (2*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (2*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp4 = tmp0 - tmp3
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp1 - tmp3
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp2 - tmp3
tmp9 = tl_math.exp(tmp8)
tmp10 = tmp7 + tmp9
tmp11 = tmp5 / tmp10
tl.store(out_ptr0 + (x2), tmp11, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13 = args
args.clear()
assert_size_stride(primals_1, (4, 32), (32, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 32), (32, 1))
assert_size_stride(primals_5, (32, ), (1, ))
assert_size_stride(primals_6, (32, 32), (32, 1))
assert_size_stride(primals_7, (32, 32), (32, 1))
assert_size_stride(primals_8, (32, ), (1, ))
assert_size_stride(primals_9, (4, 4), (4, 1))
assert_size_stride(primals_10, (16, 32), (32, 1))
assert_size_stride(primals_11, (16, ), (1, ))
assert_size_stride(primals_12, (2, 16), (16, 1))
assert_size_stride(primals_13, (2, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 32), (32, 1), torch.float32)
# Topologically Sorted Source Nodes: [potential_msgs], Original ATen: [aten.mm]
extern_kernels.mm(primals_2, primals_1, out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 32), (32, 1), torch.float32)
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
extern_kernels._mm_plus_mm(primals_3, buf0, primals_2, primals_4, out=buf1)
del primals_4
buf2 = buf0; del buf0 # reuse
buf12 = empty_strided_cuda((4, 32), (32, 1), torch.bool)
# Topologically Sorted Source Nodes: [output, x], Original ATen: [aten.add, aten.clamp, aten.ge]
stream0 = get_raw_stream(0)
triton_poi_fused_add_clamp_ge_0.run(buf1, primals_5, buf2, buf12, 128, grid=grid(128), stream=stream0)
del primals_5
buf3 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [potential_msgs_1], Original ATen: [aten.mm]
extern_kernels.mm(buf2, primals_6, out=buf3)
buf4 = empty_strided_cuda((4, 32), (32, 1), torch.float32)
# Topologically Sorted Source Nodes: [add_2], Original ATen: [aten.add]
extern_kernels.mm(primals_3, buf3, out=buf4)
buf5 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [add_2], Original ATen: [aten.add]
extern_kernels.mm(buf2, primals_7, out=buf5)
buf6 = empty_strided_cuda((4, 32), (32, 1), torch.float32)
buf11 = empty_strided_cuda((4, 32), (32, 1), torch.bool)
# Topologically Sorted Source Nodes: [add_2, output_1, x_1], Original ATen: [aten.add, aten.clamp, aten.ge]
triton_poi_fused_add_clamp_ge_1.run(buf4, buf5, primals_8, buf6, buf11, 128, grid=grid(128), stream=stream0)
del buf4
del primals_8
buf7 = buf5; del buf5 # reuse
# Topologically Sorted Source Nodes: [add_2, output_1, x_1, output_2], Original ATen: [aten.add, aten.clamp, aten.mm]
extern_kernels.mm(primals_9, buf6, out=buf7)
del buf6
buf8 = empty_strided_cuda((4, 16), (16, 1), torch.float32)
# Topologically Sorted Source Nodes: [output_3], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_11, buf7, reinterpret_tensor(primals_10, (32, 16), (1, 32), 0), alpha=1, beta=1, out=buf8)
del primals_11
buf9 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [output_4], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_13, buf8, reinterpret_tensor(primals_12, (16, 2), (1, 16), 0), alpha=1, beta=1, out=buf9)
del primals_13
buf10 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [softmax], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf9, buf10, 8, grid=grid(8), stream=stream0)
del buf9
return (buf10, buf7, buf8, buf10, primals_12, primals_10, reinterpret_tensor(primals_9, (4, 4), (1, 4), 0), buf11, reinterpret_tensor(buf2, (32, 4), (1, 32), 0), reinterpret_tensor(primals_7, (32, 32), (1, 32), 0), reinterpret_tensor(primals_3, (4, 4), (1, 4), 0), reinterpret_tensor(primals_6, (32, 32), (1, 32), 0), buf12, reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 32), (32, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 32), (32, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((32, 32), (32, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((32, 32), (32, 1), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((16, 32), (32, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((16, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_12 = rand_strided((2, 16), (16, 1), device='cuda:0', dtype=torch.float32)
primals_13 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11, primals_12, primals_13])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn.functional as F
from torch.nn.parameter import Parameter
def global_sum_pool(X, batch_mat):
if batch_mat is None or batch_mat.dim() == 1:
return torch.sum(X, dim=0).unsqueeze(0)
else:
return torch.mm(batch_mat, X)
class BasicGraphConvolutionLayer(torch.nn.Module):
def __init__(self, in_channels, out_channels):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.W2 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.W1 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.bias = Parameter(torch.zeros(out_channels, dtype=torch.float32))
def forward(self, X, A):
potential_msgs = torch.mm(X, self.W2)
propagated_msgs = torch.mm(A, potential_msgs)
root_update = torch.mm(X, self.W1)
output = propagated_msgs + root_update + self.bias
return output
class NodeNetwork(torch.nn.Module):
def __init__(self, input_features):
super().__init__()
self.conv_1 = BasicGraphConvolutionLayer(input_features, 32)
self.conv_2 = BasicGraphConvolutionLayer(32, 32)
self.fc_1 = torch.nn.Linear(32, 16)
self.out_layer = torch.nn.Linear(16, 2)
def forward(self, X, A, batch_mat):
x = self.conv_1(X, A).clamp(0)
x = self.conv_2(x, A).clamp(0)
output = global_sum_pool(x, batch_mat)
output = self.fc_1(output)
output = self.out_layer(output)
return F.softmax(output, dim=1)
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [[], {'input_features': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch.nn.parameter import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_clamp_ge_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 32
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = triton_helpers.maximum(tmp2, tmp3)
tmp5 = tmp2 >= tmp3
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp5, xmask)
@triton.jit
def triton_poi_fused_add_clamp_ge_1(in_ptr0, in_ptr1, in_ptr2, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 32
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp3 = tl.load(in_ptr2 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp5 = 0.0
tmp6 = triton_helpers.maximum(tmp4, tmp5)
tmp7 = tmp4 >= tmp5
tl.store(out_ptr0 + x2, tmp6, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 2
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 2 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 2 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp4 = tmp0 - tmp3
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp1 - tmp3
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp2 - tmp3
tmp9 = tl_math.exp(tmp8)
tmp10 = tmp7 + tmp9
tmp11 = tmp5 / tmp10
tl.store(out_ptr0 + x2, tmp11, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11, primals_12,
primals_13) = args
args.clear()
assert_size_stride(primals_1, (4, 32), (32, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4), (4, 1))
assert_size_stride(primals_4, (4, 32), (32, 1))
assert_size_stride(primals_5, (32,), (1,))
assert_size_stride(primals_6, (32, 32), (32, 1))
assert_size_stride(primals_7, (32, 32), (32, 1))
assert_size_stride(primals_8, (32,), (1,))
assert_size_stride(primals_9, (4, 4), (4, 1))
assert_size_stride(primals_10, (16, 32), (32, 1))
assert_size_stride(primals_11, (16,), (1,))
assert_size_stride(primals_12, (2, 16), (16, 1))
assert_size_stride(primals_13, (2,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 32), (32, 1), torch.float32)
extern_kernels.mm(primals_2, primals_1, out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 32), (32, 1), torch.float32)
extern_kernels._mm_plus_mm(primals_3, buf0, primals_2, primals_4,
out=buf1)
del primals_4
buf2 = buf0
del buf0
buf12 = empty_strided_cuda((4, 32), (32, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_add_clamp_ge_0[grid(128)](buf1, primals_5, buf2,
buf12, 128, XBLOCK=128, num_warps=4, num_stages=1)
del primals_5
buf3 = buf1
del buf1
extern_kernels.mm(buf2, primals_6, out=buf3)
buf4 = empty_strided_cuda((4, 32), (32, 1), torch.float32)
extern_kernels.mm(primals_3, buf3, out=buf4)
buf5 = buf3
del buf3
extern_kernels.mm(buf2, primals_7, out=buf5)
buf6 = empty_strided_cuda((4, 32), (32, 1), torch.float32)
buf11 = empty_strided_cuda((4, 32), (32, 1), torch.bool)
triton_poi_fused_add_clamp_ge_1[grid(128)](buf4, buf5, primals_8,
buf6, buf11, 128, XBLOCK=128, num_warps=4, num_stages=1)
del buf4
del primals_8
buf7 = buf5
del buf5
extern_kernels.mm(primals_9, buf6, out=buf7)
del buf6
buf8 = empty_strided_cuda((4, 16), (16, 1), torch.float32)
extern_kernels.addmm(primals_11, buf7, reinterpret_tensor(
primals_10, (32, 16), (1, 32), 0), alpha=1, beta=1, out=buf8)
del primals_11
buf9 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
extern_kernels.addmm(primals_13, buf8, reinterpret_tensor(
primals_12, (16, 2), (1, 16), 0), alpha=1, beta=1, out=buf9)
del primals_13
buf10 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
triton_poi_fused__softmax_2[grid(8)](buf9, buf10, 8, XBLOCK=8,
num_warps=1, num_stages=1)
del buf9
return (buf10, buf7, buf8, buf10, primals_12, primals_10,
reinterpret_tensor(primals_9, (4, 4), (1, 4), 0), buf11,
reinterpret_tensor(buf2, (32, 4), (1, 32), 0), reinterpret_tensor(
primals_7, (32, 32), (1, 32), 0), reinterpret_tensor(primals_3, (4,
4), (1, 4), 0), reinterpret_tensor(primals_6, (32, 32), (1, 32), 0),
buf12, reinterpret_tensor(primals_2, (4, 4), (1, 4), 0))
def global_sum_pool(X, batch_mat):
if batch_mat is None or batch_mat.dim() == 1:
return torch.sum(X, dim=0).unsqueeze(0)
else:
return torch.mm(batch_mat, X)
class BasicGraphConvolutionLayer(torch.nn.Module):
def __init__(self, in_channels, out_channels):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.W2 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.W1 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.bias = Parameter(torch.zeros(out_channels, dtype=torch.float32))
def forward(self, X, A):
potential_msgs = torch.mm(X, self.W2)
propagated_msgs = torch.mm(A, potential_msgs)
root_update = torch.mm(X, self.W1)
output = propagated_msgs + root_update + self.bias
return output
class NodeNetworkNew(torch.nn.Module):
def __init__(self, input_features):
super().__init__()
self.conv_1 = BasicGraphConvolutionLayer(input_features, 32)
self.conv_2 = BasicGraphConvolutionLayer(32, 32)
self.fc_1 = torch.nn.Linear(32, 16)
self.out_layer = torch.nn.Linear(16, 2)
def forward(self, input_0, input_1, input_2):
primals_1 = self.conv_1.W2
primals_4 = self.conv_1.W1
primals_5 = self.conv_1.bias
primals_6 = self.conv_2.W2
primals_7 = self.conv_2.W1
primals_8 = self.conv_2.bias
primals_10 = self.fc_1.weight
primals_11 = self.fc_1.bias
primals_12 = self.out_layer.weight
primals_13 = self.out_layer.bias
primals_2 = input_0
primals_3 = input_1
primals_9 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11, primals_12, primals_13])
return output[0]
| mbrukman/machine-learning-book | NodeNetwork | false | 7,192 | [
"MIT"
] | 1 | f29a0f8aafa63a77081f3bcec68866e33dd41776 | https://github.com/mbrukman/machine-learning-book/tree/f29a0f8aafa63a77081f3bcec68866e33dd41776 | import torch
import torch.nn.functional as F
from torch.nn.parameter import Parameter
def global_sum_pool(X, batch_mat):
if batch_mat is None or batch_mat.dim() == 1:
return torch.sum(X, dim=0).unsqueeze(0)
else:
return torch.mm(batch_mat, X)
class BasicGraphConvolutionLayer(torch.nn.Module):
def __init__(self, in_channels, out_channels):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.W2 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.W1 = Parameter(torch.rand((in_channels, out_channels), dtype=
torch.float32))
self.bias = Parameter(torch.zeros(out_channels, dtype=torch.float32))
def forward(self, X, A):
potential_msgs = torch.mm(X, self.W2)
propagated_msgs = torch.mm(A, potential_msgs)
root_update = torch.mm(X, self.W1)
output = propagated_msgs + root_update + self.bias
return output
class Model(torch.nn.Module):
def __init__(self, input_features):
super().__init__()
self.conv_1 = BasicGraphConvolutionLayer(input_features, 32)
self.conv_2 = BasicGraphConvolutionLayer(32, 32)
self.fc_1 = torch.nn.Linear(32, 16)
self.out_layer = torch.nn.Linear(16, 2)
def forward(self, X, A, batch_mat):
x = self.conv_1(X, A).clamp(0)
x = self.conv_2(x, A).clamp(0)
output = global_sum_pool(x, batch_mat)
output = self.fc_1(output)
output = self.out_layer(output)
return F.softmax(output, dim=1)
def get_inputs():
return [torch.rand([4, 4]), torch.rand([4, 4]), torch.rand([4, 4])]
def get_init_inputs():
return [4]
|
FFN | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/pf/cpfm5wdpxnjdoxwc64lkd5hti4p2apuv2blizip6w7zx245e2zdx.py
# Topologically Sorted Source Nodes: [mul, x], Original ATen: [aten.mul, aten.constant_pad_nd]
# Source node to ATen node mapping:
# mul => mul
# x => constant_pad_nd
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_1, %primals_2), kwargs = {})
# %constant_pad_nd : [num_users=2] = call_function[target=torch.ops.aten.constant_pad_nd.default](args = (%mul, [1, 2, 0, 0, 0, 0], 0.0), kwargs = {})
triton_poi_fused_constant_pad_nd_mul_0 = async_compile.triton('triton_poi_fused_constant_pad_nd_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_constant_pad_nd_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_constant_pad_nd_mul_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 112
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 7
x1 = (xindex // 7)
x2 = xindex
tmp0 = (-1) + x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = tl.load(in_ptr0 + ((-1) + x0 + (4*x1)), tmp5 & xmask, other=0.0)
tmp7 = tl.load(in_ptr1 + ((-1) + x0 + (4*x1)), tmp5 & xmask, other=0.0)
tmp8 = tmp6 * tmp7
tmp9 = tl.full(tmp8.shape, 0.0, tmp8.dtype)
tmp10 = tl.where(tmp5, tmp8, tmp9)
tl.store(out_ptr0 + (x2), tmp10, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/fg/cfgllpb4gyvdjlh7fdxswezh7emhepgm4i7bapbduuzso44m6c2f.py
# Topologically Sorted Source Nodes: [x_1, x_2, mul_1, x_4], Original ATen: [aten.convolution, aten.relu, aten.mul, aten.constant_pad_nd]
# Source node to ATen node mapping:
# mul_1 => mul_1
# x_1 => convolution
# x_2 => relu
# x_4 => constant_pad_nd_1
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%constant_pad_nd, %primals_3, %primals_4, [1], [0], [1], False, [0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%relu, %primals_2), kwargs = {})
# %constant_pad_nd_1 : [num_users=2] = call_function[target=torch.ops.aten.constant_pad_nd.default](args = (%mul_1, [1, 2, 0, 0, 0, 0], 0.0), kwargs = {})
triton_poi_fused_constant_pad_nd_convolution_mul_relu_1 = async_compile.triton('triton_poi_fused_constant_pad_nd_convolution_mul_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_constant_pad_nd_convolution_mul_relu_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_constant_pad_nd_convolution_mul_relu_1(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 112
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 7
x3 = (xindex // 7)
x1 = (xindex // 7) % 4
x4 = xindex
tmp0 = (-1) + x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = tl.load(in_ptr0 + ((-1) + x0 + (4*x3)), tmp5 & xmask, other=0.0)
tmp7 = tl.load(in_ptr1 + (x1), tmp5 & xmask, eviction_policy='evict_last', other=0.0)
tmp8 = tmp6 + tmp7
tmp9 = tl.full([1], 0, tl.int32)
tmp10 = triton_helpers.maximum(tmp9, tmp8)
tmp11 = tl.load(in_ptr2 + ((-1) + x0 + (4*x3)), tmp5 & xmask, other=0.0)
tmp12 = tmp10 * tmp11
tmp13 = tl.full(tmp12.shape, 0.0, tmp12.dtype)
tmp14 = tl.where(tmp5, tmp12, tmp13)
tl.store(out_ptr0 + (x4), tmp14, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/p4/cp4nytdwbukq45muq2audhwinyxcui5jzm35yncgordm6rei7uos.py
# Topologically Sorted Source Nodes: [x_5, mul_2], Original ATen: [aten.convolution, aten.mul]
# Source node to ATen node mapping:
# mul_2 => mul_2
# x_5 => convolution_1
# Graph fragment:
# %convolution_1 : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%constant_pad_nd_1, %primals_5, %primals_6, [1], [0], [1], False, [0], 1), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%convolution_1, %primals_2), kwargs = {})
triton_poi_fused_convolution_mul_2 = async_compile.triton('triton_poi_fused_convolution_mul_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_mul_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_mul_2(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (x3), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 * tmp3
tl.store(in_out_ptr0 + (x3), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/wy/cwyc5bi426x3ud3par7mhtpup3hm766bdsmojwrlcobttexvf7oe.py
# Topologically Sorted Source Nodes: [x_1, x_2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x_1 => convolution
# x_2 => relu
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%constant_pad_nd, %primals_3, %primals_4, [1], [0], [1], False, [0], 1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%convolution,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_convolution_relu_threshold_backward_3 = async_compile.triton('triton_poi_fused_convolution_relu_threshold_backward_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_relu_threshold_backward_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_3(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + (x3), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4, ), (1, ))
assert_size_stride(primals_5, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 7), (28, 7, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, x], Original ATen: [aten.mul, aten.constant_pad_nd]
stream0 = get_raw_stream(0)
triton_poi_fused_constant_pad_nd_mul_0.run(primals_1, primals_2, buf0, 112, grid=grid(112), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_3, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4), (16, 4, 1))
buf2 = empty_strided_cuda((4, 4, 7), (28, 7, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_1, x_2, mul_1, x_4], Original ATen: [aten.convolution, aten.relu, aten.mul, aten.constant_pad_nd]
triton_poi_fused_constant_pad_nd_convolution_mul_relu_1.run(buf1, primals_4, primals_2, buf2, 112, grid=grid(112), stream=stream0)
# Topologically Sorted Source Nodes: [x_5], Original ATen: [aten.convolution]
buf3 = extern_kernels.convolution(buf2, primals_5, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf3, (4, 4, 4), (16, 4, 1))
buf4 = buf3; del buf3 # reuse
# Topologically Sorted Source Nodes: [x_5, mul_2], Original ATen: [aten.convolution, aten.mul]
triton_poi_fused_convolution_mul_2.run(buf4, primals_6, primals_2, 64, grid=grid(64), stream=stream0)
del primals_6
buf5 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [x_1, x_2], Original ATen: [aten.convolution, aten.relu, aten.threshold_backward]
triton_poi_fused_convolution_relu_threshold_backward_3.run(buf1, primals_4, buf5, 64, grid=grid(64), stream=stream0)
del buf1
del primals_4
return (buf4, primals_2, primals_3, primals_5, buf0, buf2, buf5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import typing
import torch.multiprocessing
from torch import nn
from torch.nn import functional as F
import torch.optim
import torch.utils.data
import torch.distributed
class FFN(nn.Module):
def __init__(self, in_channels: 'int', out_channels: 'int',
filter_channels: 'int', kernel_size: 'int', p_dropout: 'float'=0.0,
activation: 'typing.Optional[str]'=None, causal: 'bool'=False):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.filter_channels = filter_channels
self.kernel_size = kernel_size
self.p_dropout = p_dropout
self.activation = activation
self.causal = causal
if causal:
self.padding = self._causal_padding
else:
self.padding = self._same_padding
self.conv_1 = nn.Conv1d(in_channels, filter_channels, kernel_size)
self.conv_2 = nn.Conv1d(filter_channels, out_channels, kernel_size)
self.drop = nn.Dropout(p_dropout)
def forward(self, x, x_mask):
x = self.conv_1(self.padding(x * x_mask))
if self.activation == 'gelu':
x = x * torch.sigmoid(1.702 * x)
else:
x = torch.relu(x)
x = self.drop(x)
x = self.conv_2(self.padding(x * x_mask))
return x * x_mask
def _causal_padding(self, x):
if self.kernel_size == 1:
return x
pad_l = self.kernel_size - 1
pad_r = 0
x = F.pad(x, (pad_l, pad_r, 0, 0, 0, 0))
return x
def _same_padding(self, x):
if self.kernel_size == 1:
return x
pad_l = (self.kernel_size - 1) // 2
pad_r = self.kernel_size // 2
x = F.pad(x, (pad_l, pad_r, 0, 0, 0, 0))
return x
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'filter_channels': 4,
'kernel_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import typing
import torch.multiprocessing
from torch import nn
from torch.nn import functional as F
import torch.optim
import torch.utils.data
import torch.distributed
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_constant_pad_nd_mul_0(in_ptr0, in_ptr1, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 112
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 7
x1 = xindex // 7
x2 = xindex
tmp0 = -1 + x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = tl.load(in_ptr0 + (-1 + x0 + 4 * x1), tmp5 & xmask, other=0.0)
tmp7 = tl.load(in_ptr1 + (-1 + x0 + 4 * x1), tmp5 & xmask, other=0.0)
tmp8 = tmp6 * tmp7
tmp9 = tl.full(tmp8.shape, 0.0, tmp8.dtype)
tmp10 = tl.where(tmp5, tmp8, tmp9)
tl.store(out_ptr0 + x2, tmp10, xmask)
@triton.jit
def triton_poi_fused_constant_pad_nd_convolution_mul_relu_1(in_ptr0,
in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 112
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 7
x3 = xindex // 7
x1 = xindex // 7 % 4
x4 = xindex
tmp0 = -1 + x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = tl.load(in_ptr0 + (-1 + x0 + 4 * x3), tmp5 & xmask, other=0.0)
tmp7 = tl.load(in_ptr1 + x1, tmp5 & xmask, eviction_policy='evict_last',
other=0.0)
tmp8 = tmp6 + tmp7
tmp9 = tl.full([1], 0, tl.int32)
tmp10 = triton_helpers.maximum(tmp9, tmp8)
tmp11 = tl.load(in_ptr2 + (-1 + x0 + 4 * x3), tmp5 & xmask, other=0.0)
tmp12 = tmp10 * tmp11
tmp13 = tl.full(tmp12.shape, 0.0, tmp12.dtype)
tmp14 = tl.where(tmp5, tmp12, tmp13)
tl.store(out_ptr0 + x4, tmp14, xmask)
@triton.jit
def triton_poi_fused_convolution_mul_2(in_out_ptr0, in_ptr0, in_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + x3, xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 * tmp3
tl.store(in_out_ptr0 + x3, tmp4, xmask)
@triton.jit
def triton_poi_fused_convolution_relu_threshold_backward_3(in_ptr0, in_ptr1,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(out_ptr0 + x3, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_4, (4,), (1,))
assert_size_stride(primals_5, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_6, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 7), (28, 7, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_constant_pad_nd_mul_0[grid(112)](primals_1,
primals_2, buf0, 112, XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
buf1 = extern_kernels.convolution(buf0, primals_3, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4), (16, 4, 1))
buf2 = empty_strided_cuda((4, 4, 7), (28, 7, 1), torch.float32)
triton_poi_fused_constant_pad_nd_convolution_mul_relu_1[grid(112)](buf1
, primals_4, primals_2, buf2, 112, XBLOCK=128, num_warps=4,
num_stages=1)
buf3 = extern_kernels.convolution(buf2, primals_5, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf3, (4, 4, 4), (16, 4, 1))
buf4 = buf3
del buf3
triton_poi_fused_convolution_mul_2[grid(64)](buf4, primals_6,
primals_2, 64, XBLOCK=64, num_warps=1, num_stages=1)
del primals_6
buf5 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.bool)
triton_poi_fused_convolution_relu_threshold_backward_3[grid(64)](buf1,
primals_4, buf5, 64, XBLOCK=64, num_warps=1, num_stages=1)
del buf1
del primals_4
return buf4, primals_2, primals_3, primals_5, buf0, buf2, buf5
class FFNNew(nn.Module):
def __init__(self, in_channels: 'int', out_channels: 'int',
filter_channels: 'int', kernel_size: 'int', p_dropout: 'float'=0.0,
activation: 'typing.Optional[str]'=None, causal: 'bool'=False):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.filter_channels = filter_channels
self.kernel_size = kernel_size
self.p_dropout = p_dropout
self.activation = activation
self.causal = causal
if causal:
self.padding = self._causal_padding
else:
self.padding = self._same_padding
self.conv_1 = nn.Conv1d(in_channels, filter_channels, kernel_size)
self.conv_2 = nn.Conv1d(filter_channels, out_channels, kernel_size)
self.drop = nn.Dropout(p_dropout)
def _causal_padding(self, x):
if self.kernel_size == 1:
return x
pad_l = self.kernel_size - 1
pad_r = 0
x = F.pad(x, (pad_l, pad_r, 0, 0, 0, 0))
return x
def _same_padding(self, x):
if self.kernel_size == 1:
return x
pad_l = (self.kernel_size - 1) // 2
pad_r = self.kernel_size // 2
x = F.pad(x, (pad_l, pad_r, 0, 0, 0, 0))
return x
def forward(self, input_0, input_1):
primals_1 = self.conv_1.weight
primals_4 = self.conv_1.bias
primals_2 = self.conv_2.weight
primals_6 = self.conv_2.bias
primals_3 = input_0
primals_5 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6])
return output[0]
| mbarnig/vits-train | FFN | false | 7,193 | [
"MIT"
] | 1 | cfb8a0fc91daad868fe3d062ebf85d62edbd7506 | https://github.com/mbarnig/vits-train/tree/cfb8a0fc91daad868fe3d062ebf85d62edbd7506 | import torch
import typing
import torch.multiprocessing
from torch import nn
from torch.nn import functional as F
import torch.optim
import torch.utils.data
import torch.distributed
class Model(nn.Module):
def __init__(self, in_channels: 'int', out_channels: 'int',
filter_channels: 'int', kernel_size: 'int', p_dropout: 'float'=0.0,
activation: 'typing.Optional[str]'=None, causal: 'bool'=False):
super().__init__()
self.in_channels = in_channels
self.out_channels = out_channels
self.filter_channels = filter_channels
self.kernel_size = kernel_size
self.p_dropout = p_dropout
self.activation = activation
self.causal = causal
if causal:
self.padding = self._causal_padding
else:
self.padding = self._same_padding
self.conv_1 = nn.Conv1d(in_channels, filter_channels, kernel_size)
self.conv_2 = nn.Conv1d(filter_channels, out_channels, kernel_size)
self.drop = nn.Dropout(p_dropout)
def forward(self, x, x_mask):
x = self.conv_1(self.padding(x * x_mask))
if self.activation == 'gelu':
x = x * torch.sigmoid(1.702 * x)
else:
x = torch.relu(x)
x = self.drop(x)
x = self.conv_2(self.padding(x * x_mask))
return x * x_mask
def _causal_padding(self, x):
if self.kernel_size == 1:
return x
pad_l = self.kernel_size - 1
pad_r = 0
x = F.pad(x, (pad_l, pad_r, 0, 0, 0, 0))
return x
def _same_padding(self, x):
if self.kernel_size == 1:
return x
pad_l = (self.kernel_size - 1) // 2
pad_r = self.kernel_size // 2
x = F.pad(x, (pad_l, pad_r, 0, 0, 0, 0))
return x
def get_inputs():
return [torch.rand([4, 4, 4]), torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'filter_channels': 4,
'kernel_size': 4}]
|
DiscriminatorLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/6k/c6keogbsbczhv7za5qvdnqtoyslxmpkn26o5ri5y2uquynatq3pp.py
# Topologically Sorted Source Nodes: [sub, relu, mean], Original ATen: [aten.rsub, aten.relu, aten.mean]
# Source node to ATen node mapping:
# mean => mean
# relu => relu
# sub => sub
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %arg0_1), kwargs = {})
# %relu : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%sub,), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%relu,), kwargs = {})
triton_per_fused_mean_relu_rsub_0 = async_compile.triton('triton_per_fused_mean_relu_rsub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {2: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=(2,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mean_relu_rsub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 1, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mean_relu_rsub_0(in_out_ptr0, in_ptr0, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = tl.broadcast_to(tmp4, [RBLOCK])
tmp7 = triton_helpers.promote_to_tensor(tl.sum(tmp5, 0))
tmp8 = 256.0
tmp9 = tmp7 / tmp8
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp9, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/lj/cljvcit4p6vjtwmejklnkzpuexfn22rbffq5dyzgt4sefx3jy3zk.py
# Topologically Sorted Source Nodes: [add, relu_1, mean_1], Original ATen: [aten.add, aten.relu, aten.mean]
# Source node to ATen node mapping:
# add => add
# mean_1 => mean_1
# relu_1 => relu_1
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg1_1, 1), kwargs = {})
# %relu_1 : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%add,), kwargs = {})
# %mean_1 : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%relu_1,), kwargs = {})
triton_per_fused_add_mean_relu_1 = async_compile.triton('triton_per_fused_add_mean_relu_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {2: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=(2,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_mean_relu_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 1, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_mean_relu_1(in_out_ptr0, in_ptr0, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = tl.broadcast_to(tmp4, [RBLOCK])
tmp7 = triton_helpers.promote_to_tensor(tl.sum(tmp5, 0))
tmp8 = 256.0
tmp9 = tmp7 / tmp8
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp9, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf2 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [sub, relu, mean], Original ATen: [aten.rsub, aten.relu, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_mean_relu_rsub_0.run(buf2, arg0_1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
buf1 = empty_strided_cuda((), (), torch.float32)
buf3 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [add, relu_1, mean_1], Original ATen: [aten.add, aten.relu, aten.mean]
triton_per_fused_add_mean_relu_1.run(buf3, arg1_1, 1, 256, grid=grid(1), stream=stream0)
del arg1_1
return (buf2, buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.nn.functional as F
import torch.utils.data
import torch.nn.functional
import torch.autograd
class DiscriminatorLoss(Module):
"""
## Discriminator Loss
We want to find $w$ to maximize
$$\\mathbb{E}_{x \\sim \\mathbb{P}_r} [f_w(x)]- \\mathbb{E}_{z \\sim p(z)} [f_w(g_ heta(z))]$$,
so we minimize,
$$- rac{1}{m} \\sum_{i=1}^m f_w ig(x^{(i)} ig) +
rac{1}{m} \\sum_{i=1}^m f_w ig( g_ heta(z^{(i)}) ig)$$
"""
def forward(self, f_real: 'torch.Tensor', f_fake: 'torch.Tensor'):
"""
* `f_real` is $f_w(x)$
* `f_fake` is $f_w(g_ heta(z))$
This returns the a tuple with losses for $f_w(x)$ and $f_w(g_ heta(z))$,
which are later added.
They are kept separate for logging.
"""
return F.relu(1 - f_real).mean(), F.relu(1 + f_fake).mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_mean_relu_rsub_0(in_out_ptr0, in_ptr0, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = tl.broadcast_to(tmp4, [RBLOCK])
tmp7 = triton_helpers.promote_to_tensor(tl.sum(tmp5, 0))
tmp8 = 256.0
tmp9 = tmp7 / tmp8
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp9, None)
@triton.jit
def triton_per_fused_add_mean_relu_1(in_out_ptr0, in_ptr0, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = tl.broadcast_to(tmp4, [RBLOCK])
tmp7 = triton_helpers.promote_to_tensor(tl.sum(tmp5, 0))
tmp8 = 256.0
tmp9 = tmp7 / tmp8
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp9, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf2 = buf0
del buf0
get_raw_stream(0)
triton_per_fused_mean_relu_rsub_0[grid(1)](buf2, arg0_1, 1, 256,
num_warps=2, num_stages=1)
del arg0_1
buf1 = empty_strided_cuda((), (), torch.float32)
buf3 = buf1
del buf1
triton_per_fused_add_mean_relu_1[grid(1)](buf3, arg1_1, 1, 256,
num_warps=2, num_stages=1)
del arg1_1
return buf2, buf3
class DiscriminatorLossNew(Module):
"""
## Discriminator Loss
We want to find $w$ to maximize
$$\\mathbb{E}_{x \\sim \\mathbb{P}_r} [f_w(x)]- \\mathbb{E}_{z \\sim p(z)} [f_w(g_ heta(z))]$$,
so we minimize,
$$- rac{1}{m} \\sum_{i=1}^m f_w ig(x^{(i)} ig) +
rac{1}{m} \\sum_{i=1}^m f_w ig( g_ heta(z^{(i)}) ig)$$
"""
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0], output[1]
| mcx/annotated_deep_learning_paper_implementations | DiscriminatorLoss | false | 7,194 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.nn.functional as F
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Discriminator Loss
We want to find $w$ to maximize
$$\\mathbb{E}_{x \\sim \\mathbb{P}_r} [f_w(x)]- \\mathbb{E}_{z \\sim p(z)} [f_w(g_ heta(z))]$$,
so we minimize,
$$- rac{1}{m} \\sum_{i=1}^m f_w ig(x^{(i)} ig) +
rac{1}{m} \\sum_{i=1}^m f_w ig( g_ heta(z^{(i)}) ig)$$
"""
def forward(self, f_real: 'torch.Tensor', f_fake: 'torch.Tensor'):
"""
* `f_real` is $f_w(x)$
* `f_fake` is $f_w(g_ heta(z))$
This returns the a tuple with losses for $f_w(x)$ and $f_w(g_ heta(z))$,
which are later added.
They are kept separate for logging.
"""
return F.relu(1 - f_real).mean(), F.relu(1 + f_fake).mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
Model | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/jq/cjqaq2meov3vkcgfealq7w4w35tw2oemvmhneuxmigeoumva22p7.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.sigmoid]
# Source node to ATen node mapping:
# x_1 => sigmoid
# Graph fragment:
# %sigmoid : [num_users=2] = call_function[target=torch.ops.aten.sigmoid.default](args = (%view_1,), kwargs = {})
triton_poi_fused_sigmoid_0 = async_compile.triton('triton_poi_fused_sigmoid_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sigmoid_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sigmoid_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.sigmoid(tmp2)
tl.store(in_out_ptr0 + (x2), tmp3, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/xk/cxkugsynlmnyrjhah42fewrhwovuvurnuv2qimo2qhxq27wjmq7q.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# x_3 => amax, exp, sub
# Graph fragment:
# %amax : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%view_3, [1], True), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view_3, %amax), kwargs = {})
# %exp : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub,), kwargs = {})
triton_poi_fused__softmax_1 = async_compile.triton('triton_poi_fused__softmax_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x3), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/jf/cjfzp64ny4hf7wdw5wptah3hqv5fcsh5rrw4brz7uxcy6ad57n7h.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten._softmax]
# Source node to ATen node mapping:
# x_3 => div, sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp, [1], True), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp, %sum_1), kwargs = {})
triton_poi_fused__softmax_2 = async_compile.triton('triton_poi_fused__softmax_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__softmax_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + (x3), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.sigmoid]
stream0 = get_raw_stream(0)
triton_poi_fused_sigmoid_0.run(buf1, primals_2, 256, grid=grid(256), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten._softmax]
triton_poi_fused__softmax_1.run(buf2, buf3, 256, grid=grid(256), stream=stream0)
buf4 = reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten._softmax]
triton_poi_fused__softmax_2.run(buf3, buf4, 256, grid=grid(256), stream=stream0)
del buf3
return (buf4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, buf4, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self, input_size, hidden_size, output_size):
super(Model, self).__init__()
self.layer1 = nn.Linear(input_size, hidden_size)
self.layer2 = nn.Linear(hidden_size, output_size)
def forward(self, x):
x = self.layer1(x)
x = nn.Sigmoid()(x)
x = self.layer2(x)
x = nn.Softmax(dim=1)(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'input_size': 4, 'hidden_size': 4, 'output_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_sigmoid_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.sigmoid(tmp2)
tl.store(in_out_ptr0 + x2, tmp3, xmask)
@triton.jit
def triton_poi_fused__softmax_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x3, tmp9, xmask)
@triton.jit
def triton_poi_fused__softmax_2(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tmp1 + tmp2
tmp5 = tmp3 + tmp4
tmp7 = tmp5 + tmp6
tmp8 = tmp0 / tmp7
tl.store(out_ptr0 + x3, tmp8, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf0
get_raw_stream(0)
triton_poi_fused_sigmoid_0[grid(256)](buf1, primals_2, 256, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 4), (
4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0),
alpha=1, beta=1, out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused__softmax_1[grid(256)](buf2, buf3, 256, XBLOCK=128,
num_warps=4, num_stages=1)
buf4 = reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf2
triton_poi_fused__softmax_2[grid(256)](buf3, buf4, 256, XBLOCK=256,
num_warps=4, num_stages=1)
del buf3
return buf4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf1, buf4, primals_4
class ModelNew(nn.Module):
def __init__(self, input_size, hidden_size, output_size):
super(ModelNew, self).__init__()
self.layer1 = nn.Linear(input_size, hidden_size)
self.layer2 = nn.Linear(hidden_size, output_size)
def forward(self, input_0):
primals_1 = self.layer1.weight
primals_2 = self.layer1.bias
primals_4 = self.layer2.weight
primals_5 = self.layer2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| mbrukman/machine-learning-book | Model | false | 7,195 | [
"MIT"
] | 1 | f29a0f8aafa63a77081f3bcec68866e33dd41776 | https://github.com/mbrukman/machine-learning-book/tree/f29a0f8aafa63a77081f3bcec68866e33dd41776 | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self, input_size, hidden_size, output_size):
super(Model, self).__init__()
self.layer1 = nn.Linear(input_size, hidden_size)
self.layer2 = nn.Linear(hidden_size, output_size)
def forward(self, x):
x = self.layer1(x)
x = nn.Sigmoid()(x)
x = self.layer2(x)
x = nn.Softmax(dim=1)(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4, 4]
|
ClippedValueFunctionLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ux/cuxyq6jg627wu3buhxoipzlue5jekgs24d5ikz7kqoqhqlliw5dg.py
# Topologically Sorted Source Nodes: [sub_1, pow_1, sub, neg, clamp, clipped_value, sub_2, pow_2, vf_loss, mean, mul], Original ATen: [aten.sub, aten.pow, aten.neg, aten.clamp, aten.add, aten.maximum, aten.mean, aten.mul]
# Source node to ATen node mapping:
# clamp => clamp_max, clamp_min
# clipped_value => add
# mean => mean
# mul => mul
# neg => neg
# pow_1 => pow_1
# pow_2 => pow_2
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# vf_loss => maximum
# Graph fragment:
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %arg3_1), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_1, 2), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg0_1, %arg1_1), kwargs = {})
# %neg : [num_users=1] = call_function[target=torch.ops.aten.neg.default](args = (%arg2_1,), kwargs = {})
# %clamp_min : [num_users=1] = call_function[target=torch.ops.aten.clamp_min.Tensor](args = (%sub, %neg), kwargs = {})
# %clamp_max : [num_users=1] = call_function[target=torch.ops.aten.clamp_max.Tensor](args = (%clamp_min, %arg2_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg1_1, %clamp_max), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %arg3_1), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub_2, 2), kwargs = {})
# %maximum : [num_users=1] = call_function[target=torch.ops.aten.maximum.default](args = (%pow_1, %pow_2), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%maximum,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mean, 0.5), kwargs = {})
triton_per_fused_add_clamp_maximum_mean_mul_neg_pow_sub_0 = async_compile.triton('triton_per_fused_add_clamp_maximum_mean_mul_neg_pow_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {5: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 6), equal_to_1=(5,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_clamp_maximum_mean_mul_neg_pow_sub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 4, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_clamp_maximum_mean_mul_neg_pow_sub_0(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = tl.load(in_ptr1 + (r0), None)
tmp4 = tl.load(in_ptr2 + (r0), None)
tmp6 = tl.load(in_ptr3 + (r0), None)
tmp2 = tmp0 - tmp1
tmp3 = tmp2 * tmp2
tmp5 = tmp0 - tmp4
tmp7 = -tmp6
tmp8 = triton_helpers.maximum(tmp5, tmp7)
tmp9 = triton_helpers.minimum(tmp8, tmp6)
tmp10 = tmp4 + tmp9
tmp11 = tmp10 - tmp1
tmp12 = tmp11 * tmp11
tmp13 = triton_helpers.maximum(tmp3, tmp12)
tmp14 = tl.broadcast_to(tmp13, [RBLOCK])
tmp16 = triton_helpers.promote_to_tensor(tl.sum(tmp14, 0))
tmp17 = 256.0
tmp18 = tmp16 / tmp17
tmp19 = 0.5
tmp20 = tmp18 * tmp19
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp20, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg3_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [sub_1, pow_1, sub, neg, clamp, clipped_value, sub_2, pow_2, vf_loss, mean, mul], Original ATen: [aten.sub, aten.pow, aten.neg, aten.clamp, aten.add, aten.maximum, aten.mean, aten.mul]
stream0 = get_raw_stream(0)
triton_per_fused_add_clamp_maximum_mean_mul_neg_pow_sub_0.run(buf1, arg0_1, arg3_1, arg1_1, arg2_1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
del arg2_1
del arg3_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg2_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg3_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1, arg2_1, arg3_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class ClippedValueFunctionLoss(Module):
"""
## Clipped Value Function Loss
Similarly we clip the value function update also.
egin{align}
V^{\\pi_ heta}_{CLIP}(s_t)
&= clip\\Bigl(V^{\\pi_ heta}(s_t) - \\hat{V_t}, -\\epsilon, +\\epsilon\\Bigr)
\\
\\mathcal{L}^{VF}( heta)
&= rac{1}{2} \\mathbb{E} iggl[
max\\Bigl(igl(V^{\\pi_ heta}(s_t) - R_tigr)^2,
igl(V^{\\pi_ heta}_{CLIP}(s_t) - R_tigr)^2\\Bigr)
iggr]
\\end{align}
Clipping makes sure the value function $V_ heta$ doesn't deviate
significantly from $V_{ heta_{OLD}}$.
"""
def forward(self, value: 'torch.Tensor', sampled_value: 'torch.Tensor',
sampled_return: 'torch.Tensor', clip: 'float'):
clipped_value = sampled_value + (value - sampled_value).clamp(min=-
clip, max=clip)
vf_loss = torch.max((value - sampled_return) ** 2, (clipped_value -
sampled_return) ** 2)
return 0.5 * vf_loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_add_clamp_maximum_mean_mul_neg_pow_sub_0(in_out_ptr0,
in_ptr0, in_ptr1, in_ptr2, in_ptr3, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = tl.load(in_ptr1 + r0, None)
tmp4 = tl.load(in_ptr2 + r0, None)
tmp6 = tl.load(in_ptr3 + r0, None)
tmp2 = tmp0 - tmp1
tmp3 = tmp2 * tmp2
tmp5 = tmp0 - tmp4
tmp7 = -tmp6
tmp8 = triton_helpers.maximum(tmp5, tmp7)
tmp9 = triton_helpers.minimum(tmp8, tmp6)
tmp10 = tmp4 + tmp9
tmp11 = tmp10 - tmp1
tmp12 = tmp11 * tmp11
tmp13 = triton_helpers.maximum(tmp3, tmp12)
tmp14 = tl.broadcast_to(tmp13, [RBLOCK])
tmp16 = triton_helpers.promote_to_tensor(tl.sum(tmp14, 0))
tmp17 = 256.0
tmp18 = tmp16 / tmp17
tmp19 = 0.5
tmp20 = tmp18 * tmp19
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp20, None)
def call(args):
arg0_1, arg1_1, arg2_1, arg3_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg2_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg3_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_per_fused_add_clamp_maximum_mean_mul_neg_pow_sub_0[grid(1)](buf1
, arg0_1, arg3_1, arg1_1, arg2_1, 1, 256, num_warps=2, num_stages=1
)
del arg0_1
del arg1_1
del arg2_1
del arg3_1
return buf1,
class ClippedValueFunctionLossNew(Module):
"""
## Clipped Value Function Loss
Similarly we clip the value function update also.
egin{align}
V^{\\pi_ heta}_{CLIP}(s_t)
&= clip\\Bigl(V^{\\pi_ heta}(s_t) - \\hat{V_t}, -\\epsilon, +\\epsilon\\Bigr)
\\
\\mathcal{L}^{VF}( heta)
&= rac{1}{2} \\mathbb{E} iggl[
max\\Bigl(igl(V^{\\pi_ heta}(s_t) - R_tigr)^2,
igl(V^{\\pi_ heta}_{CLIP}(s_t) - R_tigr)^2\\Bigr)
iggr]
\\end{align}
Clipping makes sure the value function $V_ heta$ doesn't deviate
significantly from $V_{ heta_{OLD}}$.
"""
def forward(self, input_0, input_1, input_2, input_3):
arg0_1 = input_0
arg1_1 = input_1
arg2_1 = input_2
arg3_1 = input_3
output = call([arg0_1, arg1_1, arg2_1, arg3_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | ClippedValueFunctionLoss | false | 7,196 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Clipped Value Function Loss
Similarly we clip the value function update also.
egin{align}
V^{\\pi_ heta}_{CLIP}(s_t)
&= clip\\Bigl(V^{\\pi_ heta}(s_t) - \\hat{V_t}, -\\epsilon, +\\epsilon\\Bigr)
\\
\\mathcal{L}^{VF}( heta)
&= rac{1}{2} \\mathbb{E} iggl[
max\\Bigl(igl(V^{\\pi_ heta}(s_t) - R_tigr)^2,
igl(V^{\\pi_ heta}_{CLIP}(s_t) - R_tigr)^2\\Bigr)
iggr]
\\end{align}
Clipping makes sure the value function $V_ heta$ doesn't deviate
significantly from $V_{ heta_{OLD}}$.
"""
def forward(self, value: 'torch.Tensor', sampled_value: 'torch.Tensor',
sampled_return: 'torch.Tensor', clip: 'float'):
clipped_value = sampled_value + (value - sampled_value).clamp(min=-
clip, max=clip)
vf_loss = torch.max((value - sampled_return) ** 2, (clipped_value -
sampled_return) ** 2)
return 0.5 * vf_loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
CrossEntropyBayesRisk | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/f7/cf7j4pwlu55ah6fc4huqphitmbwk5lpjrsflp3g4iowoictloibu.py
# Topologically Sorted Source Nodes: [alpha, strength], Original ATen: [aten.add, aten.sum]
# Source node to ATen node mapping:
# alpha => add
# strength => sum_1
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, 1.0), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%add, [-1]), kwargs = {})
triton_poi_fused_add_sum_0 = async_compile.triton('triton_poi_fused_add_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp4 = tmp3 + tmp1
tmp5 = tmp2 + tmp4
tmp7 = tmp6 + tmp1
tmp8 = tmp5 + tmp7
tmp10 = tmp9 + tmp1
tmp11 = tmp8 + tmp10
tl.store(out_ptr0 + (x0), tmp11, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ng/cng3qs5i7l7nydt5poohpevomfjvgc4tpjnfgxynhzf2fuov5xde.py
# Topologically Sorted Source Nodes: [alpha], Original ATen: [aten.add]
# Source node to ATen node mapping:
# alpha => add
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, 1.0), kwargs = {})
triton_poi_fused_add_1 = async_compile.triton('triton_poi_fused_add_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/vn/cvnzeqhjxijtgpwsm3a57xbuknnkhkk7k3udzcqfvczkfyomgkmt.py
# Topologically Sorted Source Nodes: [sub, mul, loss, mean], Original ATen: [aten.sub, aten.mul, aten.sum, aten.mean]
# Source node to ATen node mapping:
# loss => sum_2
# mean => mean
# mul => mul
# sub => sub
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%unsqueeze, %digamma_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg1_1, %sub), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [-1]), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sum_2,), kwargs = {})
triton_per_fused_mean_mul_sub_sum_2 = async_compile.triton('triton_per_fused_mean_mul_sub_sum_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {4: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 5), equal_to_1=(4,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mean_mul_sub_sum_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 12, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mean_mul_sub_sum_2(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 4
r2 = (rindex // 16)
tmp0 = tl.load(in_ptr0 + (4*r3), None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + ((4*r0) + (16*r2)), None, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr2 + (4*r3), None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*r3)), None, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr1 + (1 + (4*r0) + (16*r2)), None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr2 + (1 + (4*r3)), None, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (2 + (4*r3)), None, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (2 + (4*r0) + (16*r2)), None, eviction_policy='evict_last')
tmp13 = tl.load(in_ptr2 + (2 + (4*r3)), None, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr0 + (3 + (4*r3)), None, eviction_policy='evict_last')
tmp18 = tl.load(in_ptr1 + (3 + (4*r0) + (16*r2)), None, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr2 + (3 + (4*r3)), None, eviction_policy='evict_last')
tmp3 = tmp1 - tmp2
tmp4 = tmp0 * tmp3
tmp8 = tmp6 - tmp7
tmp9 = tmp5 * tmp8
tmp10 = tmp4 + tmp9
tmp14 = tmp12 - tmp13
tmp15 = tmp11 * tmp14
tmp16 = tmp10 + tmp15
tmp20 = tmp18 - tmp19
tmp21 = tmp17 * tmp20
tmp22 = tmp16 + tmp21
tmp23 = tl.broadcast_to(tmp22, [XBLOCK, RBLOCK])
tmp25 = tl.sum(tmp23, 1)[:, None]
tmp26 = 64.0
tmp27 = tmp25 / tmp26
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp27, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [alpha, strength], Original ATen: [aten.add, aten.sum]
stream0 = get_raw_stream(0)
triton_poi_fused_add_sum_0.run(arg0_1, buf0, 64, grid=grid(64), stream=stream0)
# Topologically Sorted Source Nodes: [alpha, strength, digamma], Original ATen: [aten.add, aten.sum, aten.digamma]
buf1 = torch.ops.aten.digamma.default(buf0)
del buf0
buf2 = buf1
del buf1
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [alpha], Original ATen: [aten.add]
triton_poi_fused_add_1.run(arg0_1, buf3, 256, grid=grid(256), stream=stream0)
del arg0_1
# Topologically Sorted Source Nodes: [alpha, digamma_1], Original ATen: [aten.add, aten.digamma]
buf4 = torch.ops.aten.digamma.default(buf3)
del buf3
buf5 = buf4
del buf4
buf7 = empty_strided_cuda((), (), torch.float32)
buf8 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [sub, mul, loss, mean], Original ATen: [aten.sub, aten.mul, aten.sum, aten.mean]
triton_per_fused_mean_mul_sub_sum_2.run(buf8, arg1_1, buf2, buf5, 1, 64, grid=grid(1), stream=stream0)
del arg1_1
del buf2
del buf5
return (buf8, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class CrossEntropyBayesRisk(Module):
"""
<a id="CrossEntropyBayesRisk"></a>
## Bayes Risk with Cross Entropy Loss
Bayes risk is the overall maximum cost of making incorrect estimates.
It takes a cost function that gives the cost of making an incorrect estimate
and sums it over all possible outcomes based on probability distribution.
Here the cost function is cross-entropy loss, for one-hot coded $\\mathbf{y}$
$$\\sum_{k=1}^K -y_k \\log p_k$$
We integrate this cost over all $\\mathbf{p}$
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\Big[ \\sum_{k=1}^K -y_k \\log p_k \\Big]
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K y_k igg( \\psi(S) - \\psi( extcolor{orange}{lpha_k} ) igg)
\\end{align}
where $\\psi(\\cdot)$ is the $digamma$ function.
"""
def forward(self, evidence: 'torch.Tensor', target: 'torch.Tensor'):
"""
* `evidence` is $\\mathbf{e} \\ge 0$ with shape `[batch_size, n_classes]`
* `target` is $\\mathbf{y}$ with shape `[batch_size, n_classes]`
"""
alpha = evidence + 1.0
strength = alpha.sum(dim=-1)
loss = (target * (torch.digamma(strength)[:, None] - torch.digamma(
alpha))).sum(dim=-1)
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp4 = tmp3 + tmp1
tmp5 = tmp2 + tmp4
tmp7 = tmp6 + tmp1
tmp8 = tmp5 + tmp7
tmp10 = tmp9 + tmp1
tmp11 = tmp8 + tmp10
tl.store(out_ptr0 + x0, tmp11, xmask)
@triton.jit
def triton_poi_fused_add_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_per_fused_mean_mul_sub_sum_2(in_out_ptr0, in_ptr0, in_ptr1,
in_ptr2, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 4
r2 = rindex // 16
tmp0 = tl.load(in_ptr0 + 4 * r3, None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4 * r0 + 16 * r2), None, eviction_policy=
'evict_last')
tmp2 = tl.load(in_ptr2 + 4 * r3, None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * r3), None, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr1 + (1 + 4 * r0 + 16 * r2), None, eviction_policy=
'evict_last')
tmp7 = tl.load(in_ptr2 + (1 + 4 * r3), None, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (2 + 4 * r3), None, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (2 + 4 * r0 + 16 * r2), None, eviction_policy
='evict_last')
tmp13 = tl.load(in_ptr2 + (2 + 4 * r3), None, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr0 + (3 + 4 * r3), None, eviction_policy='evict_last')
tmp18 = tl.load(in_ptr1 + (3 + 4 * r0 + 16 * r2), None, eviction_policy
='evict_last')
tmp19 = tl.load(in_ptr2 + (3 + 4 * r3), None, eviction_policy='evict_last')
tmp3 = tmp1 - tmp2
tmp4 = tmp0 * tmp3
tmp8 = tmp6 - tmp7
tmp9 = tmp5 * tmp8
tmp10 = tmp4 + tmp9
tmp14 = tmp12 - tmp13
tmp15 = tmp11 * tmp14
tmp16 = tmp10 + tmp15
tmp20 = tmp18 - tmp19
tmp21 = tmp17 * tmp20
tmp22 = tmp16 + tmp21
tmp23 = tl.broadcast_to(tmp22, [XBLOCK, RBLOCK])
tmp25 = tl.sum(tmp23, 1)[:, None]
tmp26 = 64.0
tmp27 = tmp25 / tmp26
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp27, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_sum_0[grid(64)](arg0_1, buf0, 64, XBLOCK=64,
num_warps=1, num_stages=1)
buf1 = torch.ops.aten.digamma.default(buf0)
del buf0
buf2 = buf1
del buf1
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_add_1[grid(256)](arg0_1, buf3, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del arg0_1
buf4 = torch.ops.aten.digamma.default(buf3)
del buf3
buf5 = buf4
del buf4
buf7 = empty_strided_cuda((), (), torch.float32)
buf8 = buf7
del buf7
triton_per_fused_mean_mul_sub_sum_2[grid(1)](buf8, arg1_1, buf2,
buf5, 1, 64, XBLOCK=1, num_warps=2, num_stages=1)
del arg1_1
del buf2
del buf5
return buf8,
class CrossEntropyBayesRiskNew(Module):
"""
<a id="CrossEntropyBayesRisk"></a>
## Bayes Risk with Cross Entropy Loss
Bayes risk is the overall maximum cost of making incorrect estimates.
It takes a cost function that gives the cost of making an incorrect estimate
and sums it over all possible outcomes based on probability distribution.
Here the cost function is cross-entropy loss, for one-hot coded $\\mathbf{y}$
$$\\sum_{k=1}^K -y_k \\log p_k$$
We integrate this cost over all $\\mathbf{p}$
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\Big[ \\sum_{k=1}^K -y_k \\log p_k \\Big]
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K y_k igg( \\psi(S) - \\psi( extcolor{orange}{lpha_k} ) igg)
\\end{align}
where $\\psi(\\cdot)$ is the $digamma$ function.
"""
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | CrossEntropyBayesRisk | false | 7,197 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
<a id="CrossEntropyBayesRisk"></a>
## Bayes Risk with Cross Entropy Loss
Bayes risk is the overall maximum cost of making incorrect estimates.
It takes a cost function that gives the cost of making an incorrect estimate
and sums it over all possible outcomes based on probability distribution.
Here the cost function is cross-entropy loss, for one-hot coded $\\mathbf{y}$
$$\\sum_{k=1}^K -y_k \\log p_k$$
We integrate this cost over all $\\mathbf{p}$
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\Big[ \\sum_{k=1}^K -y_k \\log p_k \\Big]
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K y_k igg( \\psi(S) - \\psi( extcolor{orange}{lpha_k} ) igg)
\\end{align}
where $\\psi(\\cdot)$ is the $digamma$ function.
"""
def forward(self, evidence: 'torch.Tensor', target: 'torch.Tensor'):
"""
* `evidence` is $\\mathbf{e} \\ge 0$ with shape `[batch_size, n_classes]`
* `target` is $\\mathbf{y}$ with shape `[batch_size, n_classes]`
"""
alpha = evidence + 1.0
strength = alpha.sum(dim=-1)
loss = (target * (torch.digamma(strength)[:, None] - torch.digamma(
alpha))).sum(dim=-1)
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
DPFP | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/nk/cnkn4azjs7p23wvexd663q3licov6bhbhf2h5uynp4fgwugfkg6g.py
# Topologically Sorted Source Nodes: [cat, x, roll, k, sum_1, add, truediv], Original ATen: [aten.cat, aten.relu, aten.roll, aten.mul, aten.sum, aten.add, aten.div]
# Source node to ATen node mapping:
# add => add_1
# cat => cat
# k => mul
# roll => index
# sum_1 => sum_1
# truediv => div
# x => relu
# Graph fragment:
# %cat : [num_users=1] = call_function[target=torch.ops.aten.cat.default](args = ([%arg0_1, %neg], -1), kwargs = {})
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%cat,), kwargs = {})
# %index : [num_users=1] = call_function[target=torch.ops.aten.index.Tensor](args = (%relu, [None, None, None, %fmod]), kwargs = {})
# %mul : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%relu, %index), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [-1], True), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sum_1, 1e-06), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul, %add_1), kwargs = {})
triton_per_fused_add_cat_div_mul_relu_roll_sum_0 = async_compile.triton('triton_per_fused_add_cat_div_mul_relu_roll_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[64, 8],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_cat_div_mul_relu_roll_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_cat_div_mul_relu_roll_sum_0(in_ptr0, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 64
rnumel = 8
RBLOCK: tl.constexpr = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = r1
tmp1 = tl.full([1, 1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1, 1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + ((4*x0) + r1), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tmp7 = tl.full([1, 1], 8, tl.int64)
tmp8 = tmp0 < tmp7
tmp9 = tl.load(in_ptr0 + ((4*x0) + ((-4) + r1)), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp10 = -tmp9
tmp11 = tl.full(tmp10.shape, 0.0, tmp10.dtype)
tmp12 = tl.where(tmp6, tmp10, tmp11)
tmp13 = tl.where(tmp4, tmp5, tmp12)
tmp14 = tl.full([1, 1], 0, tl.int32)
tmp15 = triton_helpers.maximum(tmp14, tmp13)
tmp16 = (7 + r1) % 8
tmp17 = tmp16 >= tmp1
tmp18 = tmp16 < tmp3
tmp19 = tl.load(in_ptr0 + ((4*x0) + ((7 + r1) % 8)), tmp18 & xmask, eviction_policy='evict_last', other=0.0)
tmp20 = tmp16 >= tmp3
tmp21 = tmp16 < tmp7
tmp22 = tl.load(in_ptr0 + ((4*x0) + ((-4) + ((7 + r1) % 8))), tmp20 & xmask, eviction_policy='evict_last', other=0.0)
tmp23 = -tmp22
tmp24 = tl.full(tmp23.shape, 0.0, tmp23.dtype)
tmp25 = tl.where(tmp20, tmp23, tmp24)
tmp26 = tl.where(tmp18, tmp19, tmp25)
tmp27 = triton_helpers.maximum(tmp14, tmp26)
tmp28 = tmp15 * tmp27
tmp29 = tl.broadcast_to(tmp28, [XBLOCK, RBLOCK])
tmp31 = tl.where(xmask, tmp29, 0)
tmp32 = tl.sum(tmp31, 1)[:, None]
tmp33 = 1e-06
tmp34 = tmp32 + tmp33
tmp35 = tmp28 / tmp34
tl.store(out_ptr2 + (r1 + (8*x0)), tmp35, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf2 = empty_strided_cuda((4, 4, 4, 8), (128, 32, 8, 1), torch.float32)
# Topologically Sorted Source Nodes: [cat, x, roll, k, sum_1, add, truediv], Original ATen: [aten.cat, aten.relu, aten.roll, aten.mul, aten.sum, aten.add, aten.div]
stream0 = get_raw_stream(0)
triton_per_fused_add_cat_div_mul_relu_roll_sum_0.run(arg0_1, buf2, 64, 8, grid=grid(64), stream=stream0)
del arg0_1
return (buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class DPFP(Module):
"""
## Deterministic Parameter Free Project (DPFP)
This is the new projection function $ extcolor{lightgreen}{\\phi}$ introduced in the paper.
DPFP projects $k$ of dimensionality $d_{key}$ to dimensionality $d_{dot} = 2 d_{key}
u$,
where $
u \\in \\{1, 2, ..., 2 d_{key} - 1 \\}$ is a hyper-parameter.
$$ extcolor{lightgreen}{\\phi_{2 d_{key} (i - 1) + j}(k)}
= ext{ReLU}\\Big(ig[k, -kig]\\Big)_{j}
ext{ReLU}\\Big(ig[k, -kig]\\Big)_{i + j}$$
where $ig[k, -kig]$ is the concatenation of $k$ and $-k$ to give a vector of
size $2 d_{key}$, $i \\in \\{1, 2, ...,
u \\}$, and $j \\in \\{1, 2, ..., 2 d_{key}\\}$.
$x_i$ is the $i$-th element of vector $x$ and is rolled around if
$i$ is larger than the number of elements in $x$.
Basically, it creates a new vector by multiplying elements of $[k, -k]$ shifted by $i$.
This produces projections that are sparse (only a few elements of $phi$ are non-zero) and
orthogonal ($ extcolor{lightgreen}{\\phi(k^{(i)})} \\cdot extcolor{lightgreen}{\\phi(k^{(j)})}
pprox 0$ for most $i, j$
unless $k^{(i)}$ and $k^{(j)}$ are very similar.
### Normalization
Paper introduces a simple normalization for $ extcolor{lightgreen}{\\phi}$,
$$ extcolor{lightgreen}{\\phi '(k)} =
rac{ extcolor{lightgreen}{\\phi(k)}}{\\sum^{d_{dot}}_{j=1} extcolor{lightgreen}{\\phi(k)_j}}$$
*Check the paper for derivation.*
"""
def __init__(self, nu: 'int'=1, eps: 'float'=1e-06):
"""
* `nu` is the hyper-parameter $
u$.
* `eps` is the small value used to make sure there is no division-by-zero when normalizing.
"""
super().__init__()
self.nu = nu
self.relu = nn.ReLU()
self.eps = eps
def forward(self, k: 'torch.Tensor'):
k = self.dpfp(k)
return k / (torch.sum(k, dim=-1, keepdim=True) + self.eps)
def dpfp(self, k: 'torch.Tensor'):
"""
$$ extcolor{lightgreen}{\\phi(k)}$$
"""
x = self.relu(torch.cat([k, -k], dim=-1))
x_rolled = [x.roll(shifts=i, dims=-1) for i in range(1, self.nu + 1)]
x_rolled = torch.cat(x_rolled, dim=-1)
x_repeat = torch.cat([x] * self.nu, dim=-1)
return x_repeat * x_rolled
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_add_cat_div_mul_relu_roll_sum_0(in_ptr0, out_ptr2,
xnumel, rnumel, XBLOCK: tl.constexpr):
xnumel = 64
RBLOCK: tl.constexpr = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r1 = rindex
x0 = xindex
tmp0 = r1
tl.full([1, 1], 0, tl.int64)
tmp3 = tl.full([1, 1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tl.load(in_ptr0 + (4 * x0 + r1), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp6 = tmp0 >= tmp3
tl.full([1, 1], 8, tl.int64)
tmp9 = tl.load(in_ptr0 + (4 * x0 + (-4 + r1)), tmp6 & xmask,
eviction_policy='evict_last', other=0.0)
tmp10 = -tmp9
tmp11 = tl.full(tmp10.shape, 0.0, tmp10.dtype)
tmp12 = tl.where(tmp6, tmp10, tmp11)
tmp13 = tl.where(tmp4, tmp5, tmp12)
tmp14 = tl.full([1, 1], 0, tl.int32)
tmp15 = triton_helpers.maximum(tmp14, tmp13)
tmp16 = (7 + r1) % 8
tmp18 = tmp16 < tmp3
tmp19 = tl.load(in_ptr0 + (4 * x0 + (7 + r1) % 8), tmp18 & xmask,
eviction_policy='evict_last', other=0.0)
tmp20 = tmp16 >= tmp3
tmp22 = tl.load(in_ptr0 + (4 * x0 + (-4 + (7 + r1) % 8)), tmp20 & xmask,
eviction_policy='evict_last', other=0.0)
tmp23 = -tmp22
tmp24 = tl.full(tmp23.shape, 0.0, tmp23.dtype)
tmp25 = tl.where(tmp20, tmp23, tmp24)
tmp26 = tl.where(tmp18, tmp19, tmp25)
tmp27 = triton_helpers.maximum(tmp14, tmp26)
tmp28 = tmp15 * tmp27
tmp29 = tl.broadcast_to(tmp28, [XBLOCK, RBLOCK])
tmp31 = tl.where(xmask, tmp29, 0)
tmp32 = tl.sum(tmp31, 1)[:, None]
tmp33 = 1e-06
tmp34 = tmp32 + tmp33
tmp35 = tmp28 / tmp34
tl.store(out_ptr2 + (r1 + 8 * x0), tmp35, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf2 = empty_strided_cuda((4, 4, 4, 8), (128, 32, 8, 1), torch.float32)
get_raw_stream(0)
triton_per_fused_add_cat_div_mul_relu_roll_sum_0[grid(64)](arg0_1,
buf2, 64, 8, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
return buf2,
class DPFPNew(Module):
"""
## Deterministic Parameter Free Project (DPFP)
This is the new projection function $ extcolor{lightgreen}{\\phi}$ introduced in the paper.
DPFP projects $k$ of dimensionality $d_{key}$ to dimensionality $d_{dot} = 2 d_{key}
u$,
where $
u \\in \\{1, 2, ..., 2 d_{key} - 1 \\}$ is a hyper-parameter.
$$ extcolor{lightgreen}{\\phi_{2 d_{key} (i - 1) + j}(k)}
= ext{ReLU}\\Big(ig[k, -kig]\\Big)_{j}
ext{ReLU}\\Big(ig[k, -kig]\\Big)_{i + j}$$
where $ig[k, -kig]$ is the concatenation of $k$ and $-k$ to give a vector of
size $2 d_{key}$, $i \\in \\{1, 2, ...,
u \\}$, and $j \\in \\{1, 2, ..., 2 d_{key}\\}$.
$x_i$ is the $i$-th element of vector $x$ and is rolled around if
$i$ is larger than the number of elements in $x$.
Basically, it creates a new vector by multiplying elements of $[k, -k]$ shifted by $i$.
This produces projections that are sparse (only a few elements of $phi$ are non-zero) and
orthogonal ($ extcolor{lightgreen}{\\phi(k^{(i)})} \\cdot extcolor{lightgreen}{\\phi(k^{(j)})}
pprox 0$ for most $i, j$
unless $k^{(i)}$ and $k^{(j)}$ are very similar.
### Normalization
Paper introduces a simple normalization for $ extcolor{lightgreen}{\\phi}$,
$$ extcolor{lightgreen}{\\phi '(k)} =
rac{ extcolor{lightgreen}{\\phi(k)}}{\\sum^{d_{dot}}_{j=1} extcolor{lightgreen}{\\phi(k)_j}}$$
*Check the paper for derivation.*
"""
def __init__(self, nu: 'int'=1, eps: 'float'=1e-06):
"""
* `nu` is the hyper-parameter $
u$.
* `eps` is the small value used to make sure there is no division-by-zero when normalizing.
"""
super().__init__()
self.nu = nu
self.relu = nn.ReLU()
self.eps = eps
def dpfp(self, k: 'torch.Tensor'):
"""
$$ extcolor{lightgreen}{\\phi(k)}$$
"""
x = self.relu(torch.cat([k, -k], dim=-1))
x_rolled = [x.roll(shifts=i, dims=-1) for i in range(1, self.nu + 1)]
x_rolled = torch.cat(x_rolled, dim=-1)
x_repeat = torch.cat([x] * self.nu, dim=-1)
return x_repeat * x_rolled
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | DPFP | false | 7,198 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Deterministic Parameter Free Project (DPFP)
This is the new projection function $ extcolor{lightgreen}{\\phi}$ introduced in the paper.
DPFP projects $k$ of dimensionality $d_{key}$ to dimensionality $d_{dot} = 2 d_{key}
u$,
where $
u \\in \\{1, 2, ..., 2 d_{key} - 1 \\}$ is a hyper-parameter.
$$ extcolor{lightgreen}{\\phi_{2 d_{key} (i - 1) + j}(k)}
= ext{ReLU}\\Big(ig[k, -kig]\\Big)_{j}
ext{ReLU}\\Big(ig[k, -kig]\\Big)_{i + j}$$
where $ig[k, -kig]$ is the concatenation of $k$ and $-k$ to give a vector of
size $2 d_{key}$, $i \\in \\{1, 2, ...,
u \\}$, and $j \\in \\{1, 2, ..., 2 d_{key}\\}$.
$x_i$ is the $i$-th element of vector $x$ and is rolled around if
$i$ is larger than the number of elements in $x$.
Basically, it creates a new vector by multiplying elements of $[k, -k]$ shifted by $i$.
This produces projections that are sparse (only a few elements of $phi$ are non-zero) and
orthogonal ($ extcolor{lightgreen}{\\phi(k^{(i)})} \\cdot extcolor{lightgreen}{\\phi(k^{(j)})}
pprox 0$ for most $i, j$
unless $k^{(i)}$ and $k^{(j)}$ are very similar.
### Normalization
Paper introduces a simple normalization for $ extcolor{lightgreen}{\\phi}$,
$$ extcolor{lightgreen}{\\phi '(k)} =
rac{ extcolor{lightgreen}{\\phi(k)}}{\\sum^{d_{dot}}_{j=1} extcolor{lightgreen}{\\phi(k)_j}}$$
*Check the paper for derivation.*
"""
def __init__(self, nu: 'int'=1, eps: 'float'=1e-06):
"""
* `nu` is the hyper-parameter $
u$.
* `eps` is the small value used to make sure there is no division-by-zero when normalizing.
"""
super().__init__()
self.nu = nu
self.relu = nn.ReLU()
self.eps = eps
def forward(self, k: 'torch.Tensor'):
k = self.dpfp(k)
return k / (torch.sum(k, dim=-1, keepdim=True) + self.eps)
def dpfp(self, k: 'torch.Tensor'):
"""
$$ extcolor{lightgreen}{\\phi(k)}$$
"""
x = self.relu(torch.cat([k, -k], dim=-1))
x_rolled = [x.roll(shifts=i, dims=-1) for i in range(1, self.nu + 1)]
x_rolled = torch.cat(x_rolled, dim=-1)
x_repeat = torch.cat([x] * self.nu, dim=-1)
return x_repeat * x_rolled
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
KLDivLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/fh/cfhl4tf45osircr7c3fcnkkmm63hrxmrewn65xg36whopnb4upq6.py
# Topologically Sorted Source Nodes: [add, pow_1, sub, exp, sub_1, mean, mul], Original ATen: [aten.add, aten.pow, aten.sub, aten.exp, aten.mean, aten.mul]
# Source node to ATen node mapping:
# add => add
# exp => exp
# mean => mean
# mul => mul
# pow_1 => pow_1
# sub => sub
# sub_1 => sub_1
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, 1), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%arg1_1, 2), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %pow_1), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%arg0_1,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub, %exp), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sub_1,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mean, -0.5), kwargs = {})
triton_per_fused_add_exp_mean_mul_pow_sub_0 = async_compile.triton('triton_per_fused_add_exp_mean_mul_pow_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_exp_mean_mul_pow_sub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 2, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_exp_mean_mul_pow_sub_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp3 = tl.load(in_ptr1 + (r0), None)
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 - tmp4
tmp6 = tl_math.exp(tmp0)
tmp7 = tmp5 - tmp6
tmp8 = tl.broadcast_to(tmp7, [RBLOCK])
tmp10 = triton_helpers.promote_to_tensor(tl.sum(tmp8, 0))
tmp11 = 256.0
tmp12 = tmp10 / tmp11
tmp13 = -0.5
tmp14 = tmp12 * tmp13
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp14, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [add, pow_1, sub, exp, sub_1, mean, mul], Original ATen: [aten.add, aten.pow, aten.sub, aten.exp, aten.mean, aten.mul]
stream0 = get_raw_stream(0)
triton_per_fused_add_exp_mean_mul_pow_sub_0.run(buf1, arg0_1, arg1_1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class KLDivLoss(Module):
"""
## KL-Divergence loss
This calculates the KL divergence between a given normal distribution and $\\mathcal{N}(0, 1)$
"""
def forward(self, sigma_hat: 'torch.Tensor', mu: 'torch.Tensor'):
return -0.5 * torch.mean(1 + sigma_hat - mu ** 2 - torch.exp(sigma_hat)
)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_add_exp_mean_mul_pow_sub_0(in_out_ptr0, in_ptr0,
in_ptr1, xnumel, rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp3 = tl.load(in_ptr1 + r0, None)
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 - tmp4
tmp6 = tl_math.exp(tmp0)
tmp7 = tmp5 - tmp6
tmp8 = tl.broadcast_to(tmp7, [RBLOCK])
tmp10 = triton_helpers.promote_to_tensor(tl.sum(tmp8, 0))
tmp11 = 256.0
tmp12 = tmp10 / tmp11
tmp13 = -0.5
tmp14 = tmp12 * tmp13
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp14, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((), (), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_per_fused_add_exp_mean_mul_pow_sub_0[grid(1)](buf1, arg0_1,
arg1_1, 1, 256, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf1,
class KLDivLossNew(Module):
"""
## KL-Divergence loss
This calculates the KL divergence between a given normal distribution and $\\mathcal{N}(0, 1)$
"""
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | KLDivLoss | false | 7,199 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## KL-Divergence loss
This calculates the KL divergence between a given normal distribution and $\\mathcal{N}(0, 1)$
"""
def forward(self, sigma_hat: 'torch.Tensor', mu: 'torch.Tensor'):
return -0.5 * torch.mean(1 + sigma_hat - mu ** 2 - torch.exp(sigma_hat)
)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
MaximumLikelihoodLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/f6/cf6outqezdomcxvjwbjchfmsgy45k2q6wtv3oau6yri4t3ezi6d7.py
# Topologically Sorted Source Nodes: [alpha, log_1, sub, mul, loss, mean], Original ATen: [aten.add, aten.log, aten.sub, aten.mul, aten.sum, aten.mean]
# Source node to ATen node mapping:
# alpha => add
# log_1 => log_1
# loss => sum_2
# mean => mean
# mul => mul
# sub => sub
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, 1.0), kwargs = {})
# %log_1 : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%add,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%unsqueeze, %log_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg1_1, %sub), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul, [-1]), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sum_2,), kwargs = {})
triton_per_fused_add_log_mean_mul_sub_sum_0 = async_compile.triton('triton_per_fused_add_log_mean_mul_sub_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {3: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 4), equal_to_1=(3,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_log_mean_mul_sub_sum_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 24, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_log_mean_mul_sub_sum_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 4
r2 = (rindex // 16)
tmp0 = tl.load(in_ptr0 + (4*r3), None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + ((16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr1 + (2 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr1 + (3 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr1 + (4*r3), None, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr0 + (1 + (4*r3)), None, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr1 + (4 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr1 + (5 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr1 + (6 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp28 = tl.load(in_ptr1 + (7 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp32 = tl.load(in_ptr1 + (1 + (4*r3)), None, eviction_policy='evict_last')
tmp38 = tl.load(in_ptr0 + (2 + (4*r3)), None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr1 + (8 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp41 = tl.load(in_ptr1 + (9 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp44 = tl.load(in_ptr1 + (10 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp47 = tl.load(in_ptr1 + (11 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp51 = tl.load(in_ptr1 + (2 + (4*r3)), None, eviction_policy='evict_last')
tmp57 = tl.load(in_ptr0 + (3 + (4*r3)), None, eviction_policy='evict_last')
tmp58 = tl.load(in_ptr1 + (12 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp60 = tl.load(in_ptr1 + (13 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp63 = tl.load(in_ptr1 + (14 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp66 = tl.load(in_ptr1 + (15 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp70 = tl.load(in_ptr1 + (3 + (4*r3)), None, eviction_policy='evict_last')
tmp2 = 1.0
tmp3 = tmp1 + tmp2
tmp5 = tmp4 + tmp2
tmp6 = tmp3 + tmp5
tmp8 = tmp7 + tmp2
tmp9 = tmp6 + tmp8
tmp11 = tmp10 + tmp2
tmp12 = tmp9 + tmp11
tmp13 = tl_math.log(tmp12)
tmp15 = tmp14 + tmp2
tmp16 = tl_math.log(tmp15)
tmp17 = tmp13 - tmp16
tmp18 = tmp0 * tmp17
tmp21 = tmp20 + tmp2
tmp23 = tmp22 + tmp2
tmp24 = tmp21 + tmp23
tmp26 = tmp25 + tmp2
tmp27 = tmp24 + tmp26
tmp29 = tmp28 + tmp2
tmp30 = tmp27 + tmp29
tmp31 = tl_math.log(tmp30)
tmp33 = tmp32 + tmp2
tmp34 = tl_math.log(tmp33)
tmp35 = tmp31 - tmp34
tmp36 = tmp19 * tmp35
tmp37 = tmp18 + tmp36
tmp40 = tmp39 + tmp2
tmp42 = tmp41 + tmp2
tmp43 = tmp40 + tmp42
tmp45 = tmp44 + tmp2
tmp46 = tmp43 + tmp45
tmp48 = tmp47 + tmp2
tmp49 = tmp46 + tmp48
tmp50 = tl_math.log(tmp49)
tmp52 = tmp51 + tmp2
tmp53 = tl_math.log(tmp52)
tmp54 = tmp50 - tmp53
tmp55 = tmp38 * tmp54
tmp56 = tmp37 + tmp55
tmp59 = tmp58 + tmp2
tmp61 = tmp60 + tmp2
tmp62 = tmp59 + tmp61
tmp64 = tmp63 + tmp2
tmp65 = tmp62 + tmp64
tmp67 = tmp66 + tmp2
tmp68 = tmp65 + tmp67
tmp69 = tl_math.log(tmp68)
tmp71 = tmp70 + tmp2
tmp72 = tl_math.log(tmp71)
tmp73 = tmp69 - tmp72
tmp74 = tmp57 * tmp73
tmp75 = tmp56 + tmp74
tmp76 = tl.broadcast_to(tmp75, [XBLOCK, RBLOCK])
tmp78 = tl.sum(tmp76, 1)[:, None]
tmp79 = 64.0
tmp80 = tmp78 / tmp79
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp80, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [alpha, log_1, sub, mul, loss, mean], Original ATen: [aten.add, aten.log, aten.sub, aten.mul, aten.sum, aten.mean]
stream0 = get_raw_stream(0)
triton_per_fused_add_log_mean_mul_sub_sum_0.run(buf2, arg1_1, arg0_1, 1, 64, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
return (buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class MaximumLikelihoodLoss(Module):
"""
<a id="MaximumLikelihoodLoss"></a>
## Type II Maximum Likelihood Loss
The distribution $D(\\mathbf{p} ert extcolor{orange}{\\mathbf{lpha}})$ is a prior on the likelihood
$Multi(\\mathbf{y} ert p)$,
and the negative log marginal likelihood is calculated by integrating over class probabilities
$\\mathbf{p}$.
If target probabilities (one-hot targets) are $y_k$ for a given sample the loss is,
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\prod_{k=1}^K p_k^{y_k}
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K y_k igg( \\log S - \\log extcolor{orange}{lpha_k} igg)
\\end{align}
"""
def forward(self, evidence: 'torch.Tensor', target: 'torch.Tensor'):
"""
* `evidence` is $\\mathbf{e} \\ge 0$ with shape `[batch_size, n_classes]`
* `target` is $\\mathbf{y}$ with shape `[batch_size, n_classes]`
"""
alpha = evidence + 1.0
strength = alpha.sum(dim=-1)
loss = (target * (strength.log()[:, None] - alpha.log())).sum(dim=-1)
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_add_log_mean_mul_sub_sum_0(in_out_ptr0, in_ptr0,
in_ptr1, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 4
r2 = rindex // 16
tmp0 = tl.load(in_ptr0 + 4 * r3, None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (16 * r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 16 * r0 + 64 * r2), None, eviction_policy
='evict_last')
tmp7 = tl.load(in_ptr1 + (2 + 16 * r0 + 64 * r2), None, eviction_policy
='evict_last')
tmp10 = tl.load(in_ptr1 + (3 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp14 = tl.load(in_ptr1 + 4 * r3, None, eviction_policy='evict_last')
tmp19 = tl.load(in_ptr0 + (1 + 4 * r3), None, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr1 + (4 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp22 = tl.load(in_ptr1 + (5 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp25 = tl.load(in_ptr1 + (6 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp28 = tl.load(in_ptr1 + (7 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp32 = tl.load(in_ptr1 + (1 + 4 * r3), None, eviction_policy='evict_last')
tmp38 = tl.load(in_ptr0 + (2 + 4 * r3), None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr1 + (8 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp41 = tl.load(in_ptr1 + (9 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp44 = tl.load(in_ptr1 + (10 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp47 = tl.load(in_ptr1 + (11 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp51 = tl.load(in_ptr1 + (2 + 4 * r3), None, eviction_policy='evict_last')
tmp57 = tl.load(in_ptr0 + (3 + 4 * r3), None, eviction_policy='evict_last')
tmp58 = tl.load(in_ptr1 + (12 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp60 = tl.load(in_ptr1 + (13 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp63 = tl.load(in_ptr1 + (14 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp66 = tl.load(in_ptr1 + (15 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp70 = tl.load(in_ptr1 + (3 + 4 * r3), None, eviction_policy='evict_last')
tmp2 = 1.0
tmp3 = tmp1 + tmp2
tmp5 = tmp4 + tmp2
tmp6 = tmp3 + tmp5
tmp8 = tmp7 + tmp2
tmp9 = tmp6 + tmp8
tmp11 = tmp10 + tmp2
tmp12 = tmp9 + tmp11
tmp13 = tl_math.log(tmp12)
tmp15 = tmp14 + tmp2
tmp16 = tl_math.log(tmp15)
tmp17 = tmp13 - tmp16
tmp18 = tmp0 * tmp17
tmp21 = tmp20 + tmp2
tmp23 = tmp22 + tmp2
tmp24 = tmp21 + tmp23
tmp26 = tmp25 + tmp2
tmp27 = tmp24 + tmp26
tmp29 = tmp28 + tmp2
tmp30 = tmp27 + tmp29
tmp31 = tl_math.log(tmp30)
tmp33 = tmp32 + tmp2
tmp34 = tl_math.log(tmp33)
tmp35 = tmp31 - tmp34
tmp36 = tmp19 * tmp35
tmp37 = tmp18 + tmp36
tmp40 = tmp39 + tmp2
tmp42 = tmp41 + tmp2
tmp43 = tmp40 + tmp42
tmp45 = tmp44 + tmp2
tmp46 = tmp43 + tmp45
tmp48 = tmp47 + tmp2
tmp49 = tmp46 + tmp48
tmp50 = tl_math.log(tmp49)
tmp52 = tmp51 + tmp2
tmp53 = tl_math.log(tmp52)
tmp54 = tmp50 - tmp53
tmp55 = tmp38 * tmp54
tmp56 = tmp37 + tmp55
tmp59 = tmp58 + tmp2
tmp61 = tmp60 + tmp2
tmp62 = tmp59 + tmp61
tmp64 = tmp63 + tmp2
tmp65 = tmp62 + tmp64
tmp67 = tmp66 + tmp2
tmp68 = tmp65 + tmp67
tmp69 = tl_math.log(tmp68)
tmp71 = tmp70 + tmp2
tmp72 = tl_math.log(tmp71)
tmp73 = tmp69 - tmp72
tmp74 = tmp57 * tmp73
tmp75 = tmp56 + tmp74
tmp76 = tl.broadcast_to(tmp75, [XBLOCK, RBLOCK])
tmp78 = tl.sum(tmp76, 1)[:, None]
tmp79 = 64.0
tmp80 = tmp78 / tmp79
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp80, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1
del buf1
get_raw_stream(0)
triton_per_fused_add_log_mean_mul_sub_sum_0[grid(1)](buf2, arg1_1,
arg0_1, 1, 64, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
return buf2,
class MaximumLikelihoodLossNew(Module):
"""
<a id="MaximumLikelihoodLoss"></a>
## Type II Maximum Likelihood Loss
The distribution $D(\\mathbf{p} ert extcolor{orange}{\\mathbf{lpha}})$ is a prior on the likelihood
$Multi(\\mathbf{y} ert p)$,
and the negative log marginal likelihood is calculated by integrating over class probabilities
$\\mathbf{p}$.
If target probabilities (one-hot targets) are $y_k$ for a given sample the loss is,
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\prod_{k=1}^K p_k^{y_k}
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K y_k igg( \\log S - \\log extcolor{orange}{lpha_k} igg)
\\end{align}
"""
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | MaximumLikelihoodLoss | false | 7,200 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
<a id="MaximumLikelihoodLoss"></a>
## Type II Maximum Likelihood Loss
The distribution $D(\\mathbf{p} ert extcolor{orange}{\\mathbf{lpha}})$ is a prior on the likelihood
$Multi(\\mathbf{y} ert p)$,
and the negative log marginal likelihood is calculated by integrating over class probabilities
$\\mathbf{p}$.
If target probabilities (one-hot targets) are $y_k$ for a given sample the loss is,
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\prod_{k=1}^K p_k^{y_k}
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K y_k igg( \\log S - \\log extcolor{orange}{lpha_k} igg)
\\end{align}
"""
def forward(self, evidence: 'torch.Tensor', target: 'torch.Tensor'):
"""
* `evidence` is $\\mathbf{e} \\ge 0$ with shape `[batch_size, n_classes]`
* `target` is $\\mathbf{y}$ with shape `[batch_size, n_classes]`
"""
alpha = evidence + 1.0
strength = alpha.sum(dim=-1)
loss = (target * (strength.log()[:, None] - alpha.log())).sum(dim=-1)
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
FCVAE | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/3q/c3qwr2d2rrpjzvnddomnmdy6cwva4hjlvrn2y5epemk4ak3k2m6c.py
# Topologically Sorted Source Nodes: [h1], Original ATen: [aten.relu]
# Source node to ATen node mapping:
# h1 => relu
# Graph fragment:
# %add_tensor_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default_2, %primals_3), kwargs = {})
# %relu : [num_users=3] = call_function[target=torch.ops.aten.relu.default](args = (%add_tensor_2,), kwargs = {})
triton_poi_fused_relu_0 = async_compile.triton('triton_poi_fused_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 400
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/zy/czy7u5vco7t6hswbiwnvj7w5efucfccq2eg7jymbbwipxow53f6t.py
# Topologically Sorted Source Nodes: [mul, std, mul_1, z], Original ATen: [aten.mul, aten.exp, aten.add]
# Source node to ATen node mapping:
# mul => mul
# mul_1 => mul_1
# std => exp
# z => add
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%addmm_2, 0.5), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%mul,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%randn, %exp), kwargs = {})
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%addmm_1, %mul_1), kwargs = {})
triton_poi_fused_add_exp_mul_1 = async_compile.triton('triton_poi_fused_add_exp_mul_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_exp_mul_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_exp_mul_1(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask)
tmp2 = tl.load(in_ptr2 + (x0), xmask)
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp1 * tmp5
tmp7 = tmp0 + tmp6
tl.store(out_ptr0 + (x0), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/eq/ceqgaaddmc4caykkz2jef7xzjkszeswmh226pkrsflrqpf5nymaa.py
# Topologically Sorted Source Nodes: [sigmoid], Original ATen: [aten.sigmoid, aten.sigmoid_backward]
# Source node to ATen node mapping:
# sigmoid => sigmoid
# Graph fragment:
# %add_tensor : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mm_default, %primals_11), kwargs = {})
# %sigmoid : [num_users=3] = call_function[target=torch.ops.aten.sigmoid.default](args = (%add_tensor,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %sigmoid), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid, %sub), kwargs = {})
triton_poi_fused_sigmoid_sigmoid_backward_2 = async_compile.triton('triton_poi_fused_sigmoid_sigmoid_backward_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[4096],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sigmoid_sigmoid_backward_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sigmoid_sigmoid_backward_2(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 3136
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 784
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.sigmoid(tmp2)
tmp4 = 1.0
tmp5 = tmp4 - tmp3
tmp6 = tmp3 * tmp5
tl.store(in_out_ptr0 + (x2), tmp3, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11 = args
args.clear()
assert_size_stride(primals_1, (4, 784), (784, 1))
assert_size_stride(primals_2, (400, 784), (784, 1))
assert_size_stride(primals_3, (400, ), (1, ))
assert_size_stride(primals_4, (2, 400), (400, 1))
assert_size_stride(primals_5, (2, ), (1, ))
assert_size_stride(primals_6, (2, 400), (400, 1))
assert_size_stride(primals_7, (2, ), (1, ))
assert_size_stride(primals_8, (400, 2), (2, 1))
assert_size_stride(primals_9, (400, ), (1, ))
assert_size_stride(primals_10, (784, 400), (400, 1))
assert_size_stride(primals_11, (784, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(primals_1, reinterpret_tensor(primals_2, (784, 400), (1, 784), 0), out=buf0)
del primals_2
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [h1], Original ATen: [aten.relu]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_0.run(buf1, primals_3, 1600, grid=grid(1600), stream=stream0)
del primals_3
buf2 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [mu], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, buf1, reinterpret_tensor(primals_4, (400, 2), (1, 400), 0), alpha=1, beta=1, out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [logvar], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_7, buf1, reinterpret_tensor(primals_6, (400, 2), (1, 400), 0), alpha=1, beta=1, out=buf3)
del primals_7
# Topologically Sorted Source Nodes: [eps], Original ATen: [aten.randn_like]
buf4 = torch.ops.aten.randn.default([4, 2], dtype=torch.float32, device=device(type='cuda', index=0), pin_memory=False)
buf5 = buf4
del buf4
buf6 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, std, mul_1, z], Original ATen: [aten.mul, aten.exp, aten.add]
triton_poi_fused_add_exp_mul_1.run(buf2, buf5, buf3, buf6, 8, grid=grid(8), stream=stream0)
buf7 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf6, reinterpret_tensor(primals_8, (2, 400), (1, 2), 0), out=buf7)
buf8 = buf7; del buf7 # reuse
# Topologically Sorted Source Nodes: [h3], Original ATen: [aten.relu]
triton_poi_fused_relu_0.run(buf8, primals_9, 1600, grid=grid(1600), stream=stream0)
del primals_9
buf9 = empty_strided_cuda((4, 784), (784, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(buf8, reinterpret_tensor(primals_10, (400, 784), (1, 400), 0), out=buf9)
buf10 = buf9; del buf9 # reuse
buf11 = empty_strided_cuda((4, 784), (784, 1), torch.float32)
# Topologically Sorted Source Nodes: [sigmoid], Original ATen: [aten.sigmoid, aten.sigmoid_backward]
triton_poi_fused_sigmoid_sigmoid_backward_2.run(buf10, primals_11, buf11, 3136, grid=grid(3136), stream=stream0)
del primals_11
return (reinterpret_tensor(buf10, (4, 1, 28, 28), (784, 784, 28, 1), 0), buf2, buf3, primals_1, buf1, buf3, buf5, buf6, buf8, buf11, primals_10, primals_8, primals_6, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 784), (784, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((400, 784), (784, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((400, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((2, 400), (400, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((2, 400), (400, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((400, 2), (2, 1), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((400, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((784, 400), (400, 1), device='cuda:0', dtype=torch.float32)
primals_11 = rand_strided((784, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10, primals_11])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch.nn import functional as F
from torch import nn
class BaseVAE(nn.Module):
"""
Base abstract class for the Variational Autoencoders
"""
def __init__(self, channels=1, width=28, height=28, z_dim=2):
"""
Constructor
Parameters:
channels - The number of channels for the image
width - The width of the image in pixels
height - The height of the image in pixels
z_dim - The dimension of the latent space
"""
super(BaseVAE, self).__init__()
self.channels = channels
self.width = width
self.height = height
self.z_dim = z_dim
def getNbChannels(self):
"""
Returns the number of channels of the handled images
"""
return self.channels
def getWidth(self):
"""
Returns the width of the handled images in pixels
"""
return self.width
def getHeight(self):
"""
Returns the height of the handled images in pixels
"""
return self.height
def getZDim(self):
"""
Returns the dimension of the latent space of the VAE
"""
return self.z_dim
def flatten(self, x):
"""
Can be used to flatten the output image. This method will only handle
images of the original size specified for the network
"""
return x.view(-1, self.channels * self.height * self.width)
def unflatten(self, x):
"""
Can be used to unflatten an image handled by the network. This method
will only handle images of the original size specified for the network
"""
return x.view(-1, self.channels, self.height, self.width)
class FCVAE(BaseVAE):
"""
Fully connected Variational Autoencoder
"""
def __init__(self, channels=1, width=28, height=28, z_dim=2):
super(FCVAE, self).__init__(channels, width, height, z_dim)
self.fc1 = nn.Linear(self.channels * self.width * self.height, 400)
self.fc21 = nn.Linear(400, self.z_dim)
self.fc22 = nn.Linear(400, self.z_dim)
self.fc3 = nn.Linear(self.z_dim, 400)
self.fc4 = nn.Linear(400, self.channels * self.width * self.height)
def encode(self, x):
h1 = F.relu(self.fc1(x))
return self.fc21(h1), self.fc22(h1)
def reparameterize(self, mu, logvar):
std = torch.exp(0.5 * logvar)
eps = torch.randn_like(std)
return mu + eps * std
def decode(self, z):
h3 = F.relu(self.fc3(z))
return torch.sigmoid(self.fc4(h3))
def forward(self, x):
mu, logvar = self.encode(self.flatten(x))
z = self.reparameterize(mu, logvar)
return self.unflatten(self.decode(z)), mu, logvar
def get_inputs():
return [torch.rand([4, 784])]
def get_init_inputs():
return [[], {}]
| import torch
from torch import device
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
from torch.nn import functional as F
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 1600
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 400
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_add_exp_mul_1(in_ptr0, in_ptr1, in_ptr2, out_ptr0,
xnumel, XBLOCK: tl.constexpr):
xnumel = 8
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask)
tmp2 = tl.load(in_ptr2 + x0, xmask)
tmp3 = 0.5
tmp4 = tmp2 * tmp3
tmp5 = tl_math.exp(tmp4)
tmp6 = tmp1 * tmp5
tmp7 = tmp0 + tmp6
tl.store(out_ptr0 + x0, tmp7, xmask)
@triton.jit
def triton_poi_fused_sigmoid_sigmoid_backward_2(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 3136
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 784
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.sigmoid(tmp2)
tmp4 = 1.0
tmp5 = tmp4 - tmp3
tmp6 = tmp3 * tmp5
tl.store(in_out_ptr0 + x2, tmp3, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10, primals_11) = args
args.clear()
assert_size_stride(primals_1, (4, 784), (784, 1))
assert_size_stride(primals_2, (400, 784), (784, 1))
assert_size_stride(primals_3, (400,), (1,))
assert_size_stride(primals_4, (2, 400), (400, 1))
assert_size_stride(primals_5, (2,), (1,))
assert_size_stride(primals_6, (2, 400), (400, 1))
assert_size_stride(primals_7, (2,), (1,))
assert_size_stride(primals_8, (400, 2), (2, 1))
assert_size_stride(primals_9, (400,), (1,))
assert_size_stride(primals_10, (784, 400), (400, 1))
assert_size_stride(primals_11, (784,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
extern_kernels.mm(primals_1, reinterpret_tensor(primals_2, (784,
400), (1, 784), 0), out=buf0)
del primals_2
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_relu_0[grid(1600)](buf1, primals_3, 1600, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_3
buf2 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
extern_kernels.addmm(primals_5, buf1, reinterpret_tensor(primals_4,
(400, 2), (1, 400), 0), alpha=1, beta=1, out=buf2)
del primals_5
buf3 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
extern_kernels.addmm(primals_7, buf1, reinterpret_tensor(primals_6,
(400, 2), (1, 400), 0), alpha=1, beta=1, out=buf3)
del primals_7
buf4 = torch.ops.aten.randn.default([4, 2], dtype=torch.float32,
device=device(type='cuda', index=0), pin_memory=False)
buf5 = buf4
del buf4
buf6 = empty_strided_cuda((4, 2), (2, 1), torch.float32)
triton_poi_fused_add_exp_mul_1[grid(8)](buf2, buf5, buf3, buf6, 8,
XBLOCK=8, num_warps=1, num_stages=1)
buf7 = empty_strided_cuda((4, 400), (400, 1), torch.float32)
extern_kernels.mm(buf6, reinterpret_tensor(primals_8, (2, 400), (1,
2), 0), out=buf7)
buf8 = buf7
del buf7
triton_poi_fused_relu_0[grid(1600)](buf8, primals_9, 1600, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_9
buf9 = empty_strided_cuda((4, 784), (784, 1), torch.float32)
extern_kernels.mm(buf8, reinterpret_tensor(primals_10, (400, 784),
(1, 400), 0), out=buf9)
buf10 = buf9
del buf9
buf11 = empty_strided_cuda((4, 784), (784, 1), torch.float32)
triton_poi_fused_sigmoid_sigmoid_backward_2[grid(3136)](buf10,
primals_11, buf11, 3136, XBLOCK=256, num_warps=4, num_stages=1)
del primals_11
return (reinterpret_tensor(buf10, (4, 1, 28, 28), (784, 784, 28, 1), 0),
buf2, buf3, primals_1, buf1, buf3, buf5, buf6, buf8, buf11,
primals_10, primals_8, primals_6, primals_4)
class BaseVAE(nn.Module):
"""
Base abstract class for the Variational Autoencoders
"""
def __init__(self, channels=1, width=28, height=28, z_dim=2):
"""
Constructor
Parameters:
channels - The number of channels for the image
width - The width of the image in pixels
height - The height of the image in pixels
z_dim - The dimension of the latent space
"""
super(BaseVAE, self).__init__()
self.channels = channels
self.width = width
self.height = height
self.z_dim = z_dim
def getNbChannels(self):
"""
Returns the number of channels of the handled images
"""
return self.channels
def getWidth(self):
"""
Returns the width of the handled images in pixels
"""
return self.width
def getHeight(self):
"""
Returns the height of the handled images in pixels
"""
return self.height
def getZDim(self):
"""
Returns the dimension of the latent space of the VAE
"""
return self.z_dim
def flatten(self, x):
"""
Can be used to flatten the output image. This method will only handle
images of the original size specified for the network
"""
return x.view(-1, self.channels * self.height * self.width)
def unflatten(self, x):
"""
Can be used to unflatten an image handled by the network. This method
will only handle images of the original size specified for the network
"""
return x.view(-1, self.channels, self.height, self.width)
class FCVAENew(BaseVAE):
"""
Fully connected Variational Autoencoder
"""
def __init__(self, channels=1, width=28, height=28, z_dim=2):
super(FCVAENew, self).__init__(channels, width, height, z_dim)
self.fc1 = nn.Linear(self.channels * self.width * self.height, 400)
self.fc21 = nn.Linear(400, self.z_dim)
self.fc22 = nn.Linear(400, self.z_dim)
self.fc3 = nn.Linear(self.z_dim, 400)
self.fc4 = nn.Linear(400, self.channels * self.width * self.height)
def encode(self, x):
h1 = F.relu(self.fc1(x))
return self.fc21(h1), self.fc22(h1)
def reparameterize(self, mu, logvar):
std = torch.exp(0.5 * logvar)
eps = torch.randn_like(std)
return mu + eps * std
def decode(self, z):
h3 = F.relu(self.fc3(z))
return torch.sigmoid(self.fc4(h3))
def forward(self, input_0):
primals_2 = self.fc1.weight
primals_3 = self.fc1.bias
primals_4 = self.fc21.weight
primals_5 = self.fc21.bias
primals_6 = self.fc22.weight
primals_7 = self.fc22.bias
primals_8 = self.fc3.weight
primals_9 = self.fc3.bias
primals_10 = self.fc4.weight
primals_11 = self.fc4.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9,
primals_10, primals_11])
return output[0], output[1], output[2]
| mbusy/vae | FCVAE | false | 7,201 | [
"MIT"
] | 1 | 455e382a557b72fc944460331e5dd010ff83a76a | https://github.com/mbusy/vae/tree/455e382a557b72fc944460331e5dd010ff83a76a | import torch
from torch.nn import functional as F
from torch import nn
class BaseVAE(nn.Module):
"""
Base abstract class for the Variational Autoencoders
"""
def __init__(self, channels=1, width=28, height=28, z_dim=2):
"""
Constructor
Parameters:
channels - The number of channels for the image
width - The width of the image in pixels
height - The height of the image in pixels
z_dim - The dimension of the latent space
"""
super().__init__()
self.channels = channels
self.width = width
self.height = height
self.z_dim = z_dim
def getNbChannels(self):
"""
Returns the number of channels of the handled images
"""
return self.channels
def getWidth(self):
"""
Returns the width of the handled images in pixels
"""
return self.width
def getHeight(self):
"""
Returns the height of the handled images in pixels
"""
return self.height
def getZDim(self):
"""
Returns the dimension of the latent space of the VAE
"""
return self.z_dim
def flatten(self, x):
"""
Can be used to flatten the output image. This method will only handle
images of the original size specified for the network
"""
return x.view(-1, self.channels * self.height * self.width)
def unflatten(self, x):
"""
Can be used to unflatten an image handled by the network. This method
will only handle images of the original size specified for the network
"""
return x.view(-1, self.channels, self.height, self.width)
class Model(BaseVAE):
"""
Fully connected Variational Autoencoder
"""
def __init__(self, channels=1, width=28, height=28, z_dim=2):
super().__init__(channels, width, height, z_dim)
self.fc1 = nn.Linear(self.channels * self.width * self.height, 400)
self.fc21 = nn.Linear(400, self.z_dim)
self.fc22 = nn.Linear(400, self.z_dim)
self.fc3 = nn.Linear(self.z_dim, 400)
self.fc4 = nn.Linear(400, self.channels * self.width * self.height)
def encode(self, x):
h1 = F.relu(self.fc1(x))
return self.fc21(h1), self.fc22(h1)
def reparameterize(self, mu, logvar):
std = torch.exp(0.5 * logvar)
eps = torch.randn_like(std)
return mu + eps * std
def decode(self, z):
h3 = F.relu(self.fc3(z))
return torch.sigmoid(self.fc4(h3))
def forward(self, x):
mu, logvar = self.encode(self.flatten(x))
z = self.reparameterize(mu, logvar)
return self.unflatten(self.decode(z)), mu, logvar
def get_inputs():
return [torch.rand([4, 784])]
def get_init_inputs():
return []
|
PatchEmbeddings | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/sr/csrhhqsexdcor6gq6tz4dawxblhadgekinzxxkt33uwojltligp6.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%primals_3, %primals_1, %primals_2, [4, 4], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(4, 4), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 1, 1), (4, 1, 1, 1))
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 16, 16), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(buf1, primals_2, 16, grid=grid(16), stream=stream0)
del primals_2
return (reinterpret_tensor(buf1, (1, 4, 4), (1, 4, 1), 0), primals_1, primals_3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class PatchEmbeddings(Module):
"""
<a id="PatchEmbeddings"></a>
## Get patch embeddings
The paper splits the image into patches of equal size and do a linear transformation
on the flattened pixels for each patch.
We implement the same thing through a convolution layer, because it's simpler to implement.
"""
def __init__(self, d_model: 'int', patch_size: 'int', in_channels: 'int'):
"""
* `d_model` is the transformer embeddings size
* `patch_size` is the size of the patch
* `in_channels` is the number of channels in the input image (3 for rgb)
"""
super().__init__()
self.conv = nn.Conv2d(in_channels, d_model, patch_size, stride=
patch_size)
def forward(self, x: 'torch.Tensor'):
"""
* `x` is the input image of shape `[batch_size, channels, height, width]`
"""
x = self.conv(x)
bs, c, h, w = x.shape
x = x.permute(2, 3, 0, 1)
x = x.view(h * w, bs, c)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'d_model': 4, 'patch_size': 4, 'in_channels': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = extern_kernels.convolution(primals_3, primals_1, stride=(4,
4), padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf0, (4, 4, 1, 1), (4, 1, 1, 1))
buf1 = reinterpret_tensor(buf0, (4, 4, 1, 1), (4, 1, 16, 16), 0)
del buf0
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(16)](buf1, primals_2, 16,
XBLOCK=16, num_warps=1, num_stages=1)
del primals_2
return reinterpret_tensor(buf1, (1, 4, 4), (1, 4, 1), 0
), primals_1, primals_3
class PatchEmbeddingsNew(Module):
"""
<a id="PatchEmbeddings"></a>
## Get patch embeddings
The paper splits the image into patches of equal size and do a linear transformation
on the flattened pixels for each patch.
We implement the same thing through a convolution layer, because it's simpler to implement.
"""
def __init__(self, d_model: 'int', patch_size: 'int', in_channels: 'int'):
"""
* `d_model` is the transformer embeddings size
* `patch_size` is the size of the patch
* `in_channels` is the number of channels in the input image (3 for rgb)
"""
super().__init__()
self.conv = nn.Conv2d(in_channels, d_model, patch_size, stride=
patch_size)
def forward(self, input_0):
primals_1 = self.conv.weight
primals_2 = self.conv.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | PatchEmbeddings | false | 7,202 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
<a id="PatchEmbeddings"></a>
## Get patch embeddings
The paper splits the image into patches of equal size and do a linear transformation
on the flattened pixels for each patch.
We implement the same thing through a convolution layer, because it's simpler to implement.
"""
def __init__(self, d_model: 'int', patch_size: 'int', in_channels: 'int'):
"""
* `d_model` is the transformer embeddings size
* `patch_size` is the size of the patch
* `in_channels` is the number of channels in the input image (3 for rgb)
"""
super().__init__()
self.conv = nn.Conv2d(in_channels, d_model, patch_size, stride=
patch_size)
def forward(self, x: 'torch.Tensor'):
"""
* `x` is the input image of shape `[batch_size, channels, height, width]`
"""
x = self.conv(x)
bs, c, h, w = x.shape
x = x.permute(2, 3, 0, 1)
x = x.view(h * w, bs, c)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4, 4]
|
LearnedPositionalEmbeddings | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/26/c26shuqxjhfilaoens27tm25ij3y3e5ctdjjfr7fuwafpcfebmdi.py
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
# Source node to ATen node mapping:
# add => add
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_2, %select), kwargs = {})
triton_poi_fused_add_0 = async_compile.triton('triton_poi_fused_add_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (16 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x2), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (5000, 1, 4), (4, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_0.run(primals_2, primals_1, buf0, 256, grid=grid(256), stream=stream0)
del primals_1
del primals_2
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((5000, 1, 4), (4, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class LearnedPositionalEmbeddings(Module):
"""
<a id="LearnedPositionalEmbeddings"></a>
## Add parameterized positional encodings
This adds learned positional embeddings to patch embeddings.
"""
def __init__(self, d_model: 'int', max_len: 'int'=5000):
"""
* `d_model` is the transformer embeddings size
* `max_len` is the maximum number of patches
"""
super().__init__()
self.positional_encodings = nn.Parameter(torch.zeros(max_len, 1,
d_model), requires_grad=True)
def forward(self, x: 'torch.Tensor'):
"""
* `x` is the patch embeddings of shape `[patches, batch_size, d_model]`
"""
pe = self.positional_encodings[x.shape[0]]
return x + pe
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'d_model': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + (16 + x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x2, tmp2, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (5000, 1, 4), (4, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_0[grid(256)](primals_2, primals_1, buf0, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_1
del primals_2
return buf0,
class LearnedPositionalEmbeddingsNew(Module):
"""
<a id="LearnedPositionalEmbeddings"></a>
## Add parameterized positional encodings
This adds learned positional embeddings to patch embeddings.
"""
def __init__(self, d_model: 'int', max_len: 'int'=5000):
"""
* `d_model` is the transformer embeddings size
* `max_len` is the maximum number of patches
"""
super().__init__()
self.positional_encodings = nn.Parameter(torch.zeros(max_len, 1,
d_model), requires_grad=True)
def forward(self, input_0):
primals_1 = self.positional_encodings
primals_2 = input_0
output = call([primals_1, primals_2])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | LearnedPositionalEmbeddings | false | 7,203 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
<a id="LearnedPositionalEmbeddings"></a>
## Add parameterized positional encodings
This adds learned positional embeddings to patch embeddings.
"""
def __init__(self, d_model: 'int', max_len: 'int'=5000):
"""
* `d_model` is the transformer embeddings size
* `max_len` is the maximum number of patches
"""
super().__init__()
self.positional_encodings = nn.Parameter(torch.zeros(max_len, 1,
d_model), requires_grad=True)
def forward(self, x: 'torch.Tensor'):
"""
* `x` is the patch embeddings of shape `[patches, batch_size, d_model]`
"""
pe = self.positional_encodings[x.shape[0]]
return x + pe
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
LSTMCell | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/lg/clgtnw2bgvc3mnanbov77yxt2a6ztfqy7qpuhudozzijvzjgbgly.py
# Topologically Sorted Source Nodes: [sigmoid, mul, sigmoid_1, tanh, mul_1, c_next, sigmoid_2, tanh_1, h_next], Original ATen: [aten.sigmoid, aten.mul, aten.tanh, aten.add, aten.sigmoid_backward]
# Source node to ATen node mapping:
# c_next => add_1
# h_next => mul_2
# mul => mul
# mul_1 => mul_1
# sigmoid => sigmoid
# sigmoid_1 => sigmoid_1
# sigmoid_2 => sigmoid_2
# tanh => tanh
# tanh_1 => tanh_1
# Graph fragment:
# %sigmoid : [num_users=3] = call_function[target=torch.ops.aten.sigmoid.default](args = (%getitem_1,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid, %primals_6), kwargs = {})
# %sigmoid_1 : [num_users=2] = call_function[target=torch.ops.aten.sigmoid.default](args = (%getitem,), kwargs = {})
# %tanh : [num_users=2] = call_function[target=torch.ops.aten.tanh.default](args = (%getitem_2,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid_1, %tanh), kwargs = {})
# %add_1 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
# %sigmoid_2 : [num_users=2] = call_function[target=torch.ops.aten.sigmoid.default](args = (%getitem_3,), kwargs = {})
# %tanh_1 : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%add_1,), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid_2, %tanh_1), kwargs = {})
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %sigmoid), kwargs = {})
# %mul_16 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid, %sub_4), kwargs = {})
triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_0 = async_compile.triton('triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: '*fp32', 8: '*fp32', 9: '*fp32', 10: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 13, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_0(in_ptr0, in_ptr1, in_ptr2, in_ptr3, out_ptr0, out_ptr1, out_ptr2, out_ptr3, out_ptr4, out_ptr5, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (16*x1)), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x0 + (16*x1)), xmask)
tmp6 = tl.load(in_ptr0 + (8 + x0 + (16*x1)), xmask)
tmp7 = tl.load(in_ptr1 + (8 + x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr2 + (8 + x0 + (16*x1)), xmask)
tmp12 = tl.load(in_ptr0 + (4 + x0 + (16*x1)), xmask)
tmp13 = tl.load(in_ptr1 + (4 + x0), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr2 + (4 + x0 + (16*x1)), xmask)
tmp18 = tl.load(in_ptr3 + (x2), xmask)
tmp25 = tl.load(in_ptr0 + (12 + x0 + (16*x1)), xmask)
tmp26 = tl.load(in_ptr1 + (12 + x0), xmask, eviction_policy='evict_last')
tmp28 = tl.load(in_ptr2 + (12 + x0 + (16*x1)), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp5 = tl.sigmoid(tmp4)
tmp8 = tmp6 + tmp7
tmp10 = tmp8 + tmp9
tmp11 = libdevice.tanh(tmp10)
tmp14 = tmp12 + tmp13
tmp16 = tmp14 + tmp15
tmp17 = tl.sigmoid(tmp16)
tmp19 = tmp17 * tmp18
tmp20 = tmp5 * tmp11
tmp21 = tmp19 + tmp20
tmp22 = 1.0
tmp23 = tmp22 - tmp17
tmp24 = tmp17 * tmp23
tmp27 = tmp25 + tmp26
tmp29 = tmp27 + tmp28
tmp30 = tl.sigmoid(tmp29)
tmp31 = libdevice.tanh(tmp21)
tmp32 = tmp30 * tmp31
tl.store(out_ptr0 + (x2), tmp5, xmask)
tl.store(out_ptr1 + (x2), tmp11, xmask)
tl.store(out_ptr2 + (x2), tmp21, xmask)
tl.store(out_ptr3 + (x2), tmp24, xmask)
tl.store(out_ptr4 + (x2), tmp30, xmask)
tl.store(out_ptr5 + (x2), tmp32, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (16, 4), (4, 1))
assert_size_stride(primals_2, (16, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (16, 4), (4, 1))
assert_size_stride(primals_5, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_6, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 16), (16, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 16), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((64, 16), (16, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_5, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 16), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [sigmoid, mul, sigmoid_1, tanh, mul_1, c_next, sigmoid_2, tanh_1, h_next], Original ATen: [aten.sigmoid, aten.mul, aten.tanh, aten.add, aten.sigmoid_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_0.run(buf0, primals_2, buf1, primals_6, buf2, buf3, buf4, buf7, buf5, buf6, 256, grid=grid(256), stream=stream0)
del buf0
del buf1
del primals_2
return (buf6, buf4, primals_6, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_5, (64, 4), (4, 1), 0), buf2, buf3, buf4, buf5, buf7, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((16, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((16, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((16, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class LSTMCell(Module):
"""
## Long Short-Term Memory Cell
LSTM Cell computes $c$, and $h$. $c$ is like the long-term memory,
and $h$ is like the short term memory.
We use the input $x$ and $h$ to update the long term memory.
In the update, some features of $c$ are cleared with a forget gate $f$,
and some features $i$ are added through a gate $g$.
The new short term memory is the $ anh$ of the long-term memory
multiplied by the output gate $o$.
Note that the cell doesn't look at long term memory $c$ when doing the update. It only modifies it.
Also $c$ never goes through a linear transformation.
This is what solves vanishing and exploding gradients.
Here's the update rule.
egin{align}
c_t &= \\sigma(f_t) \\odot c_{t-1} + \\sigma(i_t) \\odot anh(g_t) \\
h_t &= \\sigma(o_t) \\odot anh(c_t)
\\end{align}
$\\odot$ stands for element-wise multiplication.
Intermediate values and gates are computed as linear transformations of the hidden
state and input.
egin{align}
i_t &= lin_x^i(x_t) + lin_h^i(h_{t-1}) \\
f_t &= lin_x^f(x_t) + lin_h^f(h_{t-1}) \\
g_t &= lin_x^g(x_t) + lin_h^g(h_{t-1}) \\
o_t &= lin_x^o(x_t) + lin_h^o(h_{t-1})
\\end{align}
"""
def __init__(self, input_size: 'int', hidden_size: 'int', layer_norm:
'bool'=False):
super().__init__()
self.hidden_lin = nn.Linear(hidden_size, 4 * hidden_size)
self.input_lin = nn.Linear(input_size, 4 * hidden_size, bias=False)
if layer_norm:
self.layer_norm = nn.ModuleList([nn.LayerNorm(hidden_size) for
_ in range(4)])
self.layer_norm_c = nn.LayerNorm(hidden_size)
else:
self.layer_norm = nn.ModuleList([nn.Identity() for _ in range(4)])
self.layer_norm_c = nn.Identity()
def forward(self, x: 'torch.Tensor', h: 'torch.Tensor', c: 'torch.Tensor'):
ifgo = self.hidden_lin(h) + self.input_lin(x)
ifgo = ifgo.chunk(4, dim=-1)
ifgo = [self.layer_norm[i](ifgo[i]) for i in range(4)]
i, f, g, o = ifgo
c_next = torch.sigmoid(f) * c + torch.sigmoid(i) * torch.tanh(g)
h_next = torch.sigmoid(o) * torch.tanh(self.layer_norm_c(c_next))
return h_next, c_next
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [[], {'input_size': 4, 'hidden_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_0(in_ptr0,
in_ptr1, in_ptr2, in_ptr3, out_ptr0, out_ptr1, out_ptr2, out_ptr3,
out_ptr4, out_ptr5, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 16 * x1), xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x0 + 16 * x1), xmask)
tmp6 = tl.load(in_ptr0 + (8 + x0 + 16 * x1), xmask)
tmp7 = tl.load(in_ptr1 + (8 + x0), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr2 + (8 + x0 + 16 * x1), xmask)
tmp12 = tl.load(in_ptr0 + (4 + x0 + 16 * x1), xmask)
tmp13 = tl.load(in_ptr1 + (4 + x0), xmask, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr2 + (4 + x0 + 16 * x1), xmask)
tmp18 = tl.load(in_ptr3 + x2, xmask)
tmp25 = tl.load(in_ptr0 + (12 + x0 + 16 * x1), xmask)
tmp26 = tl.load(in_ptr1 + (12 + x0), xmask, eviction_policy='evict_last')
tmp28 = tl.load(in_ptr2 + (12 + x0 + 16 * x1), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp5 = tl.sigmoid(tmp4)
tmp8 = tmp6 + tmp7
tmp10 = tmp8 + tmp9
tmp11 = libdevice.tanh(tmp10)
tmp14 = tmp12 + tmp13
tmp16 = tmp14 + tmp15
tmp17 = tl.sigmoid(tmp16)
tmp19 = tmp17 * tmp18
tmp20 = tmp5 * tmp11
tmp21 = tmp19 + tmp20
tmp22 = 1.0
tmp23 = tmp22 - tmp17
tmp24 = tmp17 * tmp23
tmp27 = tmp25 + tmp26
tmp29 = tmp27 + tmp28
tmp30 = tl.sigmoid(tmp29)
tmp31 = libdevice.tanh(tmp21)
tmp32 = tmp30 * tmp31
tl.store(out_ptr0 + x2, tmp5, xmask)
tl.store(out_ptr1 + x2, tmp11, xmask)
tl.store(out_ptr2 + x2, tmp21, xmask)
tl.store(out_ptr3 + x2, tmp24, xmask)
tl.store(out_ptr4 + x2, tmp30, xmask)
tl.store(out_ptr5 + x2, tmp32, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6 = args
args.clear()
assert_size_stride(primals_1, (16, 4), (4, 1))
assert_size_stride(primals_2, (16,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (16, 4), (4, 1))
assert_size_stride(primals_5, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_6, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 16), (16, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 16), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((64, 16), (16, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_5, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 16), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf6 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_mul_sigmoid_sigmoid_backward_tanh_0[grid(256)](
buf0, primals_2, buf1, primals_6, buf2, buf3, buf4, buf7, buf5,
buf6, 256, XBLOCK=128, num_warps=4, num_stages=1)
del buf0
del buf1
del primals_2
return buf6, buf4, primals_6, reinterpret_tensor(primals_3, (64, 4), (4,
1), 0), reinterpret_tensor(primals_5, (64, 4), (4, 1), 0
), buf2, buf3, buf4, buf5, buf7
class LSTMCellNew(Module):
"""
## Long Short-Term Memory Cell
LSTM Cell computes $c$, and $h$. $c$ is like the long-term memory,
and $h$ is like the short term memory.
We use the input $x$ and $h$ to update the long term memory.
In the update, some features of $c$ are cleared with a forget gate $f$,
and some features $i$ are added through a gate $g$.
The new short term memory is the $ anh$ of the long-term memory
multiplied by the output gate $o$.
Note that the cell doesn't look at long term memory $c$ when doing the update. It only modifies it.
Also $c$ never goes through a linear transformation.
This is what solves vanishing and exploding gradients.
Here's the update rule.
egin{align}
c_t &= \\sigma(f_t) \\odot c_{t-1} + \\sigma(i_t) \\odot anh(g_t) \\
h_t &= \\sigma(o_t) \\odot anh(c_t)
\\end{align}
$\\odot$ stands for element-wise multiplication.
Intermediate values and gates are computed as linear transformations of the hidden
state and input.
egin{align}
i_t &= lin_x^i(x_t) + lin_h^i(h_{t-1}) \\
f_t &= lin_x^f(x_t) + lin_h^f(h_{t-1}) \\
g_t &= lin_x^g(x_t) + lin_h^g(h_{t-1}) \\
o_t &= lin_x^o(x_t) + lin_h^o(h_{t-1})
\\end{align}
"""
def __init__(self, input_size: 'int', hidden_size: 'int', layer_norm:
'bool'=False):
super().__init__()
self.hidden_lin = nn.Linear(hidden_size, 4 * hidden_size)
self.input_lin = nn.Linear(input_size, 4 * hidden_size, bias=False)
if layer_norm:
self.layer_norm = nn.ModuleList([nn.LayerNorm(hidden_size) for
_ in range(4)])
self.layer_norm_c = nn.LayerNorm(hidden_size)
else:
self.layer_norm = nn.ModuleList([nn.Identity() for _ in range(4)])
self.layer_norm_c = nn.Identity()
def forward(self, input_0, input_1, input_2):
primals_1 = self.hidden_lin.weight
primals_2 = self.hidden_lin.bias
primals_4 = self.input_lin.weight
primals_3 = input_0
primals_5 = input_1
primals_6 = input_2
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6])
return output[0], output[1]
| mcx/annotated_deep_learning_paper_implementations | LSTMCell | false | 7,204 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Long Short-Term Memory Cell
LSTM Cell computes $c$, and $h$. $c$ is like the long-term memory,
and $h$ is like the short term memory.
We use the input $x$ and $h$ to update the long term memory.
In the update, some features of $c$ are cleared with a forget gate $f$,
and some features $i$ are added through a gate $g$.
The new short term memory is the $ anh$ of the long-term memory
multiplied by the output gate $o$.
Note that the cell doesn't look at long term memory $c$ when doing the update. It only modifies it.
Also $c$ never goes through a linear transformation.
This is what solves vanishing and exploding gradients.
Here's the update rule.
egin{align}
c_t &= \\sigma(f_t) \\odot c_{t-1} + \\sigma(i_t) \\odot anh(g_t) \\
h_t &= \\sigma(o_t) \\odot anh(c_t)
\\end{align}
$\\odot$ stands for element-wise multiplication.
Intermediate values and gates are computed as linear transformations of the hidden
state and input.
egin{align}
i_t &= lin_x^i(x_t) + lin_h^i(h_{t-1}) \\
f_t &= lin_x^f(x_t) + lin_h^f(h_{t-1}) \\
g_t &= lin_x^g(x_t) + lin_h^g(h_{t-1}) \\
o_t &= lin_x^o(x_t) + lin_h^o(h_{t-1})
\\end{align}
"""
def __init__(self, input_size: 'int', hidden_size: 'int', layer_norm:
'bool'=False):
super().__init__()
self.hidden_lin = nn.Linear(hidden_size, 4 * hidden_size)
self.input_lin = nn.Linear(input_size, 4 * hidden_size, bias=False)
if layer_norm:
self.layer_norm = nn.ModuleList([nn.LayerNorm(hidden_size) for
_ in range(4)])
self.layer_norm_c = nn.LayerNorm(hidden_size)
else:
self.layer_norm = nn.ModuleList([nn.Identity() for _ in range(4)])
self.layer_norm_c = nn.Identity()
def forward(self, x: 'torch.Tensor', h: 'torch.Tensor', c: 'torch.Tensor'):
ifgo = self.hidden_lin(h) + self.input_lin(x)
ifgo = ifgo.chunk(4, dim=-1)
ifgo = [self.layer_norm[i](ifgo[i]) for i in range(4)]
i, f, g, o = ifgo
c_next = torch.sigmoid(f) * c + torch.sigmoid(i) * torch.tanh(g)
h_next = torch.sigmoid(o) * torch.tanh(self.layer_norm_c(c_next))
return h_next, c_next
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4]), torch.rand(
[4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
SquaredReLU | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ic/cicdjjsf5hzl3lpbazppxvx5umzuokzpuu5z3lapupqhtd2tusv6.py
# Topologically Sorted Source Nodes: [x, mul], Original ATen: [aten.relu, aten.mul]
# Source node to ATen node mapping:
# mul => mul
# x => relu
# Graph fragment:
# %relu : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%arg0_1,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%relu, %relu), kwargs = {})
triton_poi_fused_mul_relu_0 = async_compile.triton('triton_poi_fused_mul_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_relu_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_relu_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl.full([1], 0, tl.int32)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp3 = tmp2 * tmp2
tl.store(out_ptr0 + (x0), tmp3, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x, mul], Original ATen: [aten.relu, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_relu_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class SquaredReLU(Module):
"""
## Squared ReLU activation
$$y = {\\max(x, 0)}^2$$
Squared ReLU is used as the activation function in the
[position wise feedforward module](../feed_forward.html).
"""
def __init__(self):
super().__init__()
self.relu = nn.ReLU()
def forward(self, x: 'torch.Tensor'):
x = self.relu(x)
return x * x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_mul_relu_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl.full([1], 0, tl.int32)
tmp2 = triton_helpers.maximum(tmp1, tmp0)
tmp3 = tmp2 * tmp2
tl.store(out_ptr0 + x0, tmp3, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_relu_0[grid(256)](arg0_1, buf0, 256, XBLOCK=
256, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class SquaredReLUNew(Module):
"""
## Squared ReLU activation
$$y = {\\max(x, 0)}^2$$
Squared ReLU is used as the activation function in the
[position wise feedforward module](../feed_forward.html).
"""
def __init__(self):
super().__init__()
self.relu = nn.ReLU()
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | SquaredReLU | false | 7,205 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Squared ReLU activation
$$y = {\\max(x, 0)}^2$$
Squared ReLU is used as the activation function in the
[position wise feedforward module](../feed_forward.html).
"""
def __init__(self):
super().__init__()
self.relu = nn.ReLU()
def forward(self, x: 'torch.Tensor'):
x = self.relu(x)
return x * x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
MarginLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/2l/c2lqz5eg4uounnzhu2xhx6kieol5nd5d676uuijy3vfi2vfxmwhd.py
# Topologically Sorted Source Nodes: [eye, labels, pow_1, sum_1, v_norm, sub, relu, mul, sub_1, mul_1, sub_2, relu_1, mul_2, loss], Original ATen: [aten.eye, aten.index, aten.pow, aten.sum, aten.sqrt, aten.rsub, aten.relu, aten.mul, aten.sub, aten.add]
# Source node to ATen node mapping:
# eye => eq, full_default, full_default_1, iota_1, where
# labels => index
# loss => add
# mul => mul
# mul_1 => mul_1
# mul_2 => mul_2
# pow_1 => pow_1
# relu => relu
# relu_1 => relu_1
# sub => sub
# sub_1 => sub_1
# sub_2 => sub_2
# sum_1 => sum_1
# v_norm => sqrt
# Graph fragment:
# %iota_1 : [num_users=1] = call_function[target=torch.ops.prims.iota.default](args = (4,), kwargs = {start: 0, step: 1, dtype: torch.int64, device: cuda:0, requires_grad: False})
# %eq : [num_users=1] = call_function[target=torch.ops.aten.eq.Tensor](args = (%unsqueeze, %iota_1), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([1], 1), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %full_default_1 : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%eq, %full_default, %full_default_1), kwargs = {})
# %index : [num_users=2] = call_function[target=torch.ops.aten.index.Tensor](args = (%where, [%arg1_1]), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%arg0_1, 2), kwargs = {})
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [-1]), kwargs = {})
# %sqrt : [num_users=2] = call_function[target=torch.ops.aten.sqrt.default](args = (%sum_1,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (0.9, %sqrt), kwargs = {})
# %relu : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%sub,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%index, %relu), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1.0, %index), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, 0.5), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sqrt, 0.1), kwargs = {})
# %relu_1 : [num_users=1] = call_function[target=torch.ops.aten.relu.default](args = (%sub_2,), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_1, %relu_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_2), kwargs = {})
triton_poi_fused_add_eye_index_mul_pow_relu_rsub_sqrt_sub_sum_0 = async_compile.triton('triton_poi_fused_add_eye_index_mul_pow_relu_rsub_sqrt_sub_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*i64', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_eye_index_mul_pow_relu_rsub_sqrt_sub_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_eye_index_mul_pow_relu_rsub_sqrt_sub_sum_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4) % 4
x0 = xindex % 4
x3 = xindex
tmp0 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (4*x3), xmask, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr1 + (1 + (4*x3)), xmask, eviction_policy='evict_last')
tmp17 = tl.load(in_ptr1 + (2 + (4*x3)), xmask, eviction_policy='evict_last')
tmp20 = tl.load(in_ptr1 + (3 + (4*x3)), xmask, eviction_policy='evict_last')
tmp1 = tl.full([XBLOCK], 4, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tl.device_assert(((0 <= tmp4) & (tmp4 < 4)) | ~(xmask), "index out of bounds: 0 <= tmp4 < 4")
tmp6 = tmp4
tmp7 = x0
tmp8 = tmp6 == tmp7
tmp9 = 1.0
tmp10 = 0.0
tmp11 = tl.where(tmp8, tmp9, tmp10)
tmp13 = tmp12 * tmp12
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp21 = tmp20 * tmp20
tmp22 = tmp19 + tmp21
tmp23 = libdevice.sqrt(tmp22)
tmp24 = 0.9
tmp25 = tmp24 - tmp23
tmp26 = tl.full([1], 0, tl.int32)
tmp27 = triton_helpers.maximum(tmp26, tmp25)
tmp28 = tmp11 * tmp27
tmp29 = tmp9 - tmp11
tmp30 = 0.5
tmp31 = tmp29 * tmp30
tmp32 = 0.1
tmp33 = tmp23 - tmp32
tmp34 = triton_helpers.maximum(tmp26, tmp33)
tmp35 = tmp31 * tmp34
tmp36 = tmp28 + tmp35
tl.store(out_ptr0 + (x3), tmp36, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/nl/cnlounivewsan7neintygju4ung6fdl7sne7vd5zvvrfxzgid2yx.py
# Topologically Sorted Source Nodes: [sum_2, mean], Original ATen: [aten.sum, aten.mean]
# Source node to ATen node mapping:
# mean => mean
# sum_2 => sum_2
# Graph fragment:
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%add, [-1]), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sum_2,), kwargs = {})
triton_per_fused_mean_sum_1 = async_compile.triton('triton_per_fused_mean_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 16],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {2: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=(2,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_mean_sum_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_mean_sum_1(in_out_ptr0, in_ptr0, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 16
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (4*r0), None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*r0)), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*r0)), None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*r0)), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = tl.broadcast_to(tmp6, [XBLOCK, RBLOCK])
tmp9 = tl.sum(tmp7, 1)[:, None]
tmp10 = 16.0
tmp11 = tmp9 / tmp10
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp11, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [eye, labels, pow_1, sum_1, v_norm, sub, relu, mul, sub_1, mul_1, sub_2, relu_1, mul_2, loss], Original ATen: [aten.eye, aten.index, aten.pow, aten.sum, aten.sqrt, aten.rsub, aten.relu, aten.mul, aten.sub, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_eye_index_mul_pow_relu_rsub_sqrt_sub_sum_0.run(arg1_1, arg0_1, buf0, 64, grid=grid(64), stream=stream0)
del arg0_1
del arg1_1
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [sum_2, mean], Original ATen: [aten.sum, aten.mean]
triton_per_fused_mean_sum_1.run(buf2, buf0, 1, 16, grid=grid(1), stream=stream0)
del buf0
return (buf2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.int64)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.nn.functional as F
import torch.utils.data
import torch.nn.functional
import torch.autograd
class MarginLoss(Module):
'\n ## Margin loss for class existence\n\n A separate margin loss is used for each output capsule and the total loss is the sum of them.\n The length of each output capsule is the probability that class is present in the input.\n\n Loss for each output capsule or class $k$ is,\n $$\\mathcal{L}_k = T_k \\max(0, m^{+} - \\lVert\\mathbf{v}_k\rVert)^2 +\n \\lambda (1 - T_k) \\max(0, \\lVert\\mathbf{v}_k\rVert - m^{-})^2$$\n\n $T_k$ is $1$ if the class $k$ is present and $0$ otherwise.\n The first component of the loss is $0$ when the class is not present,\n and the second component is $0$ if the class is present.\n The $\\max(0, x)$ is used to avoid predictions going to extremes.\n $m^{+}$ is set to be $0.9$ and $m^{-}$ to be $0.1$ in the paper.\n\n The $\\lambda$ down-weighting is used to stop the length of all capsules from\n falling during the initial phase of training.\n '
def __init__(self, *, n_labels: int, lambda_: float=0.5, m_positive:
float=0.9, m_negative: float=0.1):
super().__init__()
self.m_negative = m_negative
self.m_positive = m_positive
self.lambda_ = lambda_
self.n_labels = n_labels
def forward(self, v: 'torch.Tensor', labels: 'torch.Tensor'):
"""
`v`, $\\mathbf{v}_j$ are the squashed output capsules.
This has shape `[batch_size, n_labels, n_features]`; that is, there is a capsule for each label.
`labels` are the labels, and has shape `[batch_size]`.
"""
v_norm = torch.sqrt((v ** 2).sum(dim=-1))
labels = torch.eye(self.n_labels, device=labels.device)[labels]
loss = labels * F.relu(self.m_positive - v_norm) + self.lambda_ * (
1.0 - labels) * F.relu(v_norm - self.m_negative)
return loss.sum(dim=-1).mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.ones([4], dtype=torch.int64)]
def get_init_inputs():
return [[], {'n_labels': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_eye_index_mul_pow_relu_rsub_sqrt_sub_sum_0(in_ptr0,
in_ptr1, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4 % 4
x0 = xindex % 4
x3 = xindex
tmp0 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + 4 * x3, xmask, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr1 + (1 + 4 * x3), xmask, eviction_policy='evict_last'
)
tmp17 = tl.load(in_ptr1 + (2 + 4 * x3), xmask, eviction_policy='evict_last'
)
tmp20 = tl.load(in_ptr1 + (3 + 4 * x3), xmask, eviction_policy='evict_last'
)
tmp1 = tl.full([XBLOCK], 4, tl.int32)
tmp2 = tmp0 + tmp1
tmp3 = tmp0 < 0
tmp4 = tl.where(tmp3, tmp2, tmp0)
tl.device_assert((0 <= tmp4) & (tmp4 < 4) | ~xmask,
'index out of bounds: 0 <= tmp4 < 4')
tmp6 = tmp4
tmp7 = x0
tmp8 = tmp6 == tmp7
tmp9 = 1.0
tmp10 = 0.0
tmp11 = tl.where(tmp8, tmp9, tmp10)
tmp13 = tmp12 * tmp12
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp21 = tmp20 * tmp20
tmp22 = tmp19 + tmp21
tmp23 = libdevice.sqrt(tmp22)
tmp24 = 0.9
tmp25 = tmp24 - tmp23
tmp26 = tl.full([1], 0, tl.int32)
tmp27 = triton_helpers.maximum(tmp26, tmp25)
tmp28 = tmp11 * tmp27
tmp29 = tmp9 - tmp11
tmp30 = 0.5
tmp31 = tmp29 * tmp30
tmp32 = 0.1
tmp33 = tmp23 - tmp32
tmp34 = triton_helpers.maximum(tmp26, tmp33)
tmp35 = tmp31 * tmp34
tmp36 = tmp28 + tmp35
tl.store(out_ptr0 + x3, tmp36, xmask)
@triton.jit
def triton_per_fused_mean_sum_1(in_out_ptr0, in_ptr0, xnumel, rnumel,
XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 16
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + 4 * r0, None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * r0), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * r0), None, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * r0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = tl.broadcast_to(tmp6, [XBLOCK, RBLOCK])
tmp9 = tl.sum(tmp7, 1)[:, None]
tmp10 = 16.0
tmp11 = tmp9 / tmp10
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp11, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_eye_index_mul_pow_relu_rsub_sqrt_sub_sum_0[grid
(64)](arg1_1, arg0_1, buf0, 64, XBLOCK=64, num_warps=1,
num_stages=1)
del arg0_1
del arg1_1
buf1 = empty_strided_cuda((), (), torch.float32)
buf2 = buf1
del buf1
triton_per_fused_mean_sum_1[grid(1)](buf2, buf0, 1, 16, XBLOCK=1,
num_warps=2, num_stages=1)
del buf0
return buf2,
class MarginLossNew(Module):
'\n ## Margin loss for class existence\n\n A separate margin loss is used for each output capsule and the total loss is the sum of them.\n The length of each output capsule is the probability that class is present in the input.\n\n Loss for each output capsule or class $k$ is,\n $$\\mathcal{L}_k = T_k \\max(0, m^{+} - \\lVert\\mathbf{v}_k\rVert)^2 +\n \\lambda (1 - T_k) \\max(0, \\lVert\\mathbf{v}_k\rVert - m^{-})^2$$\n\n $T_k$ is $1$ if the class $k$ is present and $0$ otherwise.\n The first component of the loss is $0$ when the class is not present,\n and the second component is $0$ if the class is present.\n The $\\max(0, x)$ is used to avoid predictions going to extremes.\n $m^{+}$ is set to be $0.9$ and $m^{-}$ to be $0.1$ in the paper.\n\n The $\\lambda$ down-weighting is used to stop the length of all capsules from\n falling during the initial phase of training.\n '
def __init__(self, *, n_labels: int, lambda_: float=0.5, m_positive:
float=0.9, m_negative: float=0.1):
super().__init__()
self.m_negative = m_negative
self.m_positive = m_positive
self.lambda_ = lambda_
self.n_labels = n_labels
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | MarginLoss | false | 7,206 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.nn.functional as F
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
'\n ## Margin loss for class existence\n\n A separate margin loss is used for each output capsule and the total loss is the sum of them.\n The length of each output capsule is the probability that class is present in the input.\n\n Loss for each output capsule or class $k$ is,\n $$\\mathcal{L}_k = T_k \\max(0, m^{+} - \\lVert\\mathbf{v}_k\rVert)^2 +\n \\lambda (1 - T_k) \\max(0, \\lVert\\mathbf{v}_k\rVert - m^{-})^2$$\n\n $T_k$ is $1$ if the class $k$ is present and $0$ otherwise.\n The first component of the loss is $0$ when the class is not present,\n and the second component is $0$ if the class is present.\n The $\\max(0, x)$ is used to avoid predictions going to extremes.\n $m^{+}$ is set to be $0.9$ and $m^{-}$ to be $0.1$ in the paper.\n\n The $\\lambda$ down-weighting is used to stop the length of all capsules from\n falling during the initial phase of training.\n '
def __init__(self, *, n_labels: int, lambda_: float=0.5, m_positive:
float=0.9, m_negative: float=0.1):
super().__init__()
self.m_negative = m_negative
self.m_positive = m_positive
self.lambda_ = lambda_
self.n_labels = n_labels
def forward(self, v: 'torch.Tensor', labels: 'torch.Tensor'):
"""
`v`, $\\mathbf{v}_j$ are the squashed output capsules.
This has shape `[batch_size, n_labels, n_features]`; that is, there is a capsule for each label.
`labels` are the labels, and has shape `[batch_size]`.
"""
v_norm = torch.sqrt((v ** 2).sum(dim=-1))
labels = torch.eye(self.n_labels, device=labels.device)[labels]
loss = labels * F.relu(self.m_positive - v_norm) + self.lambda_ * (
1.0 - labels) * F.relu(v_norm - self.m_negative)
return loss.sum(dim=-1).mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.ones([4], dtype=torch.int64)]
def get_init_inputs():
return [4]
|
Squash | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ym/cym3ccudngmvatoe6w7myo62gop3lu5z7s5oobh4dsa5ywnpodvj.py
# Topologically Sorted Source Nodes: [pow_1, s2, add, truediv, add_1, sqrt, truediv_1, mul], Original ATen: [aten.pow, aten.sum, aten.add, aten.div, aten.sqrt, aten.mul]
# Source node to ATen node mapping:
# add => add
# add_1 => add_1
# mul => mul
# pow_1 => pow_1
# s2 => sum_1
# sqrt => sqrt
# truediv => div
# truediv_1 => div_1
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%arg0_1, 2), kwargs = {})
# %sum_1 : [num_users=3] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%pow_1, [-1], True), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sum_1, 1), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%sum_1, %add), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sum_1, 1e-08), kwargs = {})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%add_1,), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg0_1, %sqrt), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div, %div_1), kwargs = {})
triton_poi_fused_add_div_mul_pow_sqrt_sum_0 = async_compile.triton('triton_poi_fused_add_div_mul_pow_sqrt_sum_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mul_pow_sqrt_sum_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mul_pow_sqrt_sum_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tmp0 * tmp0
tmp3 = tmp2 * tmp2
tmp4 = tmp1 + tmp3
tmp6 = tmp5 * tmp5
tmp7 = tmp4 + tmp6
tmp9 = tmp8 * tmp8
tmp10 = tmp7 + tmp9
tmp11 = 1.0
tmp12 = tmp10 + tmp11
tmp13 = tmp10 / tmp12
tmp15 = 1e-08
tmp16 = tmp10 + tmp15
tmp17 = libdevice.sqrt(tmp16)
tmp18 = tmp14 / tmp17
tmp19 = tmp13 * tmp18
tl.store(out_ptr0 + (x2), tmp19, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, s2, add, truediv, add_1, sqrt, truediv_1, mul], Original ATen: [aten.pow, aten.sum, aten.add, aten.div, aten.sqrt, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_mul_pow_sqrt_sum_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Squash(Module):
'\n ## Squash\n\n This is **squashing** function from paper, given by equation $(1)$.\n\n $$\\mathbf{v}_j = \x0crac{{\\lVert \\mathbf{s}_j \rVert}^2}{1 + {\\lVert \\mathbf{s}_j \rVert}^2}\n \x0crac{\\mathbf{s}_j}{\\lVert \\mathbf{s}_j \rVert}$$\n\n $\x0crac{\\mathbf{s}_j}{\\lVert \\mathbf{s}_j \rVert}$\n normalizes the length of all the capsules, whilst\n $\x0crac{{\\lVert \\mathbf{s}_j \rVert}^2}{1 + {\\lVert \\mathbf{s}_j \rVert}^2}$\n shrinks the capsules that have a length smaller than one .\n '
def __init__(self, epsilon=1e-08):
super().__init__()
self.epsilon = epsilon
def forward(self, s: 'torch.Tensor'):
"""
The shape of `s` is `[batch_size, n_capsules, n_features]`
"""
s2 = (s ** 2).sum(dim=-1, keepdims=True)
return s2 / (1 + s2) * (s / torch.sqrt(s2 + self.epsilon))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_mul_pow_sqrt_sum_0(in_ptr0, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tmp0 * tmp0
tmp3 = tmp2 * tmp2
tmp4 = tmp1 + tmp3
tmp6 = tmp5 * tmp5
tmp7 = tmp4 + tmp6
tmp9 = tmp8 * tmp8
tmp10 = tmp7 + tmp9
tmp11 = 1.0
tmp12 = tmp10 + tmp11
tmp13 = tmp10 / tmp12
tmp15 = 1e-08
tmp16 = tmp10 + tmp15
tmp17 = libdevice.sqrt(tmp16)
tmp18 = tmp14 / tmp17
tmp19 = tmp13 * tmp18
tl.store(out_ptr0 + x2, tmp19, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_mul_pow_sqrt_sum_0[grid(256)](arg0_1, buf0,
256, XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class SquashNew(Module):
'\n ## Squash\n\n This is **squashing** function from paper, given by equation $(1)$.\n\n $$\\mathbf{v}_j = \x0crac{{\\lVert \\mathbf{s}_j \rVert}^2}{1 + {\\lVert \\mathbf{s}_j \rVert}^2}\n \x0crac{\\mathbf{s}_j}{\\lVert \\mathbf{s}_j \rVert}$$\n\n $\x0crac{\\mathbf{s}_j}{\\lVert \\mathbf{s}_j \rVert}$\n normalizes the length of all the capsules, whilst\n $\x0crac{{\\lVert \\mathbf{s}_j \rVert}^2}{1 + {\\lVert \\mathbf{s}_j \rVert}^2}$\n shrinks the capsules that have a length smaller than one .\n '
def __init__(self, epsilon=1e-08):
super().__init__()
self.epsilon = epsilon
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | Squash | false | 7,207 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
'\n ## Squash\n\n This is **squashing** function from paper, given by equation $(1)$.\n\n $$\\mathbf{v}_j = \x0crac{{\\lVert \\mathbf{s}_j \rVert}^2}{1 + {\\lVert \\mathbf{s}_j \rVert}^2}\n \x0crac{\\mathbf{s}_j}{\\lVert \\mathbf{s}_j \rVert}$$\n\n $\x0crac{\\mathbf{s}_j}{\\lVert \\mathbf{s}_j \rVert}$\n normalizes the length of all the capsules, whilst\n $\x0crac{{\\lVert \\mathbf{s}_j \rVert}^2}{1 + {\\lVert \\mathbf{s}_j \rVert}^2}$\n shrinks the capsules that have a length smaller than one .\n '
def __init__(self, epsilon=1e-08):
super().__init__()
self.epsilon = epsilon
def forward(self, s: 'torch.Tensor'):
"""
The shape of `s` is `[batch_size, n_capsules, n_features]`
"""
s2 = (s ** 2).sum(dim=-1, keepdims=True)
return s2 / (1 + s2) * (s / torch.sqrt(s2 + self.epsilon))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
FeedForward | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/hp/chpdwpegv6lvistek2wqgimtufecqvfp6grp5rpblk5yjicjzqd2.py
# Topologically Sorted Source Nodes: [h], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# h => add, rsqrt, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%primals_1, [3]), kwargs = {correction: 0, keepdim: True})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-05), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add,), kwargs = {})
triton_poi_fused_native_layer_norm_0 = async_compile.triton('triton_poi_fused_native_layer_norm_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_0(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x0), tmp8, xmask)
tl.store(out_ptr1 + (x0), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/lh/clhh73owbiuj4adasmetdqsot2nlmw2ljupnw2q4yt3du76mikww.py
# Topologically Sorted Source Nodes: [h], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# h => add, add_1, mul, mul_1, rsqrt, sub, var_mean
# Graph fragment:
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%primals_1, [3]), kwargs = {correction: 0, keepdim: True})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-05), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%primals_1, %getitem_1), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_2), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_3), kwargs = {})
triton_poi_fused_native_layer_norm_1 = async_compile.triton('triton_poi_fused_native_layer_norm_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/gm/cgmflgdlpeeb52xctoa47uvw47ycyf7ahlj5wdscxdatpbwcboco.py
# Topologically Sorted Source Nodes: [h_2], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# h_2 => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_2 = async_compile.triton('triton_poi_fused_relu_threshold_backward_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_2(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/5q/c5qmnkuxezgezseizmolw3mx24fyy6xp3cfoz3egpqwcprxgwjre.py
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
# Source node to ATen node mapping:
# add => add_2
# Graph fragment:
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%view_3, %primals_1), kwargs = {})
triton_poi_fused_add_3 = async_compile.triton('triton_poi_fused_add_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_3(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (x2), xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf1 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
# Topologically Sorted Source Nodes: [h], Original ATen: [aten.native_layer_norm]
stream0 = get_raw_stream(0)
triton_poi_fused_native_layer_norm_0.run(primals_1, buf0, buf1, 64, grid=grid(64), stream=stream0)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [h], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_1.run(primals_1, buf0, buf1, primals_2, primals_3, buf2, 256, grid=grid(256), stream=stream0)
del buf0
del buf1
del primals_2
del primals_3
buf3 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf2, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf3)
buf4 = reinterpret_tensor(buf3, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf3 # reuse
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [h_2], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_2.run(buf4, primals_5, buf7, 256, grid=grid(256), stream=stream0)
del primals_5
buf5 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf4, (64, 4), (4, 1), 0), reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf5)
buf6 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf5 # reuse
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
triton_poi_fused_add_3.run(buf6, primals_7, primals_1, 256, grid=grid(256), stream=stream0)
del primals_7
return (buf6, primals_1, reinterpret_tensor(buf2, (64, 4), (4, 1), 0), reinterpret_tensor(buf4, (64, 4), (4, 1), 0), primals_6, buf7, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class FeedForward(nn.Module):
"""
### Position-wise Feed Forward Layer $ ext{F\\small{FW}}$
This consists of two linear layers and an activation in the middle.
"""
def __init__(self, d_model: 'int', d_ff: 'int'):
"""
* `d_model` is the number of features in transformer embeddings
* `d_ff` is the number features in the hidden layer
"""
super().__init__()
self.lin1 = nn.Linear(d_model, d_ff)
self.lin2 = nn.Linear(d_ff, d_model)
self.act = nn.ReLU()
self.norm = nn.LayerNorm(d_model)
def forward(self, h: 'torch.Tensor'):
"""
`h` are the embeddings of shape `[batch_size, seq_len, d_model]`
"""
h_res = h
h = self.norm(h)
h = self.lin1(h)
h = self.act(h)
h = self.lin2(h)
return h + h_res
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'d_model': 4, 'd_ff': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_native_layer_norm_0(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x0, tmp8, xmask)
tl.store(out_ptr1 + x0, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3,
in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
@triton.jit
def triton_poi_fused_relu_threshold_backward_2(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
@triton.jit
def triton_poi_fused_add_3(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + x2, xmask)
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tl.store(in_out_ptr0 + x2, tmp4, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf1 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
get_raw_stream(0)
triton_poi_fused_native_layer_norm_0[grid(64)](primals_1, buf0,
buf1, 64, XBLOCK=64, num_warps=1, num_stages=1)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_1[grid(256)](primals_1, buf0,
buf1, primals_2, primals_3, buf2, 256, XBLOCK=128, num_warps=4,
num_stages=1)
del buf0
del buf1
del primals_2
del primals_3
buf3 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf2, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf3)
buf4 = reinterpret_tensor(buf3, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf3
buf7 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused_relu_threshold_backward_2[grid(256)](buf4,
primals_5, buf7, 256, XBLOCK=128, num_warps=4, num_stages=1)
del primals_5
buf5 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf4, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf5)
buf6 = reinterpret_tensor(buf5, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf5
triton_poi_fused_add_3[grid(256)](buf6, primals_7, primals_1, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_7
return buf6, primals_1, reinterpret_tensor(buf2, (64, 4), (4, 1), 0
), reinterpret_tensor(buf4, (64, 4), (4, 1), 0
), primals_6, buf7, primals_4
class FeedForwardNew(nn.Module):
"""
### Position-wise Feed Forward Layer $ ext{F\\small{FW}}$
This consists of two linear layers and an activation in the middle.
"""
def __init__(self, d_model: 'int', d_ff: 'int'):
"""
* `d_model` is the number of features in transformer embeddings
* `d_ff` is the number features in the hidden layer
"""
super().__init__()
self.lin1 = nn.Linear(d_model, d_ff)
self.lin2 = nn.Linear(d_ff, d_model)
self.act = nn.ReLU()
self.norm = nn.LayerNorm(d_model)
def forward(self, input_0):
primals_4 = self.lin1.weight
primals_2 = self.lin1.bias
primals_6 = self.lin2.weight
primals_3 = self.lin2.bias
primals_5 = self.norm.weight
primals_7 = self.norm.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | FeedForward | false | 7,208 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(nn.Module):
"""
### Position-wise Feed Forward Layer $ ext{F\\small{FW}}$
This consists of two linear layers and an activation in the middle.
"""
def __init__(self, d_model: 'int', d_ff: 'int'):
"""
* `d_model` is the number of features in transformer embeddings
* `d_ff` is the number features in the hidden layer
"""
super().__init__()
self.lin1 = nn.Linear(d_model, d_ff)
self.lin2 = nn.Linear(d_ff, d_model)
self.act = nn.ReLU()
self.norm = nn.LayerNorm(d_model)
def forward(self, h: 'torch.Tensor'):
"""
`h` are the embeddings of shape `[batch_size, seq_len, d_model]`
"""
h_res = h
h = self.norm(h)
h = self.lin1(h)
h = self.act(h)
h = self.lin2(h)
return h + h_res
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
BehaviorClone | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/z5/cz5bgdo2gmhnnmtf6w7lrjkvliacxo7nomq7mbmjquxqyxqgt5bj.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# x => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_0 = async_compile.triton('triton_poi_fused_relu_threshold_backward_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (2, 4), (4, 1))
assert_size_stride(primals_2, (2, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 2), (2, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 2), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 2), (32, 8, 2, 1), 0); del buf0 # reuse
buf3 = empty_strided_cuda((4, 4, 4, 2), (32, 8, 2, 1), torch.bool)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.relu, aten.threshold_backward]
stream0 = get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0.run(buf1, primals_2, buf3, 128, grid=grid(128), stream=stream0)
del primals_2
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 2), (2, 1), 0), reinterpret_tensor(primals_4, (2, 4), (1, 2), 0), alpha=1, beta=1, out=buf2)
del primals_5
return (reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf1, (64, 2), (2, 1), 0), primals_4, buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((2, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 2), (2, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
class BehaviorClone(nn.Module):
def __init__(self, input_shape, output_shape):
super(BehaviorClone, self).__init__()
self.input_shape = input_shape
self.output_shape = output_shape
self.fc1 = nn.Linear(input_shape, input_shape // 2)
self.fc2 = nn.Linear(input_shape // 2, output_shape)
self.do = nn.Dropout(p=0.3)
def forward(self, x):
x = F.relu(self.fc1(x))
x = self.do(x)
x = self.fc2(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'input_shape': 4, 'output_shape': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_relu_threshold_backward_0(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (2, 4), (4, 1))
assert_size_stride(primals_2, (2,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 2), (2, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 2), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 2), (32, 8, 2, 1), 0)
del buf0
buf3 = empty_strided_cuda((4, 4, 4, 2), (32, 8, 2, 1), torch.bool)
get_raw_stream(0)
triton_poi_fused_relu_threshold_backward_0[grid(128)](buf1,
primals_2, buf3, 128, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 2), (
2, 1), 0), reinterpret_tensor(primals_4, (2, 4), (1, 2), 0),
alpha=1, beta=1, out=buf2)
del primals_5
return reinterpret_tensor(buf2, (4, 4, 4, 4), (64, 16, 4, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), reinterpret_tensor(buf1, (64, 2), (2, 1), 0), primals_4, buf3
class BehaviorCloneNew(nn.Module):
def __init__(self, input_shape, output_shape):
super(BehaviorCloneNew, self).__init__()
self.input_shape = input_shape
self.output_shape = output_shape
self.fc1 = nn.Linear(input_shape, input_shape // 2)
self.fc2 = nn.Linear(input_shape // 2, output_shape)
self.do = nn.Dropout(p=0.3)
def forward(self, input_0):
primals_1 = self.fc1.weight
primals_2 = self.fc1.bias
primals_4 = self.fc2.weight
primals_5 = self.fc2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| mdiephuis/Berkeley-cs294-112 | BehaviorClone | false | 7,209 | [
"MIT"
] | 1 | 99559e046b635ca8d229f19ca4ad45c2c02a1c01 | https://github.com/mdiephuis/Berkeley-cs294-112/tree/99559e046b635ca8d229f19ca4ad45c2c02a1c01 | import torch
import torch.nn as nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self, input_shape, output_shape):
super().__init__()
self.input_shape = input_shape
self.output_shape = output_shape
self.fc1 = nn.Linear(input_shape, input_shape // 2)
self.fc2 = nn.Linear(input_shape // 2, output_shape)
self.do = nn.Dropout(p=0.3)
def forward(self, x):
x = F.relu(self.fc1(x))
x = self.do(x)
x = self.fc2(x)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
SpatialDepthWiseConvolution | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/6j/c6jwd535crsaxcf2vn6fo7tivf6rewo3kc7wmj2smnibmmu6vbwp.py
# Topologically Sorted Source Nodes: [res, mul_1, iadd, mul_2, iadd_1], Original ATen: [aten.mul, aten.add]
# Source node to ATen node mapping:
# iadd => add
# iadd_1 => add_1
# mul_1 => mul_1
# mul_2 => mul_2
# res => mul
# Graph fragment:
# %mul : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_2, %view), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%slice_2, %view_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%slice_1, %mul_1), kwargs = {})
# %slice_scatter_default : [num_users=2] = call_function[target=torch.ops.aten.slice_scatter.default](args = (%mul, %add, 0, 1, 9223372036854775807), kwargs = {})
# %slice_scatter_default_1 : [num_users=2] = call_function[target=torch.ops.aten.slice_scatter.default](args = (%slice_scatter_default, %slice_3, 0, 1, 9223372036854775807), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%slice_10, %view_2), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%slice_11, %mul_2), kwargs = {})
# %slice_scatter_default_2 : [num_users=2] = call_function[target=torch.ops.aten.slice_scatter.default](args = (%slice_scatter_default_1, %add_1, 0, 2, 9223372036854775807), kwargs = {})
# %slice_scatter_default_3 : [num_users=1] = call_function[target=torch.ops.aten.slice_scatter.default](args = (%slice_scatter_default_2, %slice_12, 0, 2, 9223372036854775807), kwargs = {})
triton_poi_fused_add_mul_0 = async_compile.triton('triton_poi_fused_add_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 22, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = (xindex // 64)
x4 = xindex
x0 = xindex % 4
tmp61 = tl.load(in_ptr0 + (x4), xmask)
tmp62 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp0 = x2
tmp1 = tl.full([1], 2, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 >= tmp3
tmp5 = tmp4 & tmp2
tmp6 = tmp4 & tmp5
tmp7 = tl.load(in_ptr0 + (x4), tmp6 & xmask, other=0.0)
tmp8 = tl.load(in_ptr1 + (x0), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp9 = tmp7 * tmp8
tmp10 = tl.load(in_ptr0 + ((-64) + x4), tmp6 & xmask, other=0.0)
tmp11 = tl.load(in_ptr1 + (4 + x0), tmp6 & xmask, eviction_policy='evict_last', other=0.0)
tmp12 = tmp10 * tmp11
tmp13 = tmp9 + tmp12
tmp14 = tl.full(tmp13.shape, 0.0, tmp13.dtype)
tmp15 = tl.where(tmp6, tmp13, tmp14)
tmp16 = tl.load(in_ptr0 + (x4), tmp5 & xmask, other=0.0)
tmp17 = tl.load(in_ptr1 + (x0), tmp5 & xmask, eviction_policy='evict_last', other=0.0)
tmp18 = tmp16 * tmp17
tmp19 = tl.where(tmp4, tmp15, tmp18)
tmp20 = tl.full(tmp19.shape, 0.0, tmp19.dtype)
tmp21 = tl.where(tmp5, tmp19, tmp20)
tmp22 = tl.load(in_ptr0 + ((-64) + x4), tmp5 & xmask, other=0.0)
tmp23 = tl.load(in_ptr1 + (4 + x0), tmp5 & xmask, eviction_policy='evict_last', other=0.0)
tmp24 = tmp22 * tmp23
tmp25 = tmp18 + tmp24
tmp26 = tl.full(tmp25.shape, 0.0, tmp25.dtype)
tmp27 = tl.where(tmp5, tmp25, tmp26)
tmp28 = tl.load(in_ptr0 + (x4), tmp2 & xmask, other=0.0)
tmp29 = tl.load(in_ptr1 + (x0), tmp2 & xmask, eviction_policy='evict_last', other=0.0)
tmp30 = tmp28 * tmp29
tmp31 = tl.where(tmp4, tmp27, tmp30)
tmp32 = tl.where(tmp4, tmp21, tmp31)
tmp33 = tl.load(in_ptr0 + ((-128) + x4), tmp2 & xmask, other=0.0)
tmp34 = tl.load(in_ptr1 + (8 + x0), tmp2 & xmask, eviction_policy='evict_last', other=0.0)
tmp35 = tmp33 * tmp34
tmp36 = tmp32 + tmp35
tmp37 = tl.full(tmp36.shape, 0.0, tmp36.dtype)
tmp38 = tl.where(tmp2, tmp36, tmp37)
tmp39 = tmp4 & tmp4
tmp40 = tl.load(in_ptr0 + (x4), tmp39 & xmask, other=0.0)
tmp41 = tl.load(in_ptr1 + (x0), tmp39 & xmask, eviction_policy='evict_last', other=0.0)
tmp42 = tmp40 * tmp41
tmp43 = tl.load(in_ptr0 + ((-64) + x4), tmp39 & xmask, other=0.0)
tmp44 = tl.load(in_ptr1 + (4 + x0), tmp39 & xmask, eviction_policy='evict_last', other=0.0)
tmp45 = tmp43 * tmp44
tmp46 = tmp42 + tmp45
tmp47 = tl.full(tmp46.shape, 0.0, tmp46.dtype)
tmp48 = tl.where(tmp39, tmp46, tmp47)
tmp49 = tl.load(in_ptr0 + (x4), tmp4 & xmask, other=0.0)
tmp50 = tl.load(in_ptr1 + (x0), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp51 = tmp49 * tmp50
tmp52 = tl.where(tmp4, tmp48, tmp51)
tmp53 = tl.full(tmp52.shape, 0.0, tmp52.dtype)
tmp54 = tl.where(tmp4, tmp52, tmp53)
tmp55 = tl.load(in_ptr0 + ((-64) + x4), tmp4 & xmask, other=0.0)
tmp56 = tl.load(in_ptr1 + (4 + x0), tmp4 & xmask, eviction_policy='evict_last', other=0.0)
tmp57 = tmp55 * tmp56
tmp58 = tmp51 + tmp57
tmp59 = tl.full(tmp58.shape, 0.0, tmp58.dtype)
tmp60 = tl.where(tmp4, tmp58, tmp59)
tmp63 = tmp61 * tmp62
tmp64 = tl.where(tmp4, tmp60, tmp63)
tmp65 = tl.where(tmp4, tmp54, tmp64)
tmp66 = tl.where(tmp2, tmp38, tmp65)
tmp67 = tl.where(tmp2, tmp66, tmp66)
tl.store(in_out_ptr0 + (x4), tmp67, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (3, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf1 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [res, mul_1, iadd, mul_2, iadd_1], Original ATen: [aten.mul, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_mul_0.run(buf1, primals_2, primals_1, 256, grid=grid(256), stream=stream0)
del primals_1
return (buf1, primals_2, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((3, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import math
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class SpatialDepthWiseConvolution(Module):
"""
## Spatial Depth Wise Convolution
This is actually slower
"""
def __init__(self, d_k: 'int', kernel_size: 'int'=3):
"""
* `d_k` is the number of channels in each head
"""
super().__init__()
self.kernel_size = kernel_size
rng = 1 / math.sqrt(kernel_size)
self.kernels = nn.Parameter(torch.zeros((kernel_size, d_k)).
uniform_(-rng, rng))
def forward(self, x: 'torch.Tensor'):
"""
`x` has shape `[seq_len, batch_size, heads, d_k]`
"""
res = x * self.kernels[0].view(1, 1, 1, -1)
for i in range(1, len(self.kernels)):
res[i:] += x[:-i] * self.kernels[i].view(1, 1, 1, -1)
return res
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'d_k': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn import Module
import math
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_mul_0(in_out_ptr0, in_ptr0, in_ptr1, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex // 64
x4 = xindex
x0 = xindex % 4
tmp61 = tl.load(in_ptr0 + x4, xmask)
tmp62 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp0 = x2
tmp1 = tl.full([1], 2, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 1, tl.int64)
tmp4 = tmp0 >= tmp3
tmp5 = tmp4 & tmp2
tmp6 = tmp4 & tmp5
tmp7 = tl.load(in_ptr0 + x4, tmp6 & xmask, other=0.0)
tmp8 = tl.load(in_ptr1 + x0, tmp6 & xmask, eviction_policy='evict_last',
other=0.0)
tmp9 = tmp7 * tmp8
tmp10 = tl.load(in_ptr0 + (-64 + x4), tmp6 & xmask, other=0.0)
tmp11 = tl.load(in_ptr1 + (4 + x0), tmp6 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp12 = tmp10 * tmp11
tmp13 = tmp9 + tmp12
tmp14 = tl.full(tmp13.shape, 0.0, tmp13.dtype)
tmp15 = tl.where(tmp6, tmp13, tmp14)
tmp16 = tl.load(in_ptr0 + x4, tmp5 & xmask, other=0.0)
tmp17 = tl.load(in_ptr1 + x0, tmp5 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp18 = tmp16 * tmp17
tmp19 = tl.where(tmp4, tmp15, tmp18)
tmp20 = tl.full(tmp19.shape, 0.0, tmp19.dtype)
tmp21 = tl.where(tmp5, tmp19, tmp20)
tmp22 = tl.load(in_ptr0 + (-64 + x4), tmp5 & xmask, other=0.0)
tmp23 = tl.load(in_ptr1 + (4 + x0), tmp5 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp24 = tmp22 * tmp23
tmp25 = tmp18 + tmp24
tmp26 = tl.full(tmp25.shape, 0.0, tmp25.dtype)
tmp27 = tl.where(tmp5, tmp25, tmp26)
tmp28 = tl.load(in_ptr0 + x4, tmp2 & xmask, other=0.0)
tmp29 = tl.load(in_ptr1 + x0, tmp2 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp30 = tmp28 * tmp29
tmp31 = tl.where(tmp4, tmp27, tmp30)
tmp32 = tl.where(tmp4, tmp21, tmp31)
tmp33 = tl.load(in_ptr0 + (-128 + x4), tmp2 & xmask, other=0.0)
tmp34 = tl.load(in_ptr1 + (8 + x0), tmp2 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp35 = tmp33 * tmp34
tmp36 = tmp32 + tmp35
tmp37 = tl.full(tmp36.shape, 0.0, tmp36.dtype)
tmp38 = tl.where(tmp2, tmp36, tmp37)
tmp39 = tmp4 & tmp4
tmp40 = tl.load(in_ptr0 + x4, tmp39 & xmask, other=0.0)
tmp41 = tl.load(in_ptr1 + x0, tmp39 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp42 = tmp40 * tmp41
tmp43 = tl.load(in_ptr0 + (-64 + x4), tmp39 & xmask, other=0.0)
tmp44 = tl.load(in_ptr1 + (4 + x0), tmp39 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp45 = tmp43 * tmp44
tmp46 = tmp42 + tmp45
tmp47 = tl.full(tmp46.shape, 0.0, tmp46.dtype)
tmp48 = tl.where(tmp39, tmp46, tmp47)
tmp49 = tl.load(in_ptr0 + x4, tmp4 & xmask, other=0.0)
tmp50 = tl.load(in_ptr1 + x0, tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp51 = tmp49 * tmp50
tmp52 = tl.where(tmp4, tmp48, tmp51)
tmp53 = tl.full(tmp52.shape, 0.0, tmp52.dtype)
tmp54 = tl.where(tmp4, tmp52, tmp53)
tmp55 = tl.load(in_ptr0 + (-64 + x4), tmp4 & xmask, other=0.0)
tmp56 = tl.load(in_ptr1 + (4 + x0), tmp4 & xmask, eviction_policy=
'evict_last', other=0.0)
tmp57 = tmp55 * tmp56
tmp58 = tmp51 + tmp57
tmp59 = tl.full(tmp58.shape, 0.0, tmp58.dtype)
tmp60 = tl.where(tmp4, tmp58, tmp59)
tmp63 = tmp61 * tmp62
tmp64 = tl.where(tmp4, tmp60, tmp63)
tmp65 = tl.where(tmp4, tmp54, tmp64)
tmp66 = tl.where(tmp2, tmp38, tmp65)
tmp67 = tl.where(tmp2, tmp66, tmp66)
tl.store(in_out_ptr0 + x4, tmp67, xmask)
def call(args):
primals_1, primals_2 = args
args.clear()
assert_size_stride(primals_1, (3, 4), (4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf1 = buf0
del buf0
get_raw_stream(0)
triton_poi_fused_add_mul_0[grid(256)](buf1, primals_2, primals_1,
256, XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
return buf1, primals_2
class SpatialDepthWiseConvolutionNew(Module):
"""
## Spatial Depth Wise Convolution
This is actually slower
"""
def __init__(self, d_k: 'int', kernel_size: 'int'=3):
"""
* `d_k` is the number of channels in each head
"""
super().__init__()
self.kernel_size = kernel_size
rng = 1 / math.sqrt(kernel_size)
self.kernels = nn.Parameter(torch.zeros((kernel_size, d_k)).
uniform_(-rng, rng))
def forward(self, input_0):
primals_1 = self.kernels
primals_2 = input_0
output = call([primals_1, primals_2])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | SpatialDepthWiseConvolution | false | 7,210 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import math
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Spatial Depth Wise Convolution
This is actually slower
"""
def __init__(self, d_k: 'int', kernel_size: 'int'=3):
"""
* `d_k` is the number of channels in each head
"""
super().__init__()
self.kernel_size = kernel_size
rng = 1 / math.sqrt(kernel_size)
self.kernels = nn.Parameter(torch.zeros((kernel_size, d_k)).
uniform_(-rng, rng))
def forward(self, x: 'torch.Tensor'):
"""
`x` has shape `[seq_len, batch_size, heads, d_k]`
"""
res = x * self.kernels[0].view(1, 1, 1, -1)
for i in range(1, len(self.kernels)):
res[i:] += x[:-i] * self.kernels[i].view(1, 1, 1, -1)
return res
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
KLDivergenceLoss | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ak/cakm5e6lwgvo4ch6lbxjmcmcar3oco47d3xyxuief4vqkwlft225.py
# Topologically Sorted Source Nodes: [sub, alpha, mul, alpha_tilde], Original ATen: [aten.rsub, aten.add, aten.mul]
# Source node to ATen node mapping:
# alpha => add
# alpha_tilde => add_1
# mul => mul
# sub => sub
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %arg1_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, 1.0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %add), kwargs = {})
# %add_1 : [num_users=5] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg1_1, %mul), kwargs = {})
triton_poi_fused_add_mul_rsub_0 = async_compile.triton('triton_poi_fused_add_mul_rsub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_rsub_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_rsub_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp3 = tl.load(in_ptr1 + (x0), xmask)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp3 + tmp1
tmp5 = tmp2 * tmp4
tmp6 = tmp0 + tmp5
tl.store(out_ptr0 + (x0), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/fs/cfslznycxlykpj42fqvbg2sgr2rr5ofx6xol4bhm6igaw3efhsg5.py
# Topologically Sorted Source Nodes: [strength_tilde], Original ATen: [aten.sum]
# Source node to ATen node mapping:
# strength_tilde => sum_1
# Graph fragment:
# %sum_1 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%add_1, [-1]), kwargs = {})
triton_poi_fused_sum_1 = async_compile.triton('triton_poi_fused_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sum_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sum_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tl.store(out_ptr0 + (x0), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/2i/c2izsdx2sqgkyo4pbm7trfgkofg2h276vcut2cdf4alkpa2abztc.py
# Topologically Sorted Source Nodes: [sub, alpha, mul, alpha_tilde, sum_2, lgamma, lgamma_1, sub_1, lgamma_2, sum_3, first, sub_3, sub_4, mul_1, second, loss, mean], Original ATen: [aten.rsub, aten.add, aten.mul, aten.sum, aten.lgamma, aten.sub, aten.mean]
# Source node to ATen node mapping:
# alpha => add
# alpha_tilde => add_1
# first => sub_2
# lgamma => lgamma
# lgamma_1 => full_default
# lgamma_2 => lgamma_2
# loss => add_2
# mean => mean
# mul => mul
# mul_1 => mul_1
# second => sum_4
# sub => sub
# sub_1 => sub_1
# sub_3 => sub_3
# sub_4 => sub_4
# sum_2 => sum_2
# sum_3 => sum_3
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %arg1_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, 1.0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %add), kwargs = {})
# %add_1 : [num_users=5] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg1_1, %mul), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%add_1, [-1]), kwargs = {})
# %lgamma : [num_users=1] = call_function[target=torch.ops.aten.lgamma.default](args = (%sum_2,), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 1.7917594909667969), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%lgamma, %full_default), kwargs = {})
# %lgamma_2 : [num_users=1] = call_function[target=torch.ops.aten.lgamma.default](args = (%add_1,), kwargs = {})
# %sum_3 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%lgamma_2, [-1]), kwargs = {})
# %sub_2 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%sub_1, %sum_3), kwargs = {})
# %sub_3 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_1, 1), kwargs = {})
# %sub_4 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%digamma, %unsqueeze), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_3, %sub_4), kwargs = {})
# %sum_4 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%mul_1, [-1]), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%sub_2, %sum_4), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%add_2,), kwargs = {})
triton_per_fused_add_lgamma_mean_mul_rsub_sub_sum_2 = async_compile.triton('triton_per_fused_add_lgamma_mean_mul_rsub_sub_sum_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: 'i32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {5: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 6), equal_to_1=(5,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_lgamma_mean_mul_rsub_sub_sum_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 16, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_lgamma_mean_mul_rsub_sub_sum_2(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
r1 = rindex % 4
r3 = (rindex // 16)
tmp0 = tl.load(in_ptr0 + (4*r0), None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + (4*r0), None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (1 + (4*r0)), None, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr1 + (1 + (4*r0)), None, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr0 + (2 + (4*r0)), None, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr1 + (2 + (4*r0)), None, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (3 + (4*r0)), None, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr1 + (3 + (4*r0)), None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr2 + (4*r0), None, eviction_policy='evict_last')
tmp40 = tl.load(in_ptr3 + ((4*r1) + (16*r3)), None, eviction_policy='evict_last')
tmp44 = tl.load(in_ptr2 + (1 + (4*r0)), None, eviction_policy='evict_last')
tmp45 = tl.load(in_ptr3 + (1 + (4*r1) + (16*r3)), None, eviction_policy='evict_last')
tmp50 = tl.load(in_ptr2 + (2 + (4*r0)), None, eviction_policy='evict_last')
tmp51 = tl.load(in_ptr3 + (2 + (4*r1) + (16*r3)), None, eviction_policy='evict_last')
tmp56 = tl.load(in_ptr2 + (3 + (4*r0)), None, eviction_policy='evict_last')
tmp57 = tl.load(in_ptr3 + (3 + (4*r1) + (16*r3)), None, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp3 + tmp1
tmp5 = tmp2 * tmp4
tmp6 = tmp0 + tmp5
tmp8 = tmp1 - tmp7
tmp10 = tmp9 + tmp1
tmp11 = tmp8 * tmp10
tmp12 = tmp7 + tmp11
tmp13 = tmp6 + tmp12
tmp15 = tmp1 - tmp14
tmp17 = tmp16 + tmp1
tmp18 = tmp15 * tmp17
tmp19 = tmp14 + tmp18
tmp20 = tmp13 + tmp19
tmp22 = tmp1 - tmp21
tmp24 = tmp23 + tmp1
tmp25 = tmp22 * tmp24
tmp26 = tmp21 + tmp25
tmp27 = tmp20 + tmp26
tmp28 = libdevice.lgamma(tmp27)
tmp29 = 1.7917594909667969
tmp30 = tmp28 - tmp29
tmp31 = libdevice.lgamma(tmp6)
tmp32 = libdevice.lgamma(tmp12)
tmp33 = tmp31 + tmp32
tmp34 = libdevice.lgamma(tmp19)
tmp35 = tmp33 + tmp34
tmp36 = libdevice.lgamma(tmp26)
tmp37 = tmp35 + tmp36
tmp38 = tmp6 - tmp1
tmp41 = tmp39 - tmp40
tmp42 = tmp38 * tmp41
tmp43 = tmp12 - tmp1
tmp46 = tmp44 - tmp45
tmp47 = tmp43 * tmp46
tmp48 = tmp42 + tmp47
tmp49 = tmp19 - tmp1
tmp52 = tmp50 - tmp51
tmp53 = tmp49 * tmp52
tmp54 = tmp48 + tmp53
tmp55 = tmp26 - tmp1
tmp58 = tmp56 - tmp57
tmp59 = tmp55 * tmp58
tmp60 = tmp54 + tmp59
tmp61 = tmp30 - tmp37
tmp62 = tmp61 + tmp60
tmp63 = tl.broadcast_to(tmp62, [XBLOCK, RBLOCK])
tmp65 = tl.sum(tmp63, 1)[:, None]
tmp66 = 64.0
tmp67 = tmp65 / tmp66
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp67, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [sub, alpha, mul, alpha_tilde], Original ATen: [aten.rsub, aten.add, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_add_mul_rsub_0.run(arg1_1, arg0_1, buf2, 256, grid=grid(256), stream=stream0)
# Topologically Sorted Source Nodes: [sub, alpha, mul, alpha_tilde, digamma], Original ATen: [aten.rsub, aten.add, aten.mul, aten.digamma]
buf3 = torch.ops.aten.digamma.default(buf2)
buf4 = buf3
del buf3
buf5 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [strength_tilde], Original ATen: [aten.sum]
triton_poi_fused_sum_1.run(buf2, buf5, 64, grid=grid(64), stream=stream0)
del buf2
# Topologically Sorted Source Nodes: [strength_tilde, digamma_1], Original ATen: [aten.sum, aten.digamma]
buf6 = torch.ops.aten.digamma.default(buf5)
del buf5
buf7 = buf6
del buf6
buf9 = empty_strided_cuda((), (), torch.float32)
buf10 = buf9; del buf9 # reuse
# Topologically Sorted Source Nodes: [sub, alpha, mul, alpha_tilde, sum_2, lgamma, lgamma_1, sub_1, lgamma_2, sum_3, first, sub_3, sub_4, mul_1, second, loss, mean], Original ATen: [aten.rsub, aten.add, aten.mul, aten.sum, aten.lgamma, aten.sub, aten.mean]
triton_per_fused_add_lgamma_mean_mul_rsub_sub_sum_2.run(buf10, arg1_1, arg0_1, buf4, buf7, 1, 64, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
del buf4
del buf7
return (buf10, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class KLDivergenceLoss(Module):
"""
<a id="KLDivergenceLoss"></a>
## KL Divergence Regularization Loss
This tries to shrink the total evidence to zero if the sample cannot be correctly classified.
First we calculate $ ilde{lpha}_k = y_k + (1 - y_k) extcolor{orange}{lpha_k}$ the
Dirichlet parameters after remove the correct evidence.
egin{align}
&KL \\Big[ D(\\mathbf{p} ert \\mathbf{ ilde{lpha}}) \\Big \\Vert
D(\\mathbf{p} ert <1, \\dots, 1>\\Big] \\
&= \\log \\Bigg( rac{\\Gamma \\Big( \\sum_{k=1}^K ilde{lpha}_k \\Big)}
{\\Gamma(K) \\prod_{k=1}^K \\Gamma( ilde{lpha}_k)} \\Bigg)
+ \\sum_{k=1}^K ( ilde{lpha}_k - 1)
\\Big[ \\psi( ilde{lpha}_k) - \\psi( ilde{S}) \\Big]
\\end{align}
where $\\Gamma(\\cdot)$ is the gamma function,
$\\psi(\\cdot)$ is the $digamma$ function and
$ ilde{S} = \\sum_{k=1}^K ilde{lpha}_k$
"""
def forward(self, evidence: 'torch.Tensor', target: 'torch.Tensor'):
"""
* `evidence` is $\\mathbf{e} \\ge 0$ with shape `[batch_size, n_classes]`
* `target` is $\\mathbf{y}$ with shape `[batch_size, n_classes]`
"""
alpha = evidence + 1.0
n_classes = evidence.shape[-1]
alpha_tilde = target + (1 - target) * alpha
strength_tilde = alpha_tilde.sum(dim=-1)
first = torch.lgamma(alpha_tilde.sum(dim=-1)) - torch.lgamma(
alpha_tilde.new_tensor(float(n_classes))) - torch.lgamma(
alpha_tilde).sum(dim=-1)
second = ((alpha_tilde - 1) * (torch.digamma(alpha_tilde) - torch.
digamma(strength_tilde)[:, None])).sum(dim=-1)
loss = first + second
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_mul_rsub_0(in_ptr0, in_ptr1, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp3 = tl.load(in_ptr1 + x0, xmask)
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp3 + tmp1
tmp5 = tmp2 * tmp4
tmp6 = tmp0 + tmp5
tl.store(out_ptr0 + x0, tmp6, xmask)
@triton.jit
def triton_poi_fused_sum_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tl.store(out_ptr0 + x0, tmp6, xmask)
@triton.jit
def triton_per_fused_add_lgamma_mean_mul_rsub_sub_sum_2(in_out_ptr0,
in_ptr0, in_ptr1, in_ptr2, in_ptr3, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r0 = rindex
r1 = rindex % 4
r3 = rindex // 16
tmp0 = tl.load(in_ptr0 + 4 * r0, None, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr1 + 4 * r0, None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (1 + 4 * r0), None, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr1 + (1 + 4 * r0), None, eviction_policy='evict_last')
tmp14 = tl.load(in_ptr0 + (2 + 4 * r0), None, eviction_policy='evict_last')
tmp16 = tl.load(in_ptr1 + (2 + 4 * r0), None, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (3 + 4 * r0), None, eviction_policy='evict_last')
tmp23 = tl.load(in_ptr1 + (3 + 4 * r0), None, eviction_policy='evict_last')
tmp39 = tl.load(in_ptr2 + 4 * r0, None, eviction_policy='evict_last')
tmp40 = tl.load(in_ptr3 + (4 * r1 + 16 * r3), None, eviction_policy=
'evict_last')
tmp44 = tl.load(in_ptr2 + (1 + 4 * r0), None, eviction_policy='evict_last')
tmp45 = tl.load(in_ptr3 + (1 + 4 * r1 + 16 * r3), None, eviction_policy
='evict_last')
tmp50 = tl.load(in_ptr2 + (2 + 4 * r0), None, eviction_policy='evict_last')
tmp51 = tl.load(in_ptr3 + (2 + 4 * r1 + 16 * r3), None, eviction_policy
='evict_last')
tmp56 = tl.load(in_ptr2 + (3 + 4 * r0), None, eviction_policy='evict_last')
tmp57 = tl.load(in_ptr3 + (3 + 4 * r1 + 16 * r3), None, eviction_policy
='evict_last')
tmp1 = 1.0
tmp2 = tmp1 - tmp0
tmp4 = tmp3 + tmp1
tmp5 = tmp2 * tmp4
tmp6 = tmp0 + tmp5
tmp8 = tmp1 - tmp7
tmp10 = tmp9 + tmp1
tmp11 = tmp8 * tmp10
tmp12 = tmp7 + tmp11
tmp13 = tmp6 + tmp12
tmp15 = tmp1 - tmp14
tmp17 = tmp16 + tmp1
tmp18 = tmp15 * tmp17
tmp19 = tmp14 + tmp18
tmp20 = tmp13 + tmp19
tmp22 = tmp1 - tmp21
tmp24 = tmp23 + tmp1
tmp25 = tmp22 * tmp24
tmp26 = tmp21 + tmp25
tmp27 = tmp20 + tmp26
tmp28 = libdevice.lgamma(tmp27)
tmp29 = 1.7917594909667969
tmp30 = tmp28 - tmp29
tmp31 = libdevice.lgamma(tmp6)
tmp32 = libdevice.lgamma(tmp12)
tmp33 = tmp31 + tmp32
tmp34 = libdevice.lgamma(tmp19)
tmp35 = tmp33 + tmp34
tmp36 = libdevice.lgamma(tmp26)
tmp37 = tmp35 + tmp36
tmp38 = tmp6 - tmp1
tmp41 = tmp39 - tmp40
tmp42 = tmp38 * tmp41
tmp43 = tmp12 - tmp1
tmp46 = tmp44 - tmp45
tmp47 = tmp43 * tmp46
tmp48 = tmp42 + tmp47
tmp49 = tmp19 - tmp1
tmp52 = tmp50 - tmp51
tmp53 = tmp49 * tmp52
tmp54 = tmp48 + tmp53
tmp55 = tmp26 - tmp1
tmp58 = tmp56 - tmp57
tmp59 = tmp55 * tmp58
tmp60 = tmp54 + tmp59
tmp61 = tmp30 - tmp37
tmp62 = tmp61 + tmp60
tmp63 = tl.broadcast_to(tmp62, [XBLOCK, RBLOCK])
tmp65 = tl.sum(tmp63, 1)[:, None]
tmp66 = 64.0
tmp67 = tmp65 / tmp66
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp67, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_mul_rsub_0[grid(256)](arg1_1, arg0_1, buf2,
256, XBLOCK=128, num_warps=4, num_stages=1)
buf3 = torch.ops.aten.digamma.default(buf2)
buf4 = buf3
del buf3
buf5 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
triton_poi_fused_sum_1[grid(64)](buf2, buf5, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del buf2
buf6 = torch.ops.aten.digamma.default(buf5)
del buf5
buf7 = buf6
del buf6
buf9 = empty_strided_cuda((), (), torch.float32)
buf10 = buf9
del buf9
triton_per_fused_add_lgamma_mean_mul_rsub_sub_sum_2[grid(1)](buf10,
arg1_1, arg0_1, buf4, buf7, 1, 64, XBLOCK=1, num_warps=2,
num_stages=1)
del arg0_1
del arg1_1
del buf4
del buf7
return buf10,
class KLDivergenceLossNew(Module):
"""
<a id="KLDivergenceLoss"></a>
## KL Divergence Regularization Loss
This tries to shrink the total evidence to zero if the sample cannot be correctly classified.
First we calculate $ ilde{lpha}_k = y_k + (1 - y_k) extcolor{orange}{lpha_k}$ the
Dirichlet parameters after remove the correct evidence.
egin{align}
&KL \\Big[ D(\\mathbf{p} ert \\mathbf{ ilde{lpha}}) \\Big \\Vert
D(\\mathbf{p} ert <1, \\dots, 1>\\Big] \\
&= \\log \\Bigg( rac{\\Gamma \\Big( \\sum_{k=1}^K ilde{lpha}_k \\Big)}
{\\Gamma(K) \\prod_{k=1}^K \\Gamma( ilde{lpha}_k)} \\Bigg)
+ \\sum_{k=1}^K ( ilde{lpha}_k - 1)
\\Big[ \\psi( ilde{lpha}_k) - \\psi( ilde{S}) \\Big]
\\end{align}
where $\\Gamma(\\cdot)$ is the gamma function,
$\\psi(\\cdot)$ is the $digamma$ function and
$ ilde{S} = \\sum_{k=1}^K ilde{lpha}_k$
"""
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | KLDivergenceLoss | false | 7,211 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
<a id="KLDivergenceLoss"></a>
## KL Divergence Regularization Loss
This tries to shrink the total evidence to zero if the sample cannot be correctly classified.
First we calculate $ ilde{lpha}_k = y_k + (1 - y_k) extcolor{orange}{lpha_k}$ the
Dirichlet parameters after remove the correct evidence.
egin{align}
&KL \\Big[ D(\\mathbf{p} ert \\mathbf{ ilde{lpha}}) \\Big \\Vert
D(\\mathbf{p} ert <1, \\dots, 1>\\Big] \\
&= \\log \\Bigg( rac{\\Gamma \\Big( \\sum_{k=1}^K ilde{lpha}_k \\Big)}
{\\Gamma(K) \\prod_{k=1}^K \\Gamma( ilde{lpha}_k)} \\Bigg)
+ \\sum_{k=1}^K ( ilde{lpha}_k - 1)
\\Big[ \\psi( ilde{lpha}_k) - \\psi( ilde{S}) \\Big]
\\end{align}
where $\\Gamma(\\cdot)$ is the gamma function,
$\\psi(\\cdot)$ is the $digamma$ function and
$ ilde{S} = \\sum_{k=1}^K ilde{lpha}_k$
"""
def forward(self, evidence: 'torch.Tensor', target: 'torch.Tensor'):
"""
* `evidence` is $\\mathbf{e} \\ge 0$ with shape `[batch_size, n_classes]`
* `target` is $\\mathbf{y}$ with shape `[batch_size, n_classes]`
"""
alpha = evidence + 1.0
n_classes = evidence.shape[-1]
alpha_tilde = target + (1 - target) * alpha
strength_tilde = alpha_tilde.sum(dim=-1)
first = torch.lgamma(alpha_tilde.sum(dim=-1)) - torch.lgamma(
alpha_tilde.new_tensor(float(n_classes))) - torch.lgamma(
alpha_tilde).sum(dim=-1)
second = ((alpha_tilde - 1) * (torch.digamma(alpha_tilde) - torch.
digamma(strength_tilde)[:, None])).sum(dim=-1)
loss = first + second
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
SpatialDepthWisePerHeadConvolution | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/jb/cjbf3ssum7resbwampiwoknxcnzh4uzdy4fhoaakjojloew6qlw5.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x_2 => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%view, %primals_2, %primals_3, [1], [2], [1], False, [0], 16), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (64*x1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x1 + (4*y0)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/6z/c6zyvii2c5e5dc43sgzfkfbsfrhltqlojfzhuoqqgpavg7xdvriv.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x_2 => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%view, %primals_2, %primals_3, [1], [2], [1], False, [0], 16), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[512],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 384
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 6) % 16
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (16, 1, 3), (3, 3, 1))
assert_size_stride(primals_3, (16, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 16, 4), (64, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(primals_1, buf0, 64, 4, grid=grid(64, 4), stream=stream0)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1,), padding=(2,), dilation=(1,), transposed=False, output_padding=(0,), groups=16, bias=None)
assert_size_stride(buf1, (4, 16, 6), (96, 6, 1))
del buf0
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf2, primals_3, 384, grid=grid(384), stream=stream0)
del primals_3
return (reinterpret_tensor(buf2, (4, 4, 4, 4), (1, 96, 24, 6), 0), primals_2, reinterpret_tensor(primals_1, (4, 16, 4), (16, 1, 64), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((16, 1, 3), (3, 3, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((16, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class SpatialDepthWisePerHeadConvolution(Module):
"""
## Spatial Depth Wise Per Head Convolution
"""
def __init__(self, heads: 'int', d_k: 'int', kernel_size: 'int'=3):
"""
* `heads` is the number of heads
* `d_k` is the number of channels in each head
"""
super().__init__()
self.kernel_size = kernel_size
self.conv = nn.Conv1d(in_channels=d_k * heads, out_channels=d_k *
heads, kernel_size=(kernel_size,), padding=(kernel_size - 1,),
groups=d_k * heads)
def forward(self, x: 'torch.Tensor'):
"""
`x` has shape `[seq_len, batch_size, heads, d_k]`
"""
seq_len, batch_size, heads, d_k = x.shape
x = x.permute(1, 2, 3, 0)
x = x.view(batch_size, heads * d_k, seq_len)
x = self.conv(x)
x = x[:, :, :-(self.kernel_size - 1)]
x = x.view(batch_size, heads, d_k, seq_len)
x = x.permute(3, 0, 1, 2)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'heads': 4, 'd_k': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_0(in_ptr0, out_ptr0, ynumel, xnumel,
YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 64 * x1), xmask & ymask, eviction_policy
='evict_last')
tl.store(out_ptr0 + (x1 + 4 * y0), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 384
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 6 % 16
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (16, 1, 3), (3, 3, 1))
assert_size_stride(primals_3, (16,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 16, 4), (64, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(64, 4)](primals_1, buf0, 64, 4,
XBLOCK=4, YBLOCK=32, num_warps=4, num_stages=1)
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1,),
padding=(2,), dilation=(1,), transposed=False, output_padding=(
0,), groups=16, bias=None)
assert_size_stride(buf1, (4, 16, 6), (96, 6, 1))
del buf0
buf2 = buf1
del buf1
triton_poi_fused_convolution_1[grid(384)](buf2, primals_3, 384,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_3
return reinterpret_tensor(buf2, (4, 4, 4, 4), (1, 96, 24, 6), 0
), primals_2, reinterpret_tensor(primals_1, (4, 16, 4), (16, 1, 64), 0)
class SpatialDepthWisePerHeadConvolutionNew(Module):
"""
## Spatial Depth Wise Per Head Convolution
"""
def __init__(self, heads: 'int', d_k: 'int', kernel_size: 'int'=3):
"""
* `heads` is the number of heads
* `d_k` is the number of channels in each head
"""
super().__init__()
self.kernel_size = kernel_size
self.conv = nn.Conv1d(in_channels=d_k * heads, out_channels=d_k *
heads, kernel_size=(kernel_size,), padding=(kernel_size - 1,),
groups=d_k * heads)
def forward(self, input_0):
primals_2 = self.conv.weight
primals_3 = self.conv.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | SpatialDepthWisePerHeadConvolution | false | 7,212 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Spatial Depth Wise Per Head Convolution
"""
def __init__(self, heads: 'int', d_k: 'int', kernel_size: 'int'=3):
"""
* `heads` is the number of heads
* `d_k` is the number of channels in each head
"""
super().__init__()
self.kernel_size = kernel_size
self.conv = nn.Conv1d(in_channels=d_k * heads, out_channels=d_k *
heads, kernel_size=(kernel_size,), padding=(kernel_size - 1,),
groups=d_k * heads)
def forward(self, x: 'torch.Tensor'):
"""
`x` has shape `[seq_len, batch_size, heads, d_k]`
"""
seq_len, batch_size, heads, d_k = x.shape
x = x.permute(1, 2, 3, 0)
x = x.view(batch_size, heads * d_k, seq_len)
x = self.conv(x)
x = x[:, :, :-(self.kernel_size - 1)]
x = x.view(batch_size, heads, d_k, seq_len)
x = x.permute(3, 0, 1, 2)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
SpatialDepthWiseSharedConvolution | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/jb/cjbf3ssum7resbwampiwoknxcnzh4uzdy4fhoaakjojloew6qlw5.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x_2 => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%view, %primals_2, %primals_3, [1], [2], [1], False, [0], 1), kwargs = {})
triton_poi_fused_convolution_0 = async_compile.triton('triton_poi_fused_convolution_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64, 4], tile_hint=TileHint.SQUARE,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_0(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + (64*x1)), xmask & ymask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x1 + (4*y0)), tmp0, xmask & ymask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/e2/ce2f4ussgg6jfpqca2kweqtdut6siq46cus4r4zd4oeneykjlqi5.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# x_2 => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%view, %primals_2, %primals_3, [1], [2], [1], False, [0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[512],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 384
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tl.store(in_out_ptr0 + (x0), tmp3, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 1, 3), (3, 3, 1))
assert_size_stride(primals_3, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 1, 4), (4, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
stream0 = get_raw_stream(0)
triton_poi_fused_convolution_0.run(primals_1, buf0, 64, 4, grid=grid(64, 4), stream=stream0)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1,), padding=(2,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf1, (64, 1, 6), (6, 6, 1))
del buf0
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf2, primals_3, 384, grid=grid(384), stream=stream0)
del primals_3
return (reinterpret_tensor(buf2, (4, 4, 4, 4), (1, 96, 24, 6), 0), primals_2, reinterpret_tensor(primals_1, (64, 1, 4), (1, 256, 64), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((1, 1, 3), (3, 3, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class SpatialDepthWiseSharedConvolution(Module):
"""
## Spatial Depth Wise Shared Convolution
We share the same kernel across all channels.
"""
def __init__(self, kernel_size: 'int'=3):
"""
"""
super().__init__()
self.kernel_size = kernel_size
self.conv = nn.Conv1d(in_channels=1, out_channels=1, kernel_size=(
kernel_size,), padding=(kernel_size - 1,))
def forward(self, x: 'torch.Tensor'):
"""
`x` has shape `[seq_len, batch_size, heads, d_k]`
"""
seq_len, batch_size, heads, d_k = x.shape
x = x.permute(1, 2, 3, 0)
x = x.view(batch_size * heads * d_k, 1, seq_len)
x = self.conv(x)
x = x[:, :, :-(self.kernel_size - 1)]
x = x.view(batch_size, heads, d_k, seq_len)
x = x.permute(3, 0, 1, 2)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn import Module
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_convolution_0(in_ptr0, out_ptr0, ynumel, xnumel,
YBLOCK: tl.constexpr, XBLOCK: tl.constexpr):
ynumel = 64
xnumel = 4
yoffset = tl.program_id(1) * YBLOCK
yindex = yoffset + tl.arange(0, YBLOCK)[None, :]
ymask = yindex < ynumel
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = xindex < xnumel
x1 = xindex
y0 = yindex
tmp0 = tl.load(in_ptr0 + (y0 + 64 * x1), xmask & ymask, eviction_policy
='evict_last')
tl.store(out_ptr0 + (x1 + 4 * y0), tmp0, xmask & ymask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 384
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tl.store(in_out_ptr0 + x0, tmp3, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (1, 1, 3), (3, 3, 1))
assert_size_stride(primals_3, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 1, 4), (4, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_convolution_0[grid(64, 4)](primals_1, buf0, 64, 4,
XBLOCK=4, YBLOCK=32, num_warps=4, num_stages=1)
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1,),
padding=(2,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf1, (64, 1, 6), (6, 6, 1))
del buf0
buf2 = buf1
del buf1
triton_poi_fused_convolution_1[grid(384)](buf2, primals_3, 384,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_3
return reinterpret_tensor(buf2, (4, 4, 4, 4), (1, 96, 24, 6), 0
), primals_2, reinterpret_tensor(primals_1, (64, 1, 4), (1, 256, 64), 0
)
class SpatialDepthWiseSharedConvolutionNew(Module):
"""
## Spatial Depth Wise Shared Convolution
We share the same kernel across all channels.
"""
def __init__(self, kernel_size: 'int'=3):
"""
"""
super().__init__()
self.kernel_size = kernel_size
self.conv = nn.Conv1d(in_channels=1, out_channels=1, kernel_size=(
kernel_size,), padding=(kernel_size - 1,))
def forward(self, input_0):
primals_2 = self.conv.weight
primals_3 = self.conv.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | SpatialDepthWiseSharedConvolution | false | 7,213 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
from torch import nn
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
## Spatial Depth Wise Shared Convolution
We share the same kernel across all channels.
"""
def __init__(self, kernel_size: 'int'=3):
"""
"""
super().__init__()
self.kernel_size = kernel_size
self.conv = nn.Conv1d(in_channels=1, out_channels=1, kernel_size=(
kernel_size,), padding=(kernel_size - 1,))
def forward(self, x: 'torch.Tensor'):
"""
`x` has shape `[seq_len, batch_size, heads, d_k]`
"""
seq_len, batch_size, heads, d_k = x.shape
x = x.permute(1, 2, 3, 0)
x = x.view(batch_size * heads * d_k, 1, seq_len)
x = self.conv(x)
x = x[:, :, :-(self.kernel_size - 1)]
x = x.view(batch_size, heads, d_k, seq_len)
x = x.permute(3, 0, 1, 2)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
SquaredErrorBayesRisk | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/4m/c4mbktqccghizbaoiq4sp5jwan6vfxof7affhyhr32khwihn5hy7.py
# Topologically Sorted Source Nodes: [alpha, p], Original ATen: [aten.add, aten.div]
# Source node to ATen node mapping:
# alpha => add
# p => div
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%arg0_1, 1.0), kwargs = {})
# %div : [num_users=3] = call_function[target=torch.ops.aten.div.Tensor](args = (%add, %unsqueeze), kwargs = {})
triton_poi_fused_add_div_0 = async_compile.triton('triton_poi_fused_add_div_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp3 = tl.load(in_ptr0 + ((4*x0) + (64*x2)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (1 + (4*x0) + (64*x2)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + (4*x0) + (64*x2)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x0) + (64*x2)), xmask, eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp4 = tmp3 + tmp1
tmp6 = tmp5 + tmp1
tmp7 = tmp4 + tmp6
tmp9 = tmp8 + tmp1
tmp10 = tmp7 + tmp9
tmp12 = tmp11 + tmp1
tmp13 = tmp10 + tmp12
tmp14 = tmp2 / tmp13
tl.store(out_ptr0 + (x3), tmp14, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/kb/ckb3ux77dgnn2btgpdwtpvd62ighfwj3upxaxlwlv65qub5yeef2.py
# Topologically Sorted Source Nodes: [sub, err, sub_1, mul, add_1, var, add_2, loss, mean], Original ATen: [aten.sub, aten.pow, aten.rsub, aten.mul, aten.add, aten.div, aten.sum, aten.mean]
# Source node to ATen node mapping:
# add_1 => add_1
# add_2 => add_2
# err => pow_1
# loss => sum_2
# mean => mean
# mul => mul
# sub => sub
# sub_1 => sub_1
# var => div_1
# Graph fragment:
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%arg1_1, %div), kwargs = {})
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%sub, 2), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %div), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%div, %sub_1), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%unsqueeze_1, 1), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul, %add_1), kwargs = {})
# %add_2 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%pow_1, %div_1), kwargs = {})
# %sum_2 : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%add_2, [-1]), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%sum_2,), kwargs = {})
triton_per_fused_add_div_mean_mul_pow_rsub_sub_sum_1 = async_compile.triton('triton_per_fused_add_div_mean_mul_pow_rsub_sub_sum_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 64],
reduction_hint=ReductionHint.DEFAULT,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32', 5: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {4: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 5), equal_to_1=(4,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_add_div_mean_mul_pow_rsub_sub_sum_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 24, 'num_reduction': 1, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_add_div_mean_mul_pow_rsub_sub_sum_1(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr):
xnumel = 1
rnumel = 64
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
roffset = 0
rmask = tl.full([XBLOCK, RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 4
r2 = (rindex // 16)
tmp0 = tl.load(in_ptr0 + (4*r3), None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4*r3), None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr2 + ((16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr2 + (1 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr2 + (2 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp15 = tl.load(in_ptr2 + (3 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (1 + (4*r3)), None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr1 + (1 + (4*r3)), None, eviction_policy='evict_last')
tmp27 = tl.load(in_ptr2 + (4 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr2 + (5 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp32 = tl.load(in_ptr2 + (6 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp35 = tl.load(in_ptr2 + (7 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp42 = tl.load(in_ptr0 + (2 + (4*r3)), None, eviction_policy='evict_last')
tmp43 = tl.load(in_ptr1 + (2 + (4*r3)), None, eviction_policy='evict_last')
tmp48 = tl.load(in_ptr2 + (8 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp50 = tl.load(in_ptr2 + (9 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp53 = tl.load(in_ptr2 + (10 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp56 = tl.load(in_ptr2 + (11 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp63 = tl.load(in_ptr0 + (3 + (4*r3)), None, eviction_policy='evict_last')
tmp64 = tl.load(in_ptr1 + (3 + (4*r3)), None, eviction_policy='evict_last')
tmp69 = tl.load(in_ptr2 + (12 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp71 = tl.load(in_ptr2 + (13 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp74 = tl.load(in_ptr2 + (14 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp77 = tl.load(in_ptr2 + (15 + (16*r0) + (64*r2)), None, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp3 = tmp2 * tmp2
tmp4 = 1.0
tmp5 = tmp4 - tmp1
tmp6 = tmp1 * tmp5
tmp8 = tmp7 + tmp4
tmp10 = tmp9 + tmp4
tmp11 = tmp8 + tmp10
tmp13 = tmp12 + tmp4
tmp14 = tmp11 + tmp13
tmp16 = tmp15 + tmp4
tmp17 = tmp14 + tmp16
tmp18 = tmp17 + tmp4
tmp19 = tmp6 / tmp18
tmp20 = tmp3 + tmp19
tmp23 = tmp21 - tmp22
tmp24 = tmp23 * tmp23
tmp25 = tmp4 - tmp22
tmp26 = tmp22 * tmp25
tmp28 = tmp27 + tmp4
tmp30 = tmp29 + tmp4
tmp31 = tmp28 + tmp30
tmp33 = tmp32 + tmp4
tmp34 = tmp31 + tmp33
tmp36 = tmp35 + tmp4
tmp37 = tmp34 + tmp36
tmp38 = tmp37 + tmp4
tmp39 = tmp26 / tmp38
tmp40 = tmp24 + tmp39
tmp41 = tmp20 + tmp40
tmp44 = tmp42 - tmp43
tmp45 = tmp44 * tmp44
tmp46 = tmp4 - tmp43
tmp47 = tmp43 * tmp46
tmp49 = tmp48 + tmp4
tmp51 = tmp50 + tmp4
tmp52 = tmp49 + tmp51
tmp54 = tmp53 + tmp4
tmp55 = tmp52 + tmp54
tmp57 = tmp56 + tmp4
tmp58 = tmp55 + tmp57
tmp59 = tmp58 + tmp4
tmp60 = tmp47 / tmp59
tmp61 = tmp45 + tmp60
tmp62 = tmp41 + tmp61
tmp65 = tmp63 - tmp64
tmp66 = tmp65 * tmp65
tmp67 = tmp4 - tmp64
tmp68 = tmp64 * tmp67
tmp70 = tmp69 + tmp4
tmp72 = tmp71 + tmp4
tmp73 = tmp70 + tmp72
tmp75 = tmp74 + tmp4
tmp76 = tmp73 + tmp75
tmp78 = tmp77 + tmp4
tmp79 = tmp76 + tmp78
tmp80 = tmp79 + tmp4
tmp81 = tmp68 / tmp80
tmp82 = tmp66 + tmp81
tmp83 = tmp62 + tmp82
tmp84 = tl.broadcast_to(tmp83, [XBLOCK, RBLOCK])
tmp86 = tl.sum(tmp84, 1)[:, None]
tmp87 = 64.0
tmp88 = tmp86 / tmp87
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([XBLOCK, 1], 0, tl.int32)), tmp88, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [alpha, p], Original ATen: [aten.add, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
buf2 = empty_strided_cuda((), (), torch.float32)
buf3 = buf2; del buf2 # reuse
# Topologically Sorted Source Nodes: [sub, err, sub_1, mul, add_1, var, add_2, loss, mean], Original ATen: [aten.sub, aten.pow, aten.rsub, aten.mul, aten.add, aten.div, aten.sum, aten.mean]
triton_per_fused_add_div_mean_mul_pow_rsub_sub_sum_1.run(buf3, arg1_1, buf0, arg0_1, 1, 64, grid=grid(1), stream=stream0)
del arg0_1
del arg1_1
del buf0
return (buf3, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
arg1_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1, arg1_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class SquaredErrorBayesRisk(Module):
"""
<a id="SquaredErrorBayesRisk"></a>
## Bayes Risk with Squared Error Loss
Here the cost function is squared error,
$$\\sum_{k=1}^K (y_k - p_k)^2 = \\Vert \\mathbf{y} - \\mathbf{p} \\Vert_2^2$$
We integrate this cost over all $\\mathbf{p}$
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\Big[ \\sum_{k=1}^K (y_k - p_k)^2 \\Big]
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K \\mathbb{E} \\Big[ y_k^2 -2 y_k p_k + p_k^2 \\Big] \\
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k^2] \\Big)
\\end{align}
Where $$\\mathbb{E}[p_k] = \\hat{p}_k = rac{ extcolor{orange}{lpha_k}}{S}$$
is the expected probability when sampled from the Dirichlet distribution
and $$\\mathbb{E}[p_k^2] = \\mathbb{E}[p_k]^2 + ext{Var}(p_k)$$
where
$$ ext{Var}(p_k) = rac{ extcolor{orange}{lpha_k}(S - extcolor{orange}{lpha_k})}{S^2 (S + 1)}
= rac{\\hat{p}_k(1 - \\hat{p}_k)}{S + 1}$$
is the variance.
This gives,
egin{align}
\\mathcal{L}(\\Theta)
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k^2] \\Big) \\
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k]^2 + ext{Var}(p_k) \\Big) \\
&= \\sum_{k=1}^K \\Big( ig( y_k -\\mathbb{E}[p_k] ig)^2 + ext{Var}(p_k) \\Big) \\
&= \\sum_{k=1}^K \\Big( ( y_k -\\hat{p}_k)^2 + rac{\\hat{p}_k(1 - \\hat{p}_k)}{S + 1} \\Big)
\\end{align}
This first part of the equation $ig(y_k -\\mathbb{E}[p_k]ig)^2$ is the error term and
the second part is the variance.
"""
def forward(self, evidence: 'torch.Tensor', target: 'torch.Tensor'):
"""
* `evidence` is $\\mathbf{e} \\ge 0$ with shape `[batch_size, n_classes]`
* `target` is $\\mathbf{y}$ with shape `[batch_size, n_classes]`
"""
alpha = evidence + 1.0
strength = alpha.sum(dim=-1)
p = alpha / strength[:, None]
err = (target - p) ** 2
var = p * (1 - p) / (strength[:, None] + 1)
loss = (err + var).sum(dim=-1)
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch.nn import Module
import torch.utils.data
import torch.nn.functional
import torch.autograd
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp3 = tl.load(in_ptr0 + (4 * x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp5 = tl.load(in_ptr0 + (1 + 4 * x0 + 64 * x2), xmask, eviction_policy
='evict_last')
tmp8 = tl.load(in_ptr0 + (2 + 4 * x0 + 64 * x2), xmask, eviction_policy
='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x0 + 64 * x2), xmask,
eviction_policy='evict_last')
tmp1 = 1.0
tmp2 = tmp0 + tmp1
tmp4 = tmp3 + tmp1
tmp6 = tmp5 + tmp1
tmp7 = tmp4 + tmp6
tmp9 = tmp8 + tmp1
tmp10 = tmp7 + tmp9
tmp12 = tmp11 + tmp1
tmp13 = tmp10 + tmp12
tmp14 = tmp2 / tmp13
tl.store(out_ptr0 + x3, tmp14, xmask)
@triton.jit
def triton_per_fused_add_div_mean_mul_pow_rsub_sub_sum_1(in_out_ptr0,
in_ptr0, in_ptr1, in_ptr2, xnumel, rnumel, XBLOCK: tl.constexpr):
RBLOCK: tl.constexpr = 64
xoffset = tl.program_id(0) * XBLOCK
xoffset + tl.arange(0, XBLOCK)[:, None]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[None, :]
tl.full([XBLOCK, RBLOCK], True, tl.int1)
r3 = rindex
r0 = rindex % 4
r2 = rindex // 16
tmp0 = tl.load(in_ptr0 + 4 * r3, None, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + 4 * r3, None, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr2 + (16 * r0 + 64 * r2), None, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr2 + (1 + 16 * r0 + 64 * r2), None, eviction_policy
='evict_last')
tmp12 = tl.load(in_ptr2 + (2 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp15 = tl.load(in_ptr2 + (3 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp21 = tl.load(in_ptr0 + (1 + 4 * r3), None, eviction_policy='evict_last')
tmp22 = tl.load(in_ptr1 + (1 + 4 * r3), None, eviction_policy='evict_last')
tmp27 = tl.load(in_ptr2 + (4 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp29 = tl.load(in_ptr2 + (5 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp32 = tl.load(in_ptr2 + (6 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp35 = tl.load(in_ptr2 + (7 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp42 = tl.load(in_ptr0 + (2 + 4 * r3), None, eviction_policy='evict_last')
tmp43 = tl.load(in_ptr1 + (2 + 4 * r3), None, eviction_policy='evict_last')
tmp48 = tl.load(in_ptr2 + (8 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp50 = tl.load(in_ptr2 + (9 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp53 = tl.load(in_ptr2 + (10 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp56 = tl.load(in_ptr2 + (11 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp63 = tl.load(in_ptr0 + (3 + 4 * r3), None, eviction_policy='evict_last')
tmp64 = tl.load(in_ptr1 + (3 + 4 * r3), None, eviction_policy='evict_last')
tmp69 = tl.load(in_ptr2 + (12 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp71 = tl.load(in_ptr2 + (13 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp74 = tl.load(in_ptr2 + (14 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp77 = tl.load(in_ptr2 + (15 + 16 * r0 + 64 * r2), None,
eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp3 = tmp2 * tmp2
tmp4 = 1.0
tmp5 = tmp4 - tmp1
tmp6 = tmp1 * tmp5
tmp8 = tmp7 + tmp4
tmp10 = tmp9 + tmp4
tmp11 = tmp8 + tmp10
tmp13 = tmp12 + tmp4
tmp14 = tmp11 + tmp13
tmp16 = tmp15 + tmp4
tmp17 = tmp14 + tmp16
tmp18 = tmp17 + tmp4
tmp19 = tmp6 / tmp18
tmp20 = tmp3 + tmp19
tmp23 = tmp21 - tmp22
tmp24 = tmp23 * tmp23
tmp25 = tmp4 - tmp22
tmp26 = tmp22 * tmp25
tmp28 = tmp27 + tmp4
tmp30 = tmp29 + tmp4
tmp31 = tmp28 + tmp30
tmp33 = tmp32 + tmp4
tmp34 = tmp31 + tmp33
tmp36 = tmp35 + tmp4
tmp37 = tmp34 + tmp36
tmp38 = tmp37 + tmp4
tmp39 = tmp26 / tmp38
tmp40 = tmp24 + tmp39
tmp41 = tmp20 + tmp40
tmp44 = tmp42 - tmp43
tmp45 = tmp44 * tmp44
tmp46 = tmp4 - tmp43
tmp47 = tmp43 * tmp46
tmp49 = tmp48 + tmp4
tmp51 = tmp50 + tmp4
tmp52 = tmp49 + tmp51
tmp54 = tmp53 + tmp4
tmp55 = tmp52 + tmp54
tmp57 = tmp56 + tmp4
tmp58 = tmp55 + tmp57
tmp59 = tmp58 + tmp4
tmp60 = tmp47 / tmp59
tmp61 = tmp45 + tmp60
tmp62 = tmp41 + tmp61
tmp65 = tmp63 - tmp64
tmp66 = tmp65 * tmp65
tmp67 = tmp4 - tmp64
tmp68 = tmp64 * tmp67
tmp70 = tmp69 + tmp4
tmp72 = tmp71 + tmp4
tmp73 = tmp70 + tmp72
tmp75 = tmp74 + tmp4
tmp76 = tmp73 + tmp75
tmp78 = tmp77 + tmp4
tmp79 = tmp76 + tmp78
tmp80 = tmp79 + tmp4
tmp81 = tmp68 / tmp80
tmp82 = tmp66 + tmp81
tmp83 = tmp62 + tmp82
tmp84 = tl.broadcast_to(tmp83, [XBLOCK, RBLOCK])
tmp86 = tl.sum(tmp84, 1)[:, None]
tmp87 = 64.0
tmp88 = tmp86 / tmp87
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([XBLOCK, 1], 0, tl.int32), tmp88, None)
def call(args):
arg0_1, arg1_1 = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(arg1_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_0[grid(256)](arg0_1, buf0, 256, XBLOCK=256,
num_warps=4, num_stages=1)
buf2 = empty_strided_cuda((), (), torch.float32)
buf3 = buf2
del buf2
triton_per_fused_add_div_mean_mul_pow_rsub_sub_sum_1[grid(1)](buf3,
arg1_1, buf0, arg0_1, 1, 64, XBLOCK=1, num_warps=2, num_stages=1)
del arg0_1
del arg1_1
del buf0
return buf3,
class SquaredErrorBayesRiskNew(Module):
"""
<a id="SquaredErrorBayesRisk"></a>
## Bayes Risk with Squared Error Loss
Here the cost function is squared error,
$$\\sum_{k=1}^K (y_k - p_k)^2 = \\Vert \\mathbf{y} - \\mathbf{p} \\Vert_2^2$$
We integrate this cost over all $\\mathbf{p}$
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\Big[ \\sum_{k=1}^K (y_k - p_k)^2 \\Big]
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K \\mathbb{E} \\Big[ y_k^2 -2 y_k p_k + p_k^2 \\Big] \\
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k^2] \\Big)
\\end{align}
Where $$\\mathbb{E}[p_k] = \\hat{p}_k = rac{ extcolor{orange}{lpha_k}}{S}$$
is the expected probability when sampled from the Dirichlet distribution
and $$\\mathbb{E}[p_k^2] = \\mathbb{E}[p_k]^2 + ext{Var}(p_k)$$
where
$$ ext{Var}(p_k) = rac{ extcolor{orange}{lpha_k}(S - extcolor{orange}{lpha_k})}{S^2 (S + 1)}
= rac{\\hat{p}_k(1 - \\hat{p}_k)}{S + 1}$$
is the variance.
This gives,
egin{align}
\\mathcal{L}(\\Theta)
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k^2] \\Big) \\
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k]^2 + ext{Var}(p_k) \\Big) \\
&= \\sum_{k=1}^K \\Big( ig( y_k -\\mathbb{E}[p_k] ig)^2 + ext{Var}(p_k) \\Big) \\
&= \\sum_{k=1}^K \\Big( ( y_k -\\hat{p}_k)^2 + rac{\\hat{p}_k(1 - \\hat{p}_k)}{S + 1} \\Big)
\\end{align}
This first part of the equation $ig(y_k -\\mathbb{E}[p_k]ig)^2$ is the error term and
the second part is the variance.
"""
def forward(self, input_0, input_1):
arg0_1 = input_0
arg1_1 = input_1
output = call([arg0_1, arg1_1])
return output[0]
| mcx/annotated_deep_learning_paper_implementations | SquaredErrorBayesRisk | false | 7,214 | [
"MIT"
] | 1 | f169f3a71dd2d36eb28ad31062d3475efa367b88 | https://github.com/mcx/annotated_deep_learning_paper_implementations/tree/f169f3a71dd2d36eb28ad31062d3475efa367b88 | from torch.nn import Module
import torch
import torch.utils.data
import torch.nn.functional
import torch.autograd
class Model(Module):
"""
<a id="SquaredErrorBayesRisk"></a>
## Bayes Risk with Squared Error Loss
Here the cost function is squared error,
$$\\sum_{k=1}^K (y_k - p_k)^2 = \\Vert \\mathbf{y} - \\mathbf{p} \\Vert_2^2$$
We integrate this cost over all $\\mathbf{p}$
egin{align}
\\mathcal{L}(\\Theta)
&= -\\log \\Bigg(
\\int
\\Big[ \\sum_{k=1}^K (y_k - p_k)^2 \\Big]
rac{1}{B( extcolor{orange}{\\mathbf{lpha}})}
\\prod_{k=1}^K p_k^{ extcolor{orange}{lpha_k} - 1}
d\\mathbf{p}
\\Bigg ) \\
&= \\sum_{k=1}^K \\mathbb{E} \\Big[ y_k^2 -2 y_k p_k + p_k^2 \\Big] \\
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k^2] \\Big)
\\end{align}
Where $$\\mathbb{E}[p_k] = \\hat{p}_k = rac{ extcolor{orange}{lpha_k}}{S}$$
is the expected probability when sampled from the Dirichlet distribution
and $$\\mathbb{E}[p_k^2] = \\mathbb{E}[p_k]^2 + ext{Var}(p_k)$$
where
$$ ext{Var}(p_k) = rac{ extcolor{orange}{lpha_k}(S - extcolor{orange}{lpha_k})}{S^2 (S + 1)}
= rac{\\hat{p}_k(1 - \\hat{p}_k)}{S + 1}$$
is the variance.
This gives,
egin{align}
\\mathcal{L}(\\Theta)
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k^2] \\Big) \\
&= \\sum_{k=1}^K \\Big( y_k^2 -2 y_k \\mathbb{E}[p_k] + \\mathbb{E}[p_k]^2 + ext{Var}(p_k) \\Big) \\
&= \\sum_{k=1}^K \\Big( ig( y_k -\\mathbb{E}[p_k] ig)^2 + ext{Var}(p_k) \\Big) \\
&= \\sum_{k=1}^K \\Big( ( y_k -\\hat{p}_k)^2 + rac{\\hat{p}_k(1 - \\hat{p}_k)}{S + 1} \\Big)
\\end{align}
This first part of the equation $ig(y_k -\\mathbb{E}[p_k]ig)^2$ is the error term and
the second part is the variance.
"""
def forward(self, evidence: 'torch.Tensor', target: 'torch.Tensor'):
"""
* `evidence` is $\\mathbf{e} \\ge 0$ with shape `[batch_size, n_classes]`
* `target` is $\\mathbf{y}$ with shape `[batch_size, n_classes]`
"""
alpha = evidence + 1.0
strength = alpha.sum(dim=-1)
p = alpha / strength[:, None]
err = (target - p) ** 2
var = p * (1 - p) / (strength[:, None] + 1)
loss = (err + var).sum(dim=-1)
return loss.mean()
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
MNIST_Discriminator | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/r3/cr355avoqx5hv3qr3i6kjj3otoyequr2sr33d7jxrywusfgam6sb.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x => gt, mul, where
# Graph fragment:
# %gt : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_1, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_1, 0.2), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %view_1, %mul), kwargs = {})
triton_poi_fused_leaky_relu_0 = async_compile.triton('triton_poi_fused_leaky_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/7z/c7zsuucunqdovb2xa6tywxjxwmolzjzdk72ratro7fi3qvgyqb7c.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.sigmoid]
# Source node to ATen node mapping:
# x_1 => sigmoid
# Graph fragment:
# %sigmoid : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%view_3,), kwargs = {})
triton_poi_fused_sigmoid_1 = async_compile.triton('triton_poi_fused_sigmoid_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sigmoid_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sigmoid_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (2, 4), (4, 1))
assert_size_stride(primals_2, (2, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (1, 2), (2, 1))
assert_size_stride(primals_5, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 2), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 2), (32, 8, 2, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 2), (32, 8, 2, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.leaky_relu]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_0.run(buf0, primals_2, buf1, buf2, 128, grid=grid(128), stream=stream0)
del buf0
del primals_2
buf3 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf2, (64, 2), (2, 1), 0), reinterpret_tensor(primals_4, (2, 1), (1, 2), 0), out=buf3)
buf4 = reinterpret_tensor(buf3, (4, 4, 4, 1), (16, 4, 1, 1), 0); del buf3 # reuse
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.sigmoid]
triton_poi_fused_sigmoid_1.run(buf4, primals_5, 64, grid=grid(64), stream=stream0)
del primals_5
return (buf4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, reinterpret_tensor(buf2, (64, 2), (2, 1), 0), buf4, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((2, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((1, 2), (2, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
from torch.nn import functional as F
class MNIST_Discriminator(nn.Module):
def __init__(self, latent_size):
super(MNIST_Discriminator, self).__init__()
self.latent_size = latent_size
self.linear1 = nn.Linear(self.latent_size, self.latent_size // 2)
self.linear2 = nn.Linear(self.latent_size // 2, 1)
def forward(self, x):
x = F.leaky_relu(self.linear1(x), 0.2)
x = torch.sigmoid(self.linear2(x))
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'latent_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
@triton.jit
def triton_poi_fused_sigmoid_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + x0, tmp4, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (2, 4), (4, 1))
assert_size_stride(primals_2, (2,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (1, 2), (2, 1))
assert_size_stride(primals_5, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 2), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 2), (32, 8, 2, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 2), (32, 8, 2, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_leaky_relu_0[grid(128)](buf0, primals_2, buf1,
buf2, 128, XBLOCK=128, num_warps=4, num_stages=1)
del buf0
del primals_2
buf3 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf2, (64, 2), (2, 1), 0),
reinterpret_tensor(primals_4, (2, 1), (1, 2), 0), out=buf3)
buf4 = reinterpret_tensor(buf3, (4, 4, 4, 1), (16, 4, 1, 1), 0)
del buf3
triton_poi_fused_sigmoid_1[grid(64)](buf4, primals_5, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_5
return buf4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf1, reinterpret_tensor(buf2, (64, 2), (2, 1), 0), buf4, primals_4
class MNIST_DiscriminatorNew(nn.Module):
def __init__(self, latent_size):
super(MNIST_DiscriminatorNew, self).__init__()
self.latent_size = latent_size
self.linear1 = nn.Linear(self.latent_size, self.latent_size // 2)
self.linear2 = nn.Linear(self.latent_size // 2, 1)
def forward(self, input_0):
primals_1 = self.linear1.weight
primals_2 = self.linear1.bias
primals_4 = self.linear2.weight
primals_5 = self.linear2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| mdiephuis/adversarial-autoencoders | MNIST_Discriminator | false | 7,215 | [
"MIT"
] | 1 | a722239564362796774de21a64fd92e81dce4089 | https://github.com/mdiephuis/adversarial-autoencoders/tree/a722239564362796774de21a64fd92e81dce4089 | import torch
import torch.nn as nn
from torch.nn import functional as F
class Model(nn.Module):
def __init__(self, latent_size):
super().__init__()
self.latent_size = latent_size
self.linear1 = nn.Linear(self.latent_size, self.latent_size // 2)
self.linear2 = nn.Linear(self.latent_size // 2, 1)
def forward(self, x):
x = F.leaky_relu(self.linear1(x), 0.2)
x = torch.sigmoid(self.linear2(x))
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
MNIST_Encoder | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/o4/co4zyxcl2qkqwco2hpqjmusaq55kf7f6hyk5tkf5vs6buom4p47w.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x => gt, mul, where
# Graph fragment:
# %gt : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_1, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_1, 0.2), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %view_1, %mul), kwargs = {})
triton_poi_fused_leaky_relu_0 = async_compile.triton('triton_poi_fused_leaky_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/hj/chjzotk5iydxvuetxetlv36s7car7cdb24whkuqihxwcy5kkr4o2.py
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.tanh]
# Source node to ATen node mapping:
# x_1 => tanh
# Graph fragment:
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%view_3,), kwargs = {})
triton_poi_fused_tanh_1 = async_compile.triton('triton_poi_fused_tanh_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_tanh_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_tanh_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = libdevice.tanh(tmp2)
tl.store(in_out_ptr0 + (x2), tmp3, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.leaky_relu]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_0.run(buf0, primals_2, buf1, buf2, 256, grid=grid(256), stream=stream0)
del primals_2
buf3 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf2, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf3)
buf4 = reinterpret_tensor(buf3, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf3 # reuse
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.tanh]
triton_poi_fused_tanh_1.run(buf4, primals_5, 256, grid=grid(256), stream=stream0)
del primals_5
return (buf4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, reinterpret_tensor(buf2, (64, 4), (4, 1), 0), buf4, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
from torch.nn import functional as F
class MNIST_Encoder(nn.Module):
def __init__(self, in_channels, latent_size):
super(MNIST_Encoder, self).__init__()
self.in_channels = in_channels
self.latent_size = latent_size
self.linear1 = nn.Linear(self.in_channels, self.latent_size)
self.linear2 = nn.Linear(self.latent_size, self.latent_size)
def forward(self, x):
x = F.leaky_relu(self.linear1(x), 0.2)
x = torch.tanh(self.linear2(x))
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'latent_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
@triton.jit
def triton_poi_fused_tanh_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = libdevice.tanh(tmp2)
tl.store(in_out_ptr0 + x2, tmp3, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_leaky_relu_0[grid(256)](buf0, primals_2, buf1,
buf2, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_2
buf3 = buf0
del buf0
extern_kernels.mm(reinterpret_tensor(buf2, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf3)
buf4 = reinterpret_tensor(buf3, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf3
triton_poi_fused_tanh_1[grid(256)](buf4, primals_5, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del primals_5
return buf4, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf1, reinterpret_tensor(buf2, (64, 4), (4, 1), 0), buf4, primals_4
class MNIST_EncoderNew(nn.Module):
def __init__(self, in_channels, latent_size):
super(MNIST_EncoderNew, self).__init__()
self.in_channels = in_channels
self.latent_size = latent_size
self.linear1 = nn.Linear(self.in_channels, self.latent_size)
self.linear2 = nn.Linear(self.latent_size, self.latent_size)
def forward(self, input_0):
primals_1 = self.linear1.weight
primals_2 = self.linear1.bias
primals_4 = self.linear2.weight
primals_5 = self.linear2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| mdiephuis/adversarial-autoencoders | MNIST_Encoder | false | 7,216 | [
"MIT"
] | 1 | a722239564362796774de21a64fd92e81dce4089 | https://github.com/mdiephuis/adversarial-autoencoders/tree/a722239564362796774de21a64fd92e81dce4089 | import torch
import torch.nn as nn
from torch.nn import functional as F
class Model(nn.Module):
def __init__(self, in_channels, latent_size):
super().__init__()
self.in_channels = in_channels
self.latent_size = latent_size
self.linear1 = nn.Linear(self.in_channels, self.latent_size)
self.linear2 = nn.Linear(self.latent_size, self.latent_size)
def forward(self, x):
x = F.leaky_relu(self.linear1(x), 0.2)
x = torch.tanh(self.linear2(x))
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
MNIST_Generator | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/o4/co4zyxcl2qkqwco2hpqjmusaq55kf7f6hyk5tkf5vs6buom4p47w.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x => gt, mul, where
# Graph fragment:
# %gt : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_1, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_1, 0.2), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %view_1, %mul), kwargs = {})
triton_poi_fused_leaky_relu_0 = async_compile.triton('triton_poi_fused_leaky_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr1 + (x2), tmp7, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.leaky_relu]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_0.run(buf0, primals_2, buf1, buf2, 256, grid=grid(256), stream=stream0)
del primals_2
buf3 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf2, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_0.run(buf3, primals_5, buf4, buf5, 256, grid=grid(256), stream=stream0)
del buf3
del primals_5
return (buf5, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, reinterpret_tensor(buf2, (64, 4), (4, 1), 0), buf4, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
from torch.nn import functional as F
class MNIST_Generator(nn.Module):
def __init__(self, out_channels, latent_size):
super(MNIST_Generator, self).__init__()
self.out_channels = out_channels
self.latent_size = latent_size
self.linear1 = nn.Linear(self.latent_size, self.out_channels)
self.linear2 = nn.Linear(self.out_channels, self.out_channels)
def forward(self, x):
x = F.leaky_relu(self.linear1(x), 0.2)
x = F.leaky_relu(self.linear2(x), 0.2)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'out_channels': 4, 'latent_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr1 + x2, tmp7, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_leaky_relu_0[grid(256)](buf0, primals_2, buf1,
buf2, 256, XBLOCK=256, num_warps=4, num_stages=1)
del primals_2
buf3 = buf0
del buf0
extern_kernels.mm(reinterpret_tensor(buf2, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_leaky_relu_0[grid(256)](buf3, primals_5, buf4,
buf5, 256, XBLOCK=256, num_warps=4, num_stages=1)
del buf3
del primals_5
return buf5, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf1, reinterpret_tensor(buf2, (64, 4), (4, 1), 0), buf4, primals_4
class MNIST_GeneratorNew(nn.Module):
def __init__(self, out_channels, latent_size):
super(MNIST_GeneratorNew, self).__init__()
self.out_channels = out_channels
self.latent_size = latent_size
self.linear1 = nn.Linear(self.latent_size, self.out_channels)
self.linear2 = nn.Linear(self.out_channels, self.out_channels)
def forward(self, input_0):
primals_1 = self.linear1.weight
primals_2 = self.linear1.bias
primals_4 = self.linear2.weight
primals_5 = self.linear2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| mdiephuis/adversarial-autoencoders | MNIST_Generator | false | 7,217 | [
"MIT"
] | 1 | a722239564362796774de21a64fd92e81dce4089 | https://github.com/mdiephuis/adversarial-autoencoders/tree/a722239564362796774de21a64fd92e81dce4089 | import torch
import torch.nn as nn
from torch.nn import functional as F
class Model(nn.Module):
def __init__(self, out_channels, latent_size):
super().__init__()
self.out_channels = out_channels
self.latent_size = latent_size
self.linear1 = nn.Linear(self.latent_size, self.out_channels)
self.linear2 = nn.Linear(self.out_channels, self.out_channels)
def forward(self, x):
x = F.leaky_relu(self.linear1(x), 0.2)
x = F.leaky_relu(self.linear2(x), 0.2)
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
Discriminator | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/54/c54qznm5wnmv6tjkruzet2cvzzwnoisovduku6yewqmcgprk5hli.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.leaky_relu]
# Source node to ATen node mapping:
# x => gt, mul, where
# Graph fragment:
# %gt : [num_users=2] = call_function[target=torch.ops.aten.gt.Scalar](args = (%view_1, 0), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%view_1, 0.2), kwargs = {})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%gt, %view_1, %mul), kwargs = {})
triton_poi_fused_leaky_relu_0 = async_compile.triton('triton_poi_fused_leaky_relu_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[8192],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_leaky_relu_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 8192
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 128
tmp0 = tl.load(in_ptr0 + (x2), None)
tmp1 = tl.load(in_ptr1 + (x0), None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + (x2), tmp4, None)
tl.store(out_ptr1 + (x2), tmp7, None)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/7z/c7zsuucunqdovb2xa6tywxjxwmolzjzdk72ratro7fi3qvgyqb7c.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.sigmoid]
# Source node to ATen node mapping:
# x_2 => sigmoid
# Graph fragment:
# %sigmoid : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%view_5,), kwargs = {})
triton_poi_fused_sigmoid_1 = async_compile.triton('triton_poi_fused_sigmoid_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_sigmoid_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_sigmoid_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + (x0), xmask)
tmp1 = tl.load(in_ptr0 + (0))
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (128, 4), (4, 1))
assert_size_stride(primals_2, (128, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (128, 128), (128, 1))
assert_size_stride(primals_5, (128, ), (1, ))
assert_size_stride(primals_6, (1, 128), (128, 1))
assert_size_stride(primals_7, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 128), (128, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 128), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 128), (2048, 512, 128, 1), torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 128), (2048, 512, 128, 1), torch.float32)
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.leaky_relu]
stream0 = get_raw_stream(0)
triton_poi_fused_leaky_relu_0.run(buf0, primals_2, buf1, buf2, 8192, grid=grid(8192), stream=stream0)
del primals_2
buf3 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf2, (64, 128), (128, 1), 0), reinterpret_tensor(primals_4, (128, 128), (1, 128), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4, 128), (2048, 512, 128, 1), torch.bool)
buf5 = empty_strided_cuda((4, 4, 4, 128), (2048, 512, 128, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_1], Original ATen: [aten.leaky_relu]
triton_poi_fused_leaky_relu_0.run(buf3, primals_5, buf4, buf5, 8192, grid=grid(8192), stream=stream0)
del buf3
del primals_5
buf6 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf5, (64, 128), (128, 1), 0), reinterpret_tensor(primals_6, (128, 1), (1, 128), 0), out=buf6)
buf7 = reinterpret_tensor(buf6, (4, 4, 4, 1), (16, 4, 1, 1), 0); del buf6 # reuse
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.sigmoid]
triton_poi_fused_sigmoid_1.run(buf7, primals_7, 64, grid=grid(64), stream=stream0)
del primals_7
return (buf7, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, reinterpret_tensor(buf2, (64, 128), (128, 1), 0), buf4, reinterpret_tensor(buf5, (64, 128), (128, 1), 0), buf7, primals_6, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((128, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((128, 128), (128, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((128, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((1, 128), (128, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
from torch.nn import functional as F
class Discriminator(nn.Module):
def __init__(self, latent_size, d=128):
super(Discriminator, self).__init__()
self.latent_size = latent_size
self.d = d
self.linear1 = nn.Linear(self.latent_size, self.d)
self.linear2 = nn.Linear(self.d, self.d)
self.linear3 = nn.Linear(self.d, 1)
def forward(self, x):
x = F.leaky_relu(self.linear1(x), 0.2)
x = F.leaky_relu(self.linear2(x), 0.2)
x = torch.sigmoid(self.linear3(x))
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'latent_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_leaky_relu_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
tl.full([XBLOCK], True, tl.int1)
x2 = xindex
x0 = xindex % 128
tmp0 = tl.load(in_ptr0 + x2, None)
tmp1 = tl.load(in_ptr1 + x0, None, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 0.0
tmp4 = tmp2 > tmp3
tmp5 = 0.2
tmp6 = tmp2 * tmp5
tmp7 = tl.where(tmp4, tmp2, tmp6)
tl.store(out_ptr0 + x2, tmp4, None)
tl.store(out_ptr1 + x2, tmp7, None)
@triton.jit
def triton_poi_fused_sigmoid_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_out_ptr0 + x0, xmask)
tmp1 = tl.load(in_ptr0 + 0)
tmp2 = tl.broadcast_to(tmp1, [XBLOCK])
tmp3 = tmp0 + tmp2
tmp4 = tl.sigmoid(tmp3)
tl.store(in_out_ptr0 + x0, tmp4, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (128, 4), (4, 1))
assert_size_stride(primals_2, (128,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (128, 128), (128, 1))
assert_size_stride(primals_5, (128,), (1,))
assert_size_stride(primals_6, (1, 128), (128, 1))
assert_size_stride(primals_7, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 128), (128, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 128), (1, 4), 0), out=buf0)
del primals_1
buf1 = empty_strided_cuda((4, 4, 4, 128), (2048, 512, 128, 1),
torch.bool)
buf2 = empty_strided_cuda((4, 4, 4, 128), (2048, 512, 128, 1),
torch.float32)
get_raw_stream(0)
triton_poi_fused_leaky_relu_0[grid(8192)](buf0, primals_2, buf1,
buf2, 8192, XBLOCK=128, num_warps=4, num_stages=1)
del primals_2
buf3 = buf0
del buf0
extern_kernels.mm(reinterpret_tensor(buf2, (64, 128), (128, 1), 0),
reinterpret_tensor(primals_4, (128, 128), (1, 128), 0), out=buf3)
buf4 = empty_strided_cuda((4, 4, 4, 128), (2048, 512, 128, 1),
torch.bool)
buf5 = empty_strided_cuda((4, 4, 4, 128), (2048, 512, 128, 1),
torch.float32)
triton_poi_fused_leaky_relu_0[grid(8192)](buf3, primals_5, buf4,
buf5, 8192, XBLOCK=128, num_warps=4, num_stages=1)
del buf3
del primals_5
buf6 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf5, (64, 128), (128, 1), 0),
reinterpret_tensor(primals_6, (128, 1), (1, 128), 0), out=buf6)
buf7 = reinterpret_tensor(buf6, (4, 4, 4, 1), (16, 4, 1, 1), 0)
del buf6
triton_poi_fused_sigmoid_1[grid(64)](buf7, primals_7, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del primals_7
return buf7, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0
), buf1, reinterpret_tensor(buf2, (64, 128), (128, 1), 0
), buf4, reinterpret_tensor(buf5, (64, 128), (128, 1), 0
), buf7, primals_6, primals_4
class DiscriminatorNew(nn.Module):
def __init__(self, latent_size, d=128):
super(DiscriminatorNew, self).__init__()
self.latent_size = latent_size
self.d = d
self.linear1 = nn.Linear(self.latent_size, self.d)
self.linear2 = nn.Linear(self.d, self.d)
self.linear3 = nn.Linear(self.d, 1)
def forward(self, input_0):
primals_1 = self.linear1.weight
primals_2 = self.linear1.bias
primals_4 = self.linear2.weight
primals_5 = self.linear2.bias
primals_6 = self.linear3.weight
primals_7 = self.linear3.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0]
| mdiephuis/adversarial-autoencoders | Discriminator | false | 7,218 | [
"MIT"
] | 1 | a722239564362796774de21a64fd92e81dce4089 | https://github.com/mdiephuis/adversarial-autoencoders/tree/a722239564362796774de21a64fd92e81dce4089 | import torch
import torch.nn as nn
from torch.nn import functional as F
class Model(nn.Module):
def __init__(self, latent_size, d=128):
super().__init__()
self.latent_size = latent_size
self.d = d
self.linear1 = nn.Linear(self.latent_size, self.d)
self.linear2 = nn.Linear(self.d, self.d)
self.linear3 = nn.Linear(self.d, 1)
def forward(self, x):
x = F.leaky_relu(self.linear1(x), 0.2)
x = F.leaky_relu(self.linear2(x), 0.2)
x = torch.sigmoid(self.linear3(x))
return x
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
MemoryEfficientPFLU | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ls/clsrn64j7q7762c2hcdfpkojqxyovclgjgcdjmwor2g5tyqdiafb.py
# Topologically Sorted Source Nodes: [mul, add, sqrt, truediv, add_1, mul_1, truediv_1], Original ATen: [aten.mul, aten.add, aten.sqrt, aten.div]
# Source node to ATen node mapping:
# add => add
# add_1 => add_1
# mul => mul
# mul_1 => mul_1
# sqrt => sqrt
# truediv => div
# truediv_1 => div_1
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, %arg0_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, 1), kwargs = {})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%add,), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg0_1, %sqrt), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%div, 1), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, %add_1), kwargs = {})
# %div_1 : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%mul_1, 2), kwargs = {})
triton_poi_fused_add_div_mul_sqrt_0 = async_compile.triton('triton_poi_fused_add_div_mul_sqrt_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mul_sqrt_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mul_sqrt_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tmp0 * tmp0
tmp2 = 1.0
tmp3 = tmp1 + tmp2
tmp4 = libdevice.sqrt(tmp3)
tmp5 = tmp0 / tmp4
tmp6 = tmp5 + tmp2
tmp7 = tmp0 * tmp6
tmp8 = 0.5
tmp9 = tmp7 * tmp8
tl.store(out_ptr0 + (x0), tmp9, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, add, sqrt, truediv, add_1, mul_1, truediv_1], Original ATen: [aten.mul, aten.add, aten.sqrt, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_mul_sqrt_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from torch.autograd import Function
import torch
from torch import nn
class PFLUFunction(Function):
@staticmethod
def forward(ctx, x):
ctx.save_for_backward(x)
return x * (1 + x / torch.sqrt(1 + x * x)) / 2
@staticmethod
def backward(ctx, grad_output):
x, = ctx.saved_tensors
grad_x = None
if ctx.needs_input_grad[0]:
t = 1 / (1 + x * x)
grad_x = grad_output * (1 + x * torch.sqrt(t) * (1 + t)) / 2
return grad_x
class MemoryEfficientPFLU(nn.Module):
def forward(self, x):
return PFLUFunction.apply(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
from torch.autograd import Function
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_mul_sqrt_0(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tmp0 * tmp0
tmp2 = 1.0
tmp3 = tmp1 + tmp2
tmp4 = libdevice.sqrt(tmp3)
tmp5 = tmp0 / tmp4
tmp6 = tmp5 + tmp2
tmp7 = tmp0 * tmp6
tmp8 = 0.5
tmp9 = tmp7 * tmp8
tl.store(out_ptr0 + x0, tmp9, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_mul_sqrt_0[grid(256)](arg0_1, buf0, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class PFLUFunction(Function):
@staticmethod
def forward(ctx, x):
ctx.save_for_backward(x)
return x * (1 + x / torch.sqrt(1 + x * x)) / 2
@staticmethod
def backward(ctx, grad_output):
x, = ctx.saved_tensors
grad_x = None
if ctx.needs_input_grad[0]:
t = 1 / (1 + x * x)
grad_x = grad_output * (1 + x * torch.sqrt(t) * (1 + t)) / 2
return grad_x
class MemoryEfficientPFLUNew(nn.Module):
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mengzhu0308/PFLU-FPFLU | MemoryEfficientPFLU | false | 7,219 | [
"Apache-2.0"
] | 1 | 628cd472db2913e555e902bdf35af834f84a284b | https://github.com/mengzhu0308/PFLU-FPFLU/tree/628cd472db2913e555e902bdf35af834f84a284b | from torch.autograd import Function
import torch
from torch import nn
class PFLUFunction(Function):
@staticmethod
def forward(ctx, x):
ctx.save_for_backward(x)
return x * (1 + x / torch.sqrt(1 + x * x)) / 2
@staticmethod
def backward(ctx, grad_output):
x, = ctx.saved_tensors
grad_x = None
if ctx.needs_input_grad[0]:
t = 1 / (1 + x * x)
grad_x = grad_output * (1 + x * torch.sqrt(t) * (1 + t)) / 2
return grad_x
class Model(nn.Module):
def forward(self, x):
return PFLUFunction.apply(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
FPFLU | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/pc/cpcqogmmdcds5f6z3gnt2ohs2k62oalb5pqrhpxdtc3bir44f2jz.py
# Topologically Sorted Source Nodes: [mul, add, truediv, maximum], Original ATen: [aten.mul, aten.add, aten.div, aten.maximum]
# Source node to ATen node mapping:
# add => add
# maximum => maximum
# mul => mul
# truediv => div
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%arg0_1, %arg0_1), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, 1), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg0_1, %add), kwargs = {})
# %maximum : [num_users=1] = call_function[target=torch.ops.aten.maximum.default](args = (%arg0_1, %div), kwargs = {})
triton_poi_fused_add_div_maximum_mul_0 = async_compile.triton('triton_poi_fused_add_div_maximum_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_maximum_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_maximum_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tmp0 * tmp0
tmp2 = 1.0
tmp3 = tmp1 + tmp2
tmp4 = tmp0 / tmp3
tmp5 = triton_helpers.maximum(tmp0, tmp4)
tl.store(out_ptr0 + (x0), tmp5, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, add, truediv, maximum], Original ATen: [aten.mul, aten.add, aten.div, aten.maximum]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_maximum_mul_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
class FPFLU(nn.Module):
def forward(self, x):
return torch.maximum(x, x / (1 + x * x))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_maximum_mul_0(in_ptr0, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tmp0 * tmp0
tmp2 = 1.0
tmp3 = tmp1 + tmp2
tmp4 = tmp0 / tmp3
tmp5 = triton_helpers.maximum(tmp0, tmp4)
tl.store(out_ptr0 + x0, tmp5, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_maximum_mul_0[grid(256)](arg0_1, buf0, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class FPFLUNew(nn.Module):
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mengzhu0308/PFLU-FPFLU | FPFLU | false | 7,220 | [
"Apache-2.0"
] | 1 | 628cd472db2913e555e902bdf35af834f84a284b | https://github.com/mengzhu0308/PFLU-FPFLU/tree/628cd472db2913e555e902bdf35af834f84a284b | import torch
from torch import nn
class Model(nn.Module):
def forward(self, x):
return torch.maximum(x, x / (1 + x * x))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
WQ | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/q2/cq2qrtyf6mtvmdrqgmyex3qru5qkrvvs4wppvbxy3li4uxc3exl4.py
# Topologically Sorted Source Nodes: [std, mul, abs_1, m, mul_1, alpha_w], Original ATen: [aten.std, aten.mul, aten.abs, aten.mean, aten.sub]
# Source node to ATen node mapping:
# abs_1 => abs_1
# alpha_w => sub
# m => mean
# mul => mul
# mul_1 => mul_1
# std => sqrt, var
# Graph fragment:
# %var : [num_users=1] = call_function[target=torch.ops.aten.var.correction](args = (%arg0_1,), kwargs = {correction: 1.0})
# %sqrt : [num_users=1] = call_function[target=torch.ops.aten.sqrt.default](args = (%var,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sqrt, 12.987012987012987), kwargs = {})
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%arg0_1,), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.default](args = (%abs_1,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mean, 13.155844155844155), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%mul, %mul_1), kwargs = {})
triton_per_fused_abs_mean_mul_std_sub_0 = async_compile.triton('triton_per_fused_abs_mean_mul_std_sub_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.persistent_reduction(
size_hints=[1, 256],
reduction_hint=ReductionHint.INNER,
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {2: 1}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=(2,))]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_per_fused_abs_mean_mul_std_sub_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': True, 'num_load': 1, 'num_reduction': 4, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}
)
@triton.jit
def triton_per_fused_abs_mean_mul_std_sub_0(in_out_ptr0, in_ptr0, xnumel, rnumel):
xnumel = 1
XBLOCK: tl.constexpr = 1
rnumel = 256
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = tl.full([1], xoffset, tl.int32)
xmask = tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
roffset = 0
rmask = tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + (r0), None)
tmp1 = tl.broadcast_to(tmp0, [RBLOCK])
tmp3 = tl.broadcast_to(tmp1, [RBLOCK])
tmp5 = triton_helpers.promote_to_tensor(tl.sum(tmp3, 0))
tmp6 = tl.full([1], 256, tl.int32)
tmp7 = tmp6.to(tl.float32)
tmp8 = tmp5 / tmp7
tmp9 = tmp1 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tl.broadcast_to(tmp10, [RBLOCK])
tmp13 = triton_helpers.promote_to_tensor(tl.sum(tmp11, 0))
tmp14 = tl_math.abs(tmp0)
tmp15 = tl.broadcast_to(tmp14, [RBLOCK])
tmp17 = triton_helpers.promote_to_tensor(tl.sum(tmp15, 0))
tmp18 = 255.0
tmp19 = tmp13 / tmp18
tmp20 = libdevice.sqrt(tmp19)
tmp21 = 12.987012987012987
tmp22 = tmp20 * tmp21
tmp23 = 256.0
tmp24 = tmp17 / tmp23
tmp25 = 13.155844155844155
tmp26 = tmp24 * tmp25
tmp27 = tmp22 - tmp26
tl.debug_barrier()
tl.store(in_out_ptr0 + (tl.full([1], 0, tl.int32)), tmp27, None)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf4 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [std, mul, abs_1, m, mul_1, alpha_w], Original ATen: [aten.std, aten.mul, aten.abs, aten.mean, aten.sub]
stream0 = get_raw_stream(0)
triton_per_fused_abs_mean_mul_std_sub_0.run(buf4, arg0_1, 1, 256, grid=grid(1), stream=stream0)
del arg0_1
return (buf4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
def stats_quant(x, nbit, qmode='symm', dequantize=True):
z_typical = {'4bit': [0.077, 1.013], '8bit': [0.027, 1.114]}
z = z_typical[f'{int(nbit)}bit']
m = x.abs().mean()
std = x.std()
if qmode == 'symm':
n_lv = 2 ** (nbit - 1) - 1
alpha_w = 1 / z[0] * std - z[1] / z[0] * m
elif qmode == 'asymm':
n_lv = (2 ** nbit - 1) / 2
alpha_w = 2 * m
else:
raise NotImplementedError
x = x.clamp(-alpha_w.item(), alpha_w.item())
scale = n_lv / alpha_w
xq = x.mul(scale).round()
if len(xq.unique()) > 2 ** nbit:
xq = xq.clamp(-2 ** nbit // 2, 2 ** nbit // 2 - 1)
if dequantize:
xq = xq.div(scale)
return xq, scale
class RoundQ(torch.autograd.Function):
@staticmethod
def forward(ctx, input, wbit, qmode):
input_q, _scale = stats_quant(input, wbit, qmode)
ctx.save_for_backward(input)
return input_q
@staticmethod
def backward(ctx, grad_output):
grad_input = grad_output.clone()
return grad_input, None, None
class WQ(nn.Module):
"""
Weight quantizer
"""
def __init__(self, wbit, qmode='symm'):
super(WQ, self).__init__()
self.wbit = wbit
self.qmode = qmode
def forward(self, x):
weight_q = RoundQ.apply(x, self.wbit, self.qmode)
return weight_q
def extra_repr(self):
return super(WQ, self).extra_repr() + 'qmode={}'.format(self.qmode)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'wbit': 4}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_per_fused_abs_mean_mul_std_sub_0(in_out_ptr0, in_ptr0, xnumel,
rnumel):
XBLOCK: tl.constexpr = 1
RBLOCK: tl.constexpr = 256
xoffset = tl.program_id(0) * XBLOCK
tl.full([1], xoffset, tl.int32)
tl.full([RBLOCK], True, tl.int1)
rindex = tl.arange(0, RBLOCK)[:]
tl.full([RBLOCK], True, tl.int1)
r0 = rindex
tmp0 = tl.load(in_ptr0 + r0, None)
tmp1 = tl.broadcast_to(tmp0, [RBLOCK])
tmp3 = tl.broadcast_to(tmp1, [RBLOCK])
tmp5 = triton_helpers.promote_to_tensor(tl.sum(tmp3, 0))
tmp6 = tl.full([1], 256, tl.int32)
tmp7 = tmp6.to(tl.float32)
tmp8 = tmp5 / tmp7
tmp9 = tmp1 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tl.broadcast_to(tmp10, [RBLOCK])
tmp13 = triton_helpers.promote_to_tensor(tl.sum(tmp11, 0))
tmp14 = tl_math.abs(tmp0)
tmp15 = tl.broadcast_to(tmp14, [RBLOCK])
tmp17 = triton_helpers.promote_to_tensor(tl.sum(tmp15, 0))
tmp18 = 255.0
tmp19 = tmp13 / tmp18
tmp20 = libdevice.sqrt(tmp19)
tmp21 = 12.987012987012987
tmp22 = tmp20 * tmp21
tmp23 = 256.0
tmp24 = tmp17 / tmp23
tmp25 = 13.155844155844155
tmp26 = tmp24 * tmp25
tmp27 = tmp22 - tmp26
tl.debug_barrier()
tl.store(in_out_ptr0 + tl.full([1], 0, tl.int32), tmp27, None)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf1 = empty_strided_cuda((), (), torch.float32)
buf4 = buf1
del buf1
get_raw_stream(0)
triton_per_fused_abs_mean_mul_std_sub_0[grid(1)](buf4, arg0_1, 1,
256, num_warps=2, num_stages=1)
del arg0_1
return buf4,
def stats_quant(x, nbit, qmode='symm', dequantize=True):
z_typical = {'4bit': [0.077, 1.013], '8bit': [0.027, 1.114]}
z = z_typical[f'{int(nbit)}bit']
m = x.abs().mean()
std = x.std()
if qmode == 'symm':
n_lv = 2 ** (nbit - 1) - 1
alpha_w = 1 / z[0] * std - z[1] / z[0] * m
elif qmode == 'asymm':
n_lv = (2 ** nbit - 1) / 2
alpha_w = 2 * m
else:
raise NotImplementedError
x = x.clamp(-alpha_w.item(), alpha_w.item())
scale = n_lv / alpha_w
xq = x.mul(scale).round()
if len(xq.unique()) > 2 ** nbit:
xq = xq.clamp(-2 ** nbit // 2, 2 ** nbit // 2 - 1)
if dequantize:
xq = xq.div(scale)
return xq, scale
class RoundQ(torch.autograd.Function):
@staticmethod
def forward(ctx, input, wbit, qmode):
input_q, _scale = stats_quant(input, wbit, qmode)
ctx.save_for_backward(input)
return input_q
@staticmethod
def backward(ctx, grad_output):
grad_input = grad_output.clone()
return grad_input, None, None
class WQNew(nn.Module):
"""
Weight quantizer
"""
def __init__(self, wbit, qmode='symm'):
super(WQNew, self).__init__()
self.wbit = wbit
self.qmode = qmode
def extra_repr(self):
return super(WQNew, self).extra_repr() + 'qmode={}'.format(self.qmode)
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mengjian0502/TorchInference_SRAM | WQ | false | 7,221 | [
"MIT"
] | 1 | fcc465c73b79f2ab670b6af03aa53f9bb47c64ca | https://github.com/mengjian0502/TorchInference_SRAM/tree/fcc465c73b79f2ab670b6af03aa53f9bb47c64ca | import torch
import torch.nn as nn
def stats_quant(x, nbit, qmode='symm', dequantize=True):
z_typical = {'4bit': [0.077, 1.013], '8bit': [0.027, 1.114]}
z = z_typical[f'{int(nbit)}bit']
m = x.abs().mean()
std = x.std()
if qmode == 'symm':
n_lv = 2 ** (nbit - 1) - 1
alpha_w = 1 / z[0] * std - z[1] / z[0] * m
elif qmode == 'asymm':
n_lv = (2 ** nbit - 1) / 2
alpha_w = 2 * m
else:
raise NotImplementedError
x = x.clamp(-alpha_w.item(), alpha_w.item())
scale = n_lv / alpha_w
xq = x.mul(scale).round()
if len(xq.unique()) > 2 ** nbit:
xq = xq.clamp(-2 ** nbit // 2, 2 ** nbit // 2 - 1)
if dequantize:
xq = xq.div(scale)
return xq, scale
class RoundQ(torch.autograd.Function):
@staticmethod
def forward(ctx, input, wbit, qmode):
input_q, _scale = stats_quant(input, wbit, qmode)
ctx.save_for_backward(input)
return input_q
@staticmethod
def backward(ctx, grad_output):
grad_input = grad_output.clone()
return grad_input, None, None
class Model(nn.Module):
"""
Weight quantizer
"""
def __init__(self, wbit, qmode='symm'):
super().__init__()
self.wbit = wbit
self.qmode = qmode
def forward(self, x):
weight_q = RoundQ.apply(x, self.wbit, self.qmode)
return weight_q
def extra_repr(self):
return super(WQ, self).extra_repr() + 'qmode={}'.format(self.qmode)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
Coxnnet | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ak/caktgvad6egrpdwim6fuv7sc4ceemq43bhsh6px5fk7rhadyseq3.py
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.tanh]
# Source node to ATen node mapping:
# x => tanh
# Graph fragment:
# %tanh : [num_users=2] = call_function[target=torch.ops.aten.tanh.default](args = (%view_1,), kwargs = {})
triton_poi_fused_tanh_0 = async_compile.triton('triton_poi_fused_tanh_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_tanh_0', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_tanh_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = libdevice.tanh(tmp2)
tl.store(in_out_ptr0 + (x2), tmp3, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (2, 4), (4, 1))
assert_size_stride(primals_2, (2, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (1, 2), (2, 1))
assert_size_stride(primals_5, (1, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 2), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 2), (32, 8, 2, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.tanh]
stream0 = get_raw_stream(0)
triton_poi_fused_tanh_0.run(buf1, primals_2, 128, grid=grid(128), stream=stream0)
del primals_2
buf3 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 2), (2, 1), 0), reinterpret_tensor(primals_4, (2, 1), (1, 2), 0), alpha=1, beta=1, out=buf3)
del primals_5
return (reinterpret_tensor(buf3, (4, 4, 4, 1), (16, 4, 1, 1), 0), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, primals_4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((2, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((2, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((1, 2), (2, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((1, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import numpy as np
import torch.nn as nn
class Coxnnet(nn.Module):
def __init__(self, nfeat):
super(Coxnnet, self).__init__()
self.fc1 = nn.Linear(nfeat, int(np.ceil(nfeat ** 0.5)))
self.dropout = nn.Dropout(0.5)
self.fc2 = nn.Linear(int(np.ceil(nfeat ** 0.5)), 1)
self.init_hidden()
def forward(self, x, coo=None):
x = torch.tanh(self.fc1(x))
x = self.dropout(x)
x = self.fc2(x)
return x
def init_hidden(self):
nn.init.xavier_normal_(self.fc1.weight)
nn.init.xavier_normal_(self.fc2.weight)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'nfeat': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import numpy as np
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_tanh_0(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 128
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 2
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = libdevice.tanh(tmp2)
tl.store(in_out_ptr0 + x2, tmp3, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (2, 4), (4, 1))
assert_size_stride(primals_2, (2,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (1, 2), (2, 1))
assert_size_stride(primals_5, (1,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 2), (2, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_1, (4, 2), (1, 4), 0), out=buf0)
del primals_1
buf1 = reinterpret_tensor(buf0, (4, 4, 4, 2), (32, 8, 2, 1), 0)
del buf0
get_raw_stream(0)
triton_poi_fused_tanh_0[grid(128)](buf1, primals_2, 128, XBLOCK=128,
num_warps=4, num_stages=1)
del primals_2
buf3 = empty_strided_cuda((64, 1), (1, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(buf1, (64, 2), (
2, 1), 0), reinterpret_tensor(primals_4, (2, 1), (1, 2), 0),
alpha=1, beta=1, out=buf3)
del primals_5
return reinterpret_tensor(buf3, (4, 4, 4, 1), (16, 4, 1, 1), 0
), reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, primals_4
class CoxnnetNew(nn.Module):
def __init__(self, nfeat):
super(CoxnnetNew, self).__init__()
self.fc1 = nn.Linear(nfeat, int(np.ceil(nfeat ** 0.5)))
self.dropout = nn.Dropout(0.5)
self.fc2 = nn.Linear(int(np.ceil(nfeat ** 0.5)), 1)
self.init_hidden()
def init_hidden(self):
nn.init.xavier_normal_(self.fc1.weight)
nn.init.xavier_normal_(self.fc2.weight)
def forward(self, input_0):
primals_1 = self.fc1.weight
primals_2 = self.fc1.bias
primals_4 = self.fc2.weight
primals_5 = self.fc2.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0]
| menggerSherry/SAVAE-Cox | Coxnnet | false | 7,222 | [
"Apache-2.0"
] | 1 | c087ab4f267da28db7eb497c844bea59e65ed125 | https://github.com/menggerSherry/SAVAE-Cox/tree/c087ab4f267da28db7eb497c844bea59e65ed125 | import torch
import numpy as np
import torch.nn as nn
class Model(nn.Module):
def __init__(self, nfeat):
super().__init__()
self.fc1 = nn.Linear(nfeat, int(np.ceil(nfeat ** 0.5)))
self.dropout = nn.Dropout(0.5)
self.fc2 = nn.Linear(int(np.ceil(nfeat ** 0.5)), 1)
self.init_hidden()
def forward(self, x, coo=None):
x = torch.tanh(self.fc1(x))
x = self.dropout(x)
x = self.fc2(x)
return x
def init_hidden(self):
nn.init.xavier_normal_(self.fc1.weight)
nn.init.xavier_normal_(self.fc2.weight)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
MVNormalNetwork | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/v4/cv4f7wcky6nab2f7hkdgxju2cexi5th7qqdlhkbjt2rokrlz7ncr.py
# Topologically Sorted Source Nodes: [diag_embed], Original ATen: [aten.diag_embed]
# Source node to ATen node mapping:
# diag_embed => full_default, where
# Graph fragment:
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([], 0.0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where : [num_users=1] = call_function[target=torch.ops.aten.where.self](args = (%view_4, %permute_2, %full_default), kwargs = {})
triton_poi_fused_diag_embed_0 = async_compile.triton('triton_poi_fused_diag_embed_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_diag_embed_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_diag_embed_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4) % 4
x2 = (xindex // 16)
x3 = xindex
tmp3 = tl.load(in_ptr0 + (x0 + (4*x2)), xmask, eviction_policy='evict_last')
tmp0 = x0
tmp1 = x1
tmp2 = tmp0 == tmp1
tmp4 = tl_math.exp(tmp3)
tmp5 = 0.0
tmp6 = tl.where(tmp2, tmp4, tmp5)
tl.store(out_ptr0 + (x3), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, ), (1, ))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mean], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [sc], Original ATen: [aten.addmm]
extern_kernels.addmm(primals_5, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), alpha=1, beta=1, out=buf1)
del primals_4
del primals_5
buf2 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [diag_embed], Original ATen: [aten.diag_embed]
stream0 = get_raw_stream(0)
triton_poi_fused_diag_embed_0.run(buf1, buf2, 1024, grid=grid(1024), stream=stream0)
return (reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0), buf2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class MVNormalNetwork(nn.Module):
def __init__(self, latent_dim):
super().__init__()
self.mean = nn.Linear(latent_dim, latent_dim)
self.sc = nn.Linear(latent_dim, latent_dim)
def forward(self, x):
mean = self.mean(x)
sc = self.sc(x)
return mean, torch.diag_embed(torch.exp(sc))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'latent_dim': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_diag_embed_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4 % 4
x2 = xindex // 16
x3 = xindex
tmp3 = tl.load(in_ptr0 + (x0 + 4 * x2), xmask, eviction_policy='evict_last'
)
tmp0 = x0
tmp1 = x1
tmp2 = tmp0 == tmp1
tmp4 = tl_math.exp(tmp3)
tmp5 = 0.0
tmp6 = tl.where(tmp2, tmp4, tmp5)
tl.store(out_ptr0 + x3, tmp6, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4,), (1,))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_2, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_1, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf0)
del primals_1
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.addmm(primals_5, reinterpret_tensor(primals_3, (64,
4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0
), alpha=1, beta=1, out=buf1)
del primals_4
del primals_5
buf2 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1),
torch.float32)
get_raw_stream(0)
triton_poi_fused_diag_embed_0[grid(1024)](buf1, buf2, 1024, XBLOCK=
128, num_warps=4, num_stages=1)
return reinterpret_tensor(buf0, (4, 4, 4, 4), (64, 16, 4, 1), 0
), buf2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1
class MVNormalNetworkNew(nn.Module):
def __init__(self, latent_dim):
super().__init__()
self.mean = nn.Linear(latent_dim, latent_dim)
self.sc = nn.Linear(latent_dim, latent_dim)
def forward(self, input_0):
primals_1 = self.mean.weight
primals_2 = self.mean.bias
primals_4 = self.sc.weight
primals_5 = self.sc.bias
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4, primals_5])
return output[0], output[1]
| mgb45/OC-notebooks | MVNormalNetwork | false | 7,223 | [
"MIT"
] | 1 | 67b1899d1fb3455ab3caab58f94429b9f432164b | https://github.com/mgb45/OC-notebooks/tree/67b1899d1fb3455ab3caab58f94429b9f432164b | import torch
import torch.nn as nn
class Model(nn.Module):
def __init__(self, latent_dim):
super().__init__()
self.mean = nn.Linear(latent_dim, latent_dim)
self.sc = nn.Linear(latent_dim, latent_dim)
def forward(self, x):
mean = self.mean(x)
sc = self.sc(x)
return mean, torch.diag_embed(torch.exp(sc))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4]
|
Conv1d_samePadding | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/5d/c5dikgnvjj47gew67o3wnlvktf4rgbccmu4bxfbwcb7bldgpilyt.py
# Topologically Sorted Source Nodes: [input_1], Original ATen: [aten.constant_pad_nd]
# Source node to ATen node mapping:
# input_1 => constant_pad_nd
# Graph fragment:
# %constant_pad_nd : [num_users=2] = call_function[target=torch.ops.aten.constant_pad_nd.default](args = (%primals_1, [1, 2], 0.0), kwargs = {})
triton_poi_fused_constant_pad_nd_0 = async_compile.triton('triton_poi_fused_constant_pad_nd_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[128],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_constant_pad_nd_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_constant_pad_nd_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 112
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 7
x1 = (xindex // 7)
x2 = xindex
tmp0 = (-1) + x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = tl.load(in_ptr0 + ((-1) + x0 + (4*x1)), tmp5 & xmask, other=0.0)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/au/cau4pihcaptiev5y2ewn2o2nvrwhk7hogc72cofmmtbyv4rxc2oy.py
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# conv1d => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%constant_pad_nd, %primals_2, %primals_3, [1], [0], [1], False, [0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 4) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 7), (28, 7, 1), torch.float32)
# Topologically Sorted Source Nodes: [input_1], Original ATen: [aten.constant_pad_nd]
stream0 = get_raw_stream(0)
triton_poi_fused_constant_pad_nd_0.run(primals_1, buf0, 112, grid=grid(112), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1,), padding=(0,), dilation=(1,), transposed=False, output_padding=(0,), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4), (16, 4, 1))
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [conv1d], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf2, primals_3, 64, grid=grid(64), stream=stream0)
del primals_3
return (buf2, primals_2, buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4), (16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
import torch.nn.functional as F
class Conv1d_samePadding(nn.Conv1d):
def __init__(self, *args, padding: int=0, **kwargs):
assert padding == 0, "no additional padding on top of 'same' padding"
kwargs['padding'] = 0
super().__init__(*args, **kwargs)
def same_padding_1d(self, input):
input_duration = input.size(2)
filter_duration = self.weight.size(2)
out_duration = (input_duration + self.stride[0] - 1) // self.stride[0]
padding_duration = max(0, (out_duration - 1) * self.stride[0] + (
filter_duration - 1) * self.dilation[0] + 1 - input_duration)
duration_odd = padding_duration % 2
input = F.pad(input, (padding_duration // 2, padding_duration // 2 +
int(duration_odd)))
return input
def forward(self, input):
input = self.same_padding_1d(input)
return super().forward(input)
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4, 'kernel_size': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch import nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_constant_pad_nd_0(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 112
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 7
x1 = xindex // 7
x2 = xindex
tmp0 = -1 + x0
tmp1 = tl.full([1], 0, tl.int64)
tmp2 = tmp0 >= tmp1
tmp3 = tl.full([1], 4, tl.int64)
tmp4 = tmp0 < tmp3
tmp5 = tmp2 & tmp4
tmp6 = tl.load(in_ptr0 + (-1 + x0 + 4 * x1), tmp5 & xmask, other=0.0)
tl.store(out_ptr0 + x2, tmp6, xmask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 4 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4), (16, 4, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 7), (28, 7, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_constant_pad_nd_0[grid(112)](primals_1, buf0, 112,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1,),
padding=(0,), dilation=(1,), transposed=False, output_padding=(
0,), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4), (16, 4, 1))
buf2 = buf1
del buf1
triton_poi_fused_convolution_1[grid(64)](buf2, primals_3, 64,
XBLOCK=64, num_warps=1, num_stages=1)
del primals_3
return buf2, primals_2, buf0
class Conv1d_samePaddingNew(nn.Conv1d):
def __init__(self, *args, padding: int=0, **kwargs):
assert padding == 0, "no additional padding on top of 'same' padding"
kwargs['padding'] = 0
super().__init__(*args, **kwargs)
def same_padding_1d(self, input):
input_duration = input.size(2)
filter_duration = self.weight.size(2)
out_duration = (input_duration + self.stride[0] - 1) // self.stride[0]
padding_duration = max(0, (out_duration - 1) * self.stride[0] + (
filter_duration - 1) * self.dilation[0] + 1 - input_duration)
duration_odd = padding_duration % 2
input = F.pad(input, (padding_duration // 2, padding_duration // 2 +
int(duration_odd)))
return input
def forward(self, input_0):
primals_1 = self.weight
primals_3 = self.bias
primals_2 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| mgrachten/crepe-pytorch | Conv1d_samePadding | false | 7,224 | [
"MIT"
] | 1 | 94305a78d2d82e414c251d50b63dc021af277c75 | https://github.com/mgrachten/crepe-pytorch/tree/94305a78d2d82e414c251d50b63dc021af277c75 | import torch
from torch import nn
import torch.nn.functional as F
class Model(nn.Conv1d):
def __init__(self, *args, padding: int=0, **kwargs):
assert padding == 0, "no additional padding on top of 'same' padding"
kwargs['padding'] = 0
super().__init__(*args, **kwargs)
def same_padding_1d(self, input):
input_duration = input.size(2)
filter_duration = self.weight.size(2)
out_duration = (input_duration + self.stride[0] - 1) // self.stride[0]
padding_duration = max(0, (out_duration - 1) * self.stride[0] + (
filter_duration - 1) * self.dilation[0] + 1 - input_duration)
duration_odd = padding_duration % 2
input = F.pad(input, (padding_duration // 2, padding_duration // 2 +
int(duration_odd)))
return input
def forward(self, input):
input = self.same_padding_1d(input)
return super().forward(input)
def get_inputs():
return [torch.rand([4, 4, 4])]
def get_init_inputs():
return [4, 4, 4]
|
NALUCell | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/lz/clz2gei2r4v6zp73e5uwuxfqwprorq7wyulpi7djhsfrhn32mgcm.py
# Topologically Sorted Source Nodes: [tanh, sigmoid, W], Original ATen: [aten.tanh, aten.sigmoid, aten.mul]
# Source node to ATen node mapping:
# W => mul
# sigmoid => sigmoid
# tanh => tanh
# Graph fragment:
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%primals_1,), kwargs = {})
# %sigmoid : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%primals_2,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%tanh, %sigmoid), kwargs = {})
triton_poi_fused_mul_sigmoid_tanh_0 = async_compile.triton('triton_poi_fused_mul_sigmoid_tanh_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_sigmoid_tanh_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_sigmoid_tanh_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp2 = tl.load(in_ptr1 + (x0), xmask)
tmp1 = libdevice.tanh(tmp0)
tmp3 = tl.sigmoid(tmp2)
tmp4 = tmp1 * tmp3
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/fp/cfpz76nln2digsmp6v4tkbkx5al74go27d54wsrtmt6k33dheeoi.py
# Topologically Sorted Source Nodes: [abs_1, add, log_input], Original ATen: [aten.abs, aten.add, aten.log]
# Source node to ATen node mapping:
# abs_1 => abs_1
# add => add
# log_input => log
# Graph fragment:
# %abs_1 : [num_users=1] = call_function[target=torch.ops.aten.abs.default](args = (%primals_3,), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%abs_1, 1e-10), kwargs = {})
# %log : [num_users=1] = call_function[target=torch.ops.aten.log.default](args = (%add,), kwargs = {})
triton_poi_fused_abs_add_log_1 = async_compile.triton('triton_poi_fused_abs_add_log_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_abs_add_log_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_abs_add_log_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = tl_math.abs(tmp0)
tmp2 = 1e-10
tmp3 = tmp1 + tmp2
tmp4 = tl_math.log(tmp3)
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/66/c66ysiitxesxde3gyvusomexzaxhpi4yob7lg235t2wcmjvocoop.py
# Topologically Sorted Source Nodes: [g, add_sub, m, sub, mul_div, y], Original ATen: [aten.sigmoid, aten.mul, aten.exp, aten.rsub, aten.add]
# Source node to ATen node mapping:
# add_sub => mul_1
# g => sigmoid_1
# m => exp
# mul_div => mul_3
# sub => sub
# y => add_1
# Graph fragment:
# %sigmoid_1 : [num_users=2] = call_function[target=torch.ops.aten.sigmoid.default](args = (%view_3,), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sigmoid_1, %view_1), kwargs = {})
# %exp : [num_users=1] = call_function[target=torch.ops.aten.exp.default](args = (%view_5,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (1, %sigmoid_1), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %exp), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %mul_3), kwargs = {})
triton_poi_fused_add_exp_mul_rsub_sigmoid_2 = async_compile.triton('triton_poi_fused_add_exp_mul_rsub_sigmoid_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_exp_mul_rsub_sigmoid_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_exp_mul_rsub_sigmoid_2(in_ptr0, in_ptr1, in_ptr2, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp2 = tl.load(in_ptr1 + (x0), xmask)
tmp6 = tl.load(in_ptr2 + (x0), xmask)
tmp1 = tl.sigmoid(tmp0)
tmp3 = tmp1 * tmp2
tmp4 = 1.0
tmp5 = tmp4 - tmp1
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 * tmp7
tmp9 = tmp3 + tmp8
tl.store(out_ptr0 + (x0), tmp9, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [tanh, sigmoid, W], Original ATen: [aten.tanh, aten.sigmoid, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_sigmoid_tanh_0.run(primals_1, primals_2, buf0, 16, grid=grid(16), stream=stream0)
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [a], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf0, (4, 4), (1, 4), 0), out=buf1)
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_1], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf2)
del primals_4
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [abs_1, add, log_input], Original ATen: [aten.abs, aten.add, aten.log]
triton_poi_fused_abs_add_log_1.run(primals_3, buf3, 256, grid=grid(256), stream=stream0)
buf4 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear_2], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(buf3, (64, 4), (4, 1), 0), reinterpret_tensor(buf0, (4, 4), (1, 4), 0), out=buf4)
del buf0
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [g, add_sub, m, sub, mul_div, y], Original ATen: [aten.sigmoid, aten.mul, aten.exp, aten.rsub, aten.add]
triton_poi_fused_add_exp_mul_rsub_sigmoid_2.run(buf2, buf1, buf4, buf5, 256, grid=grid(256), stream=stream0)
return (buf5, primals_1, primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), buf1, buf2, reinterpret_tensor(buf3, (64, 4), (4, 1), 0), buf4, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn import init
from torch.nn.parameter import Parameter
class NeuralAccumulatorCell(nn.Module):
"""A Neural Accumulator (NAC) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.W_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.M_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.register_parameter('W_hat', self.W_hat)
self.register_parameter('M_hat', self.M_hat)
self.register_parameter('bias', None)
self._reset_params()
def _reset_params(self):
init.kaiming_uniform_(self.W_hat)
init.kaiming_uniform_(self.M_hat)
def forward(self, input):
W = torch.tanh(self.W_hat) * torch.sigmoid(self.M_hat)
return F.linear(input, W, self.bias)
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
class NALUCell(nn.Module):
"""A Neural Arithmetic Logic Unit (NALU) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.eps = 1e-10
self.G = Parameter(torch.Tensor(out_dim, in_dim))
self.nac = NeuralAccumulatorCell(in_dim, out_dim)
self.register_parameter('bias', None)
init.kaiming_uniform_(self.G, a=math.sqrt(5))
def forward(self, input: 'torch.Tensor'):
a = self.nac(input)
g = F.linear(input, self.G, self.bias).sigmoid()
add_sub = g * a
log_input = (input.abs() + self.eps).log()
m = self.nac(log_input).exp()
mul_div = (1 - g) * m
y = add_sub + mul_div
return y
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_dim': 4, 'out_dim': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
import math
import torch.nn as nn
import torch.nn.functional as F
from torch.nn import init
from torch.nn.parameter import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_mul_sigmoid_tanh_0(in_ptr0, in_ptr1, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp2 = tl.load(in_ptr1 + x0, xmask)
tmp1 = libdevice.tanh(tmp0)
tmp3 = tl.sigmoid(tmp2)
tmp4 = tmp1 * tmp3
tl.store(out_ptr0 + x0, tmp4, xmask)
@triton.jit
def triton_poi_fused_abs_add_log_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.
constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = tl_math.abs(tmp0)
tmp2 = 1e-10
tmp3 = tmp1 + tmp2
tmp4 = tl_math.log(tmp3)
tl.store(out_ptr0 + x0, tmp4, xmask)
@triton.jit
def triton_poi_fused_add_exp_mul_rsub_sigmoid_2(in_ptr0, in_ptr1, in_ptr2,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp2 = tl.load(in_ptr1 + x0, xmask)
tmp6 = tl.load(in_ptr2 + x0, xmask)
tmp1 = tl.sigmoid(tmp0)
tmp3 = tmp1 * tmp2
tmp4 = 1.0
tmp5 = tmp4 - tmp1
tmp7 = tl_math.exp(tmp6)
tmp8 = tmp5 * tmp7
tmp9 = tmp3 + tmp8
tl.store(out_ptr0 + x0, tmp9, xmask)
def call(args):
primals_1, primals_2, primals_3, primals_4 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_4, (4, 4), (4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_sigmoid_tanh_0[grid(16)](primals_1, primals_2,
buf0, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(buf0, (4, 4), (1, 4), 0), out=buf1)
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf2)
del primals_4
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_abs_add_log_1[grid(256)](primals_3, buf3, 256,
XBLOCK=128, num_warps=4, num_stages=1)
buf4 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf3, (64, 4), (4, 1), 0),
reinterpret_tensor(buf0, (4, 4), (1, 4), 0), out=buf4)
del buf0
buf5 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_add_exp_mul_rsub_sigmoid_2[grid(256)](buf2, buf1,
buf4, buf5, 256, XBLOCK=256, num_warps=4, num_stages=1)
return buf5, primals_1, primals_2, reinterpret_tensor(primals_3, (64, 4
), (4, 1), 0), buf1, buf2, reinterpret_tensor(buf3, (64, 4), (4, 1), 0
), buf4
class NeuralAccumulatorCell(nn.Module):
"""A Neural Accumulator (NAC) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.W_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.M_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.register_parameter('W_hat', self.W_hat)
self.register_parameter('M_hat', self.M_hat)
self.register_parameter('bias', None)
self._reset_params()
def _reset_params(self):
init.kaiming_uniform_(self.W_hat)
init.kaiming_uniform_(self.M_hat)
def forward(self, input):
W = torch.tanh(self.W_hat) * torch.sigmoid(self.M_hat)
return F.linear(input, W, self.bias)
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
class NALUCellNew(nn.Module):
"""A Neural Arithmetic Logic Unit (NALU) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.eps = 1e-10
self.G = Parameter(torch.Tensor(out_dim, in_dim))
self.nac = NeuralAccumulatorCell(in_dim, out_dim)
self.register_parameter('bias', None)
init.kaiming_uniform_(self.G, a=math.sqrt(5))
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
def forward(self, input_0):
primals_1 = self.G
primals_2 = self.nac.W_hat
primals_4 = self.nac.M_hat
primals_3 = input_0
output = call([primals_1, primals_2, primals_3, primals_4])
return output[0]
| mikomel/machine-number-sense | NALUCell | false | 7,225 | [
"MIT"
] | 1 | 173b67e4f25bd8249ba4a41904d4cd4af26bae05 | https://github.com/mikomel/machine-number-sense/tree/173b67e4f25bd8249ba4a41904d4cd4af26bae05 | import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn import init
from torch.nn.parameter import Parameter
class NeuralAccumulatorCell(nn.Module):
"""A Neural Accumulator (NAC) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.W_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.M_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.register_parameter('W_hat', self.W_hat)
self.register_parameter('M_hat', self.M_hat)
self.register_parameter('bias', None)
self._reset_params()
def _reset_params(self):
init.kaiming_uniform_(self.W_hat)
init.kaiming_uniform_(self.M_hat)
def forward(self, input):
W = torch.tanh(self.W_hat) * torch.sigmoid(self.M_hat)
return F.linear(input, W, self.bias)
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
class Model(nn.Module):
"""A Neural Arithmetic Logic Unit (NALU) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.eps = 1e-10
self.G = Parameter(torch.Tensor(out_dim, in_dim))
self.nac = NeuralAccumulatorCell(in_dim, out_dim)
self.register_parameter('bias', None)
init.kaiming_uniform_(self.G, a=math.sqrt(5))
def forward(self, input: 'torch.Tensor'):
a = self.nac(input)
g = F.linear(input, self.G, self.bias).sigmoid()
add_sub = g * a
log_input = (input.abs() + self.eps).log()
m = self.nac(log_input).exp()
mul_div = (1 - g) * m
y = add_sub + mul_div
return y
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
MHAttention | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/es/cessyszuy24k7hdnytcsgt5xuwettl2skzip7jijkrie3ydgjozj.py
# Topologically Sorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
# Graph fragment:
# %mul_scalar : [num_users=1] = call_function[target=torch.ops.aten.mul.Scalar](args = (%permute_default, 1.0), kwargs = {})
# %clone_default : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_default,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_0 = async_compile.triton('triton_poi_fused_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4) % 4
x2 = (xindex // 16) % 4
x3 = (xindex // 64)
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (4*x2) + (16*x1) + (64*x3)), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 1.0
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + (x4), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ro/crok5xh52iy35hijarlrh7rm5fjekyqx2yqdv3e43wmmetnqrhyk.py
# Topologically Sorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
# Graph fragment:
# %amax_default : [num_users=1] = call_function[target=torch.ops.aten.amax.default](args = (%view_default_2, [-1], True), kwargs = {})
# %sub_tensor : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%view_default_2, %amax_default), kwargs = {})
# %exp_default : [num_users=2] = call_function[target=torch.ops.aten.exp.default](args = (%sub_tensor,), kwargs = {})
triton_poi_fused_1 = async_compile.triton('triton_poi_fused_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + (x2), tmp9, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/oy/coy2yxxou5ycpqu7c6bk2tzcpttyp3yhrpr6fcr56o2svgksjlpv.py
# Topologically Sorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
# Graph fragment:
# %sum_dim_int_list : [num_users=1] = call_function[target=torch.ops.aten.sum.dim_IntList](args = (%exp_default, [-1], True), kwargs = {})
# %div_tensor : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%exp_default, %sum_dim_int_list), kwargs = {})
# %eq_scalar : [num_users=1] = call_function[target=torch.ops.aten.eq.Scalar](args = (%view_default_2, -inf), kwargs = {})
# %logical_not_default : [num_users=1] = call_function[target=torch.ops.aten.logical_not.default](args = (%eq_scalar,), kwargs = {})
# %any_dim : [num_users=1] = call_function[target=torch.ops.aten.any.dim](args = (%logical_not_default, -1, True), kwargs = {})
# %logical_not_default_1 : [num_users=1] = call_function[target=torch.ops.aten.logical_not.default](args = (%any_dim,), kwargs = {})
# %full_default : [num_users=1] = call_function[target=torch.ops.aten.full.default](args = ([4, 4, 4, 4, 4], 0), kwargs = {dtype: torch.float32, layout: torch.strided, device: cuda:0, pin_memory: False})
# %where_self : [num_users=2] = call_function[target=torch.ops.aten.where.self](args = (%logical_not_default_1, %full_default, %div_tensor), kwargs = {})
triton_poi_fused_2 = async_compile.triton('triton_poi_fused_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_2', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 9, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = (xindex // 4)
x2 = xindex
tmp0 = tl.load(in_ptr0 + (4*x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp18 = tl.load(in_ptr0 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp25 = tl.load(in_ptr1 + (x2), xmask)
tmp26 = tl.load(in_ptr1 + (4*x1), xmask, eviction_policy='evict_last')
tmp27 = tl.load(in_ptr1 + (1 + (4*x1)), xmask, eviction_policy='evict_last')
tmp29 = tl.load(in_ptr1 + (2 + (4*x1)), xmask, eviction_policy='evict_last')
tmp31 = tl.load(in_ptr1 + (3 + (4*x1)), xmask, eviction_policy='evict_last')
tmp1 = float("-inf")
tmp2 = tmp0 == tmp1
tmp3 = tmp2 == 0
tmp4 = tmp3.to(tl.int64)
tmp5 = (tmp4 != 0)
tmp7 = tmp6 == tmp1
tmp8 = tmp7 == 0
tmp9 = tmp8.to(tl.int64)
tmp10 = (tmp9 != 0)
tmp11 = tmp5 | tmp10
tmp13 = tmp12 == tmp1
tmp14 = tmp13 == 0
tmp15 = tmp14.to(tl.int64)
tmp16 = (tmp15 != 0)
tmp17 = tmp11 | tmp16
tmp19 = tmp18 == tmp1
tmp20 = tmp19 == 0
tmp21 = tmp20.to(tl.int64)
tmp22 = (tmp21 != 0)
tmp23 = tmp17 | tmp22
tmp24 = tmp23 == 0
tmp28 = tmp26 + tmp27
tmp30 = tmp28 + tmp29
tmp32 = tmp30 + tmp31
tmp33 = tmp25 / tmp32
tmp34 = 0.0
tmp35 = tl.where(tmp24, tmp34, tmp33)
tl.store(out_ptr0 + (x2), tmp35, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/dn/cdnhfw6jof6hdivjstzo4v5fqsocihi7yin5e7dcduosczadyxxx.py
# Topologically Sorted Source Nodes: [], Original ATen: []
# Source node to ATen node mapping:
# Graph fragment:
# %clone_default_2 : [num_users=1] = call_function[target=torch.ops.aten.clone.default](args = (%expand_default_3,), kwargs = {memory_format: torch.contiguous_format})
triton_poi_fused_3 = async_compile.triton('triton_poi_fused_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_3', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_3(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = (xindex // 4) % 4
x2 = (xindex // 16) % 4
x3 = (xindex // 64)
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + (4*x2) + (16*x1) + (64*x3)), xmask)
tmp1 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + (x4), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4, ), (1, ))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0), reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf0)
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0), reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0), reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf2)
del primals_6
buf3 = empty_strided_cuda((4, 4, 4, 4, 1), (64, 16, 4, 1, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
stream0 = get_raw_stream(0)
triton_poi_fused_0.run(buf0, primals_3, buf3, 256, grid=grid(256), stream=stream0)
del primals_3
buf4 = reinterpret_tensor(buf0, (4, 4, 4, 1, 4), (64, 16, 4, 4, 1), 0); del buf0 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
triton_poi_fused_0.run(buf1, primals_5, buf4, 256, grid=grid(256), stream=stream0)
del primals_5
buf5 = empty_strided_cuda((64, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.bmm(reinterpret_tensor(buf3, (64, 4, 1), (4, 1, 0), 0), reinterpret_tensor(buf4, (64, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
triton_poi_fused_1.run(buf5, buf6, 1024, grid=grid(1024), stream=stream0)
buf7 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
triton_poi_fused_2.run(buf5, buf6, buf7, 1024, grid=grid(1024), stream=stream0)
del buf5
del buf6
buf8 = reinterpret_tensor(buf1, (4, 4, 4, 4, 1), (64, 16, 4, 1, 1), 0); del buf1 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
triton_poi_fused_3.run(buf2, primals_7, buf8, 256, grid=grid(256), stream=stream0)
del primals_7
buf9 = reinterpret_tensor(buf2, (64, 4, 1), (4, 1, 1), 0); del buf2 # reuse
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.bmm(reinterpret_tensor(buf7, (64, 4, 4), (16, 4, 1), 0), reinterpret_tensor(buf8, (64, 4, 1), (4, 1, 0), 0), out=buf9)
return (reinterpret_tensor(buf9, (4, 4, 4, 4), (64, 16, 1, 4), 0), reinterpret_tensor(primals_1, (64, 4), (4, 1), 0), buf7, reinterpret_tensor(buf8, (64, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf3, (64, 1, 4), (4, 1, 1), 0), reinterpret_tensor(buf4, (64, 4, 1), (4, 1, 4), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import math
import torch
from torch import nn
import torch.nn.functional as F
class MHAttention(nn.Module):
def __init__(self, ninp, nhead, dropout):
super(MHAttention, self).__init__()
if ninp % nhead != 0:
raise ValueError(
'The hidden size is not a multiple of the number of attention heads'
)
self.nhead = nhead
self.ninp = ninp
self.fc_query = nn.Linear(ninp, ninp)
self.fc_key = nn.Linear(ninp, ninp)
self.fc_value = nn.Linear(ninp, ninp)
self.dropout = nn.Dropout(dropout)
def transpose_for_scores(self, x):
"""
x has shape (*, L, C)
return shape (*, nhead, L, C/nhead)
"""
new_shape = x.shape[:-1] + (self.nhead, -1)
x = x.view(*new_shape)
return x.transpose(-3, -2)
def forward_fn(self, x):
"""
x has shape (*, L, C)
return shape (*, L, C)
"""
query = self.transpose_for_scores(self.fc_query(x))
key = self.transpose_for_scores(self.fc_key(x))
value = self.transpose_for_scores(self.fc_value(x))
attention_scores = torch.matmul(query, key.transpose(-1, -2))
attention_scores = attention_scores / math.sqrt(self.ninp / self.nhead)
attention_weights = F.softmax(attention_scores, dim=-1)
attention_weights = self.dropout(attention_weights)
x = torch.matmul(attention_weights, value)
x = x.transpose(-3, -2)
x = x.reshape(*x.shape[:-2], -1)
return x
def forward(self, x):
chunk_size = 100000 // x.shape[2]
outputs = []
for i in range(0, x.shape[1], chunk_size):
ed = min(i + chunk_size, x.shape[1])
partial = self.forward_fn(x[:, i:ed])
outputs.append(partial)
return torch.cat(outputs, dim=1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'ninp': 4, 'nhead': 4, 'dropout': 0.5}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import math as tl_math
import math
from torch import nn
import torch.nn.functional as F
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4 % 4
x2 = xindex // 16 % 4
x3 = xindex // 64
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 4 * x2 + 16 * x1 + 64 * x3), xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = 1.0
tmp4 = tmp2 * tmp3
tl.store(out_ptr0 + x4, tmp4, xmask)
@triton.jit
def triton_poi_fused_1(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp2 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last')
tmp3 = triton_helpers.maximum(tmp1, tmp2)
tmp5 = triton_helpers.maximum(tmp3, tmp4)
tmp7 = triton_helpers.maximum(tmp5, tmp6)
tmp8 = tmp0 - tmp7
tmp9 = tl_math.exp(tmp8)
tl.store(out_ptr0 + x2, tmp9, xmask)
@triton.jit
def triton_poi_fused_2(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 1024
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x1 = xindex // 4
x2 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x1, xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (1 + 4 * x1), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr0 + (2 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp18 = tl.load(in_ptr0 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp25 = tl.load(in_ptr1 + x2, xmask)
tmp26 = tl.load(in_ptr1 + 4 * x1, xmask, eviction_policy='evict_last')
tmp27 = tl.load(in_ptr1 + (1 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp29 = tl.load(in_ptr1 + (2 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp31 = tl.load(in_ptr1 + (3 + 4 * x1), xmask, eviction_policy='evict_last'
)
tmp1 = float('-inf')
tmp2 = tmp0 == tmp1
tmp3 = tmp2 == 0
tmp4 = tmp3.to(tl.int64)
tmp5 = tmp4 != 0
tmp7 = tmp6 == tmp1
tmp8 = tmp7 == 0
tmp9 = tmp8.to(tl.int64)
tmp10 = tmp9 != 0
tmp11 = tmp5 | tmp10
tmp13 = tmp12 == tmp1
tmp14 = tmp13 == 0
tmp15 = tmp14.to(tl.int64)
tmp16 = tmp15 != 0
tmp17 = tmp11 | tmp16
tmp19 = tmp18 == tmp1
tmp20 = tmp19 == 0
tmp21 = tmp20.to(tl.int64)
tmp22 = tmp21 != 0
tmp23 = tmp17 | tmp22
tmp24 = tmp23 == 0
tmp28 = tmp26 + tmp27
tmp30 = tmp28 + tmp29
tmp32 = tmp30 + tmp31
tmp33 = tmp25 / tmp32
tmp34 = 0.0
tmp35 = tl.where(tmp24, tmp34, tmp33)
tl.store(out_ptr0 + x2, tmp35, xmask)
@triton.jit
def triton_poi_fused_3(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 4
x1 = xindex // 4 % 4
x2 = xindex // 16 % 4
x3 = xindex // 64
x4 = xindex
tmp0 = tl.load(in_ptr0 + (x0 + 4 * x2 + 16 * x1 + 64 * x3), xmask)
tmp1 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(out_ptr0 + x4, tmp2, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (4, 4), (4, 1))
assert_size_stride(primals_5, (4,), (1,))
assert_size_stride(primals_6, (4, 4), (4, 1))
assert_size_stride(primals_7, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_2, (4, 4), (1, 4), 0), out=buf0)
del primals_2
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_4, (4, 4), (1, 4), 0), out=buf1)
del primals_4
buf2 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_1, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_6, (4, 4), (1, 4), 0), out=buf2)
del primals_6
buf3 = empty_strided_cuda((4, 4, 4, 4, 1), (64, 16, 4, 1, 1), torch
.float32)
get_raw_stream(0)
triton_poi_fused_0[grid(256)](buf0, primals_3, buf3, 256, XBLOCK=
256, num_warps=4, num_stages=1)
del primals_3
buf4 = reinterpret_tensor(buf0, (4, 4, 4, 1, 4), (64, 16, 4, 4, 1), 0)
del buf0
triton_poi_fused_0[grid(256)](buf1, primals_5, buf4, 256, XBLOCK=
256, num_warps=4, num_stages=1)
del primals_5
buf5 = empty_strided_cuda((64, 4, 4), (16, 4, 1), torch.float32)
extern_kernels.bmm(reinterpret_tensor(buf3, (64, 4, 1), (4, 1, 0),
0), reinterpret_tensor(buf4, (64, 1, 4), (4, 0, 1), 0), out=buf5)
buf6 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1),
torch.float32)
triton_poi_fused_1[grid(1024)](buf5, buf6, 1024, XBLOCK=256,
num_warps=4, num_stages=1)
buf7 = empty_strided_cuda((4, 4, 4, 4, 4), (256, 64, 16, 4, 1),
torch.float32)
triton_poi_fused_2[grid(1024)](buf5, buf6, buf7, 1024, XBLOCK=256,
num_warps=4, num_stages=1)
del buf5
del buf6
buf8 = reinterpret_tensor(buf1, (4, 4, 4, 4, 1), (64, 16, 4, 1, 1), 0)
del buf1
triton_poi_fused_3[grid(256)](buf2, primals_7, buf8, 256, XBLOCK=
128, num_warps=4, num_stages=1)
del primals_7
buf9 = reinterpret_tensor(buf2, (64, 4, 1), (4, 1, 1), 0)
del buf2
extern_kernels.bmm(reinterpret_tensor(buf7, (64, 4, 4), (16, 4, 1),
0), reinterpret_tensor(buf8, (64, 4, 1), (4, 1, 0), 0), out=buf9)
return reinterpret_tensor(buf9, (4, 4, 4, 4), (64, 16, 1, 4), 0
), reinterpret_tensor(primals_1, (64, 4), (4, 1), 0
), buf7, reinterpret_tensor(buf8, (64, 1, 4), (4, 1, 1), 0
), reinterpret_tensor(buf3, (64, 1, 4), (4, 1, 1), 0
), reinterpret_tensor(buf4, (64, 4, 1), (4, 1, 4), 0)
class MHAttentionNew(nn.Module):
def __init__(self, ninp, nhead, dropout):
super(MHAttentionNew, self).__init__()
if ninp % nhead != 0:
raise ValueError(
'The hidden size is not a multiple of the number of attention heads'
)
self.nhead = nhead
self.ninp = ninp
self.fc_query = nn.Linear(ninp, ninp)
self.fc_key = nn.Linear(ninp, ninp)
self.fc_value = nn.Linear(ninp, ninp)
self.dropout = nn.Dropout(dropout)
def transpose_for_scores(self, x):
"""
x has shape (*, L, C)
return shape (*, nhead, L, C/nhead)
"""
new_shape = x.shape[:-1] + (self.nhead, -1)
x = x.view(*new_shape)
return x.transpose(-3, -2)
def forward_fn(self, x):
"""
x has shape (*, L, C)
return shape (*, L, C)
"""
query = self.transpose_for_scores(self.fc_query(x))
key = self.transpose_for_scores(self.fc_key(x))
value = self.transpose_for_scores(self.fc_value(x))
attention_scores = torch.matmul(query, key.transpose(-1, -2))
attention_scores = attention_scores / math.sqrt(self.ninp / self.nhead)
attention_weights = F.softmax(attention_scores, dim=-1)
attention_weights = self.dropout(attention_weights)
x = torch.matmul(attention_weights, value)
x = x.transpose(-3, -2)
x = x.reshape(*x.shape[:-2], -1)
return x
def forward(self, input_0):
primals_2 = self.fc_query.weight
primals_3 = self.fc_query.bias
primals_4 = self.fc_key.weight
primals_5 = self.fc_key.bias
primals_6 = self.fc_value.weight
primals_7 = self.fc_value.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7])
return output[0]
| microsoft/Protein-Folding | MHAttention | false | 7,226 | [
"MIT"
] | 1 | f534b2dd1e3f192fbcdadf234f25828c7f458a58 | https://github.com/microsoft/Protein-Folding/tree/f534b2dd1e3f192fbcdadf234f25828c7f458a58 | import math
import torch
from torch import nn
import torch.nn.functional as F
class Model(nn.Module):
def __init__(self, ninp, nhead, dropout):
super().__init__()
if ninp % nhead != 0:
raise ValueError(
'The hidden size is not a multiple of the number of attention heads'
)
self.nhead = nhead
self.ninp = ninp
self.fc_query = nn.Linear(ninp, ninp)
self.fc_key = nn.Linear(ninp, ninp)
self.fc_value = nn.Linear(ninp, ninp)
self.dropout = nn.Dropout(dropout)
def transpose_for_scores(self, x):
"""
x has shape (*, L, C)
return shape (*, nhead, L, C/nhead)
"""
new_shape = x.shape[:-1] + (self.nhead, -1)
x = x.view(*new_shape)
return x.transpose(-3, -2)
def forward_fn(self, x):
"""
x has shape (*, L, C)
return shape (*, L, C)
"""
query = self.transpose_for_scores(self.fc_query(x))
key = self.transpose_for_scores(self.fc_key(x))
value = self.transpose_for_scores(self.fc_value(x))
attention_scores = torch.matmul(query, key.transpose(-1, -2))
attention_scores = attention_scores / math.sqrt(self.ninp / self.nhead)
attention_weights = F.softmax(attention_scores, dim=-1)
attention_weights = self.dropout(attention_weights)
x = torch.matmul(attention_weights, value)
x = x.transpose(-3, -2)
x = x.reshape(*x.shape[:-2], -1)
return x
def forward(self, x):
chunk_size = 100000 // x.shape[2]
outputs = []
for i in range(0, x.shape[1], chunk_size):
ed = min(i + chunk_size, x.shape[1])
partial = self.forward_fn(x[:, i:ed])
outputs.append(partial)
return torch.cat(outputs, dim=1)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4, 0.5]
|
FeedForward | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/gi/cgi45knaonkz7f2ymhhwdszxotrkx3bl6ehavmfo5zp3bzzio3zk.py
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# x => add
# x_1 => var_mean
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_1, %primals_2), kwargs = {})
# %var_mean : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add, [3]), kwargs = {correction: 0, keepdim: True})
triton_poi_fused_add_native_layer_norm_0 = async_compile.triton('triton_poi_fused_add_native_layer_norm_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_0(in_ptr0, in_ptr1, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + (4*x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr1 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + (x0), tmp16, xmask)
tl.store(out_ptr1 + (x0), tmp28, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/nh/cnhbjzi35eshrjhk7tjj644pcpxhgyi7ajxmdjzgjtdqbnbek7qt.py
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.add, aten.native_layer_norm]
# Source node to ATen node mapping:
# x => add
# x_1 => add_1, add_2, mul, mul_1, rsqrt, sub
# Graph fragment:
# %add : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%primals_1, %primals_2), kwargs = {})
# %add_1 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem, 1e-05), kwargs = {})
# %rsqrt : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_1,), kwargs = {})
# %sub : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add, %getitem_1), kwargs = {})
# %mul : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub, %rsqrt), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul, %primals_3), kwargs = {})
# %add_2 : [num_users=2] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_1, %primals_4), kwargs = {})
triton_poi_fused_add_native_layer_norm_1 = async_compile.triton('triton_poi_fused_add_native_layer_norm_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: '*fp32', 7: '*fp32', 8: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_native_layer_norm_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 6, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_native_layer_norm_1(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x2), xmask)
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x1), xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-05
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + (x2), tmp9, xmask)
tl.store(out_ptr1 + (x2), tmp13, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/gm/cgmflgdlpeeb52xctoa47uvw47ycyf7ahlj5wdscxdatpbwcboco.py
# Topologically Sorted Source Nodes: [relu], Original ATen: [aten.relu, aten.threshold_backward]
# Source node to ATen node mapping:
# relu => relu
# Graph fragment:
# %relu : [num_users=2] = call_function[target=torch.ops.aten.relu.default](args = (%view_1,), kwargs = {})
# %le : [num_users=1] = call_function[target=torch.ops.aten.le.Scalar](args = (%relu, 0), kwargs = {})
triton_poi_fused_relu_threshold_backward_2 = async_compile.triton('triton_poi_fused_relu_threshold_backward_2', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*i1', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_relu_threshold_backward_2', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_relu_threshold_backward_2(in_out_ptr0, in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr0 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
tl.store(out_ptr0 + (x2), tmp6, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/b5/cb5kfengtwtso2cktophcv2j56lrz4uz52si4koheb5liewfkvtt.py
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.add]
# Source node to ATen node mapping:
# x_2 => add_3
# Graph fragment:
# %add_3 : [num_users=3] = call_function[target=torch.ops.aten.add.Tensor](args = (%add_2, %view_3), kwargs = {})
triton_poi_fused_add_3 = async_compile.triton('triton_poi_fused_add_3', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_3', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 3, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_3(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_out_ptr0 + (x2), xmask)
tmp2 = tl.load(in_ptr1 + (x0), xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = tmp0 + tmp3
tl.store(in_out_ptr0 + (x2), tmp4, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/jt/cjtlevawtbqokt4uvrkcbjuuzlruz6jq7ejlurchfnrudj2srbgt.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# x_3 => add_4, rsqrt_1, var_mean_1
# Graph fragment:
# %var_mean_1 : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add_3, [3]), kwargs = {correction: 0, keepdim: True})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_2, 1e-05), kwargs = {})
# %rsqrt_1 : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_4,), kwargs = {})
triton_poi_fused_native_layer_norm_4 = async_compile.triton('triton_poi_fused_native_layer_norm_4', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_4', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 4, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_4(in_ptr0, out_ptr0, out_ptr1, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (4*x0), xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + (4*x0)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + (4*x0)), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + (4*x0)), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + (x0), tmp8, xmask)
tl.store(out_ptr1 + (x0), tmp23, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/xx/cxxrwl3itdrd647cc4gbcbwnrcxegxemlfvpyuyno2wav523n6ad.py
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.native_layer_norm]
# Source node to ATen node mapping:
# x_3 => add_4, add_5, mul_2, mul_3, rsqrt_1, sub_1, var_mean_1
# Graph fragment:
# %var_mean_1 : [num_users=2] = call_function[target=torch.ops.aten.var_mean.correction](args = (%add_3, [3]), kwargs = {correction: 0, keepdim: True})
# %add_4 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%getitem_2, 1e-05), kwargs = {})
# %rsqrt_1 : [num_users=1] = call_function[target=torch.ops.aten.rsqrt.default](args = (%add_4,), kwargs = {})
# %sub_1 : [num_users=1] = call_function[target=torch.ops.aten.sub.Tensor](args = (%add_3, %getitem_3), kwargs = {})
# %mul_2 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%sub_1, %rsqrt_1), kwargs = {})
# %mul_3 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%mul_2, %primals_9), kwargs = {})
# %add_5 : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul_3, %primals_10), kwargs = {})
triton_poi_fused_native_layer_norm_5 = async_compile.triton('triton_poi_fused_native_layer_norm_5', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: '*fp32', 4: '*fp32', 5: '*fp32', 6: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_native_layer_norm_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_native_layer_norm_5(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = (xindex // 4)
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + (x2), xmask)
tmp1 = tl.load(in_ptr1 + (x1), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + (x1), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + (x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + (x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + (x2), tmp8, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
assert_size_stride(primals_4, (4, ), (1, ))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4, ), (1, ))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (4, ), (1, ))
assert_size_stride(primals_9, (4, ), (1, ))
assert_size_stride(primals_10, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf1 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.add, aten.native_layer_norm]
stream0 = get_raw_stream(0)
triton_poi_fused_add_native_layer_norm_0.run(primals_1, primals_2, buf0, buf1, 64, grid=grid(64), stream=stream0)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x, x_1], Original ATen: [aten.add, aten.native_layer_norm]
triton_poi_fused_add_native_layer_norm_1.run(primals_1, primals_2, buf0, buf1, primals_3, primals_4, buf2, buf3, 256, grid=grid(256), stream=stream0)
del primals_1
del primals_2
del primals_3
del primals_4
buf4 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf3, (64, 4), (4, 1), 0), reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf4)
buf5 = reinterpret_tensor(buf4, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf4 # reuse
buf11 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
# Topologically Sorted Source Nodes: [relu], Original ATen: [aten.relu, aten.threshold_backward]
triton_poi_fused_relu_threshold_backward_2.run(buf5, primals_6, buf11, 256, grid=grid(256), stream=stream0)
del primals_6
buf6 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [], Original ATen: []
extern_kernels.mm(reinterpret_tensor(buf5, (64, 4), (4, 1), 0), reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf6)
buf7 = reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0); del buf6 # reuse
# Topologically Sorted Source Nodes: [x_2], Original ATen: [aten.add]
triton_poi_fused_add_3.run(buf7, buf3, primals_8, 256, grid=grid(256), stream=stream0)
del primals_8
buf8 = buf1; del buf1 # reuse
buf9 = buf0; del buf0 # reuse
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_4.run(buf7, buf8, buf9, 64, grid=grid(64), stream=stream0)
buf10 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [x_3], Original ATen: [aten.native_layer_norm]
triton_poi_fused_native_layer_norm_5.run(buf7, buf8, buf9, primals_9, primals_10, buf10, 256, grid=grid(256), stream=stream0)
del buf8
del buf9
del primals_10
return (buf10, primals_9, buf2, reinterpret_tensor(buf3, (64, 4), (4, 1), 0), reinterpret_tensor(buf5, (64, 4), (4, 1), 0), buf7, primals_7, buf11, primals_5, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_4 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_5 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_6 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_7 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_8 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_9 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
primals_10 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3, primals_4, primals_5, primals_6, primals_7, primals_8, primals_9, primals_10])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
from torch import nn
class FeedForward(nn.Module):
def __init__(self, ninp, dim_feedforward, dropout):
super(FeedForward, self).__init__()
self.linear1 = nn.Linear(ninp, dim_feedforward)
self.dropout = nn.Dropout(dropout)
self.linear2 = nn.Linear(dim_feedforward, ninp)
self.norm1 = nn.LayerNorm(ninp)
self.norm2 = nn.LayerNorm(ninp)
self.dropout1 = nn.Dropout(dropout)
self.dropout2 = nn.Dropout(dropout)
self.activation = nn.ReLU()
def forward_fn(self, x, branch):
x = x + self.dropout1(branch)
x = self.norm1(x)
branch = self.linear2(self.dropout(self.activation(self.linear1(x))))
x = x + self.dropout2(branch)
x = self.norm2(x)
return x
def forward(self, x, branch):
chunk_size = 100000 // x.shape[2]
outputs = []
for i in range(0, x.shape[1], chunk_size):
ed = min(i + chunk_size, x.shape[1])
partial = self.forward_fn(x[:, i:ed], branch[:, i:ed])
outputs.append(partial)
return torch.cat(outputs, dim=1)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'ninp': 4, 'dim_feedforward': 4, 'dropout': 0.5}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime import triton_helpers
from torch._inductor.runtime.triton_helpers import libdevice
from torch import nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_add_native_layer_norm_0(in_ptr0, in_ptr1, out_ptr0,
out_ptr1, xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr1 + 4 * x0, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp4 = tl.load(in_ptr1 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp8 = tl.load(in_ptr1 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp11 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp12 = tl.load(in_ptr1 + (3 + 4 * x0), xmask, eviction_policy='evict_last'
)
tmp2 = tmp0 + tmp1
tmp5 = tmp3 + tmp4
tmp6 = tmp2 + tmp5
tmp9 = tmp7 + tmp8
tmp10 = tmp6 + tmp9
tmp13 = tmp11 + tmp12
tmp14 = tmp10 + tmp13
tmp15 = 4.0
tmp16 = tmp14 / tmp15
tmp17 = tmp2 - tmp16
tmp18 = tmp17 * tmp17
tmp19 = tmp5 - tmp16
tmp20 = tmp19 * tmp19
tmp21 = tmp18 + tmp20
tmp22 = tmp9 - tmp16
tmp23 = tmp22 * tmp22
tmp24 = tmp21 + tmp23
tmp25 = tmp13 - tmp16
tmp26 = tmp25 * tmp25
tmp27 = tmp24 + tmp26
tmp28 = tmp27 / tmp15
tl.store(out_ptr0 + x0, tmp16, xmask)
tl.store(out_ptr1 + x0, tmp28, xmask)
@triton.jit
def triton_poi_fused_add_native_layer_norm_1(in_ptr0, in_ptr1, in_ptr2,
in_ptr3, in_ptr4, in_ptr5, out_ptr0, out_ptr1, xnumel, XBLOCK: tl.constexpr
):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x2, xmask)
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x1, xmask, eviction_policy='evict_last')
tmp10 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp12 = tl.load(in_ptr5 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 - tmp3
tmp6 = 1e-05
tmp7 = tmp5 + tmp6
tmp8 = libdevice.rsqrt(tmp7)
tmp9 = tmp4 * tmp8
tmp11 = tmp9 * tmp10
tmp13 = tmp11 + tmp12
tl.store(out_ptr0 + x2, tmp9, xmask)
tl.store(out_ptr1 + x2, tmp13, xmask)
@triton.jit
def triton_poi_fused_relu_threshold_backward_2(in_out_ptr0, in_ptr0,
out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_out_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr0 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp3 = tl.full([1], 0, tl.int32)
tmp4 = triton_helpers.maximum(tmp3, tmp2)
tmp5 = 0.0
tmp6 = tmp4 <= tmp5
tl.store(in_out_ptr0 + x2, tmp4, xmask)
tl.store(out_ptr0 + x2, tmp6, xmask)
@triton.jit
def triton_poi_fused_add_3(in_out_ptr0, in_ptr0, in_ptr1, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_out_ptr0 + x2, xmask)
tmp2 = tl.load(in_ptr1 + x0, xmask, eviction_policy='evict_last')
tmp3 = tmp1 + tmp2
tmp4 = tmp0 + tmp3
tl.store(in_out_ptr0 + x2, tmp4, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_4(in_ptr0, out_ptr0, out_ptr1,
xnumel, XBLOCK: tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + 4 * x0, xmask, eviction_policy='evict_last')
tmp1 = tl.load(in_ptr0 + (1 + 4 * x0), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (2 + 4 * x0), xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr0 + (3 + 4 * x0), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tmp4 = tmp2 + tmp3
tmp6 = tmp4 + tmp5
tmp7 = 4.0
tmp8 = tmp6 / tmp7
tmp9 = tmp0 - tmp8
tmp10 = tmp9 * tmp9
tmp11 = tmp1 - tmp8
tmp12 = tmp11 * tmp11
tmp13 = tmp10 + tmp12
tmp14 = tmp3 - tmp8
tmp15 = tmp14 * tmp14
tmp16 = tmp13 + tmp15
tmp17 = tmp5 - tmp8
tmp18 = tmp17 * tmp17
tmp19 = tmp16 + tmp18
tmp20 = tmp19 / tmp7
tmp21 = 1e-05
tmp22 = tmp20 + tmp21
tmp23 = libdevice.rsqrt(tmp22)
tl.store(out_ptr0 + x0, tmp8, xmask)
tl.store(out_ptr1 + x0, tmp23, xmask)
@triton.jit
def triton_poi_fused_native_layer_norm_5(in_ptr0, in_ptr1, in_ptr2, in_ptr3,
in_ptr4, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x2 = xindex
x1 = xindex // 4
x0 = xindex % 4
tmp0 = tl.load(in_ptr0 + x2, xmask)
tmp1 = tl.load(in_ptr1 + x1, xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr2 + x1, xmask, eviction_policy='evict_last')
tmp5 = tl.load(in_ptr3 + x0, xmask, eviction_policy='evict_last')
tmp7 = tl.load(in_ptr4 + x0, xmask, eviction_policy='evict_last')
tmp2 = tmp0 - tmp1
tmp4 = tmp2 * tmp3
tmp6 = tmp4 * tmp5
tmp8 = tmp6 + tmp7
tl.store(out_ptr0 + x2, tmp8, xmask)
def call(args):
(primals_1, primals_2, primals_3, primals_4, primals_5, primals_6,
primals_7, primals_8, primals_9, primals_10) = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4,), (1,))
assert_size_stride(primals_4, (4,), (1,))
assert_size_stride(primals_5, (4, 4), (4, 1))
assert_size_stride(primals_6, (4,), (1,))
assert_size_stride(primals_7, (4, 4), (4, 1))
assert_size_stride(primals_8, (4,), (1,))
assert_size_stride(primals_9, (4,), (1,))
assert_size_stride(primals_10, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
buf1 = empty_strided_cuda((4, 4, 4, 1), (16, 4, 1, 64), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_native_layer_norm_0[grid(64)](primals_1,
primals_2, buf0, buf1, 64, XBLOCK=64, num_warps=1, num_stages=1)
buf2 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
buf3 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_add_native_layer_norm_1[grid(256)](primals_1,
primals_2, buf0, buf1, primals_3, primals_4, buf2, buf3, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
del primals_2
del primals_3
del primals_4
buf4 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf3, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_5, (4, 4), (1, 4), 0), out=buf4)
buf5 = reinterpret_tensor(buf4, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf4
buf11 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.bool)
triton_poi_fused_relu_threshold_backward_2[grid(256)](buf5,
primals_6, buf11, 256, XBLOCK=128, num_warps=4, num_stages=1)
del primals_6
buf6 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(buf5, (64, 4), (4, 1), 0),
reinterpret_tensor(primals_7, (4, 4), (1, 4), 0), out=buf6)
buf7 = reinterpret_tensor(buf6, (4, 4, 4, 4), (64, 16, 4, 1), 0)
del buf6
triton_poi_fused_add_3[grid(256)](buf7, buf3, primals_8, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_8
buf8 = buf1
del buf1
buf9 = buf0
del buf0
triton_poi_fused_native_layer_norm_4[grid(64)](buf7, buf8, buf9, 64,
XBLOCK=64, num_warps=1, num_stages=1)
buf10 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
triton_poi_fused_native_layer_norm_5[grid(256)](buf7, buf8, buf9,
primals_9, primals_10, buf10, 256, XBLOCK=128, num_warps=4,
num_stages=1)
del buf8
del buf9
del primals_10
return buf10, primals_9, buf2, reinterpret_tensor(buf3, (64, 4), (4, 1), 0
), reinterpret_tensor(buf5, (64, 4), (4, 1), 0
), buf7, primals_7, buf11, primals_5
class FeedForwardNew(nn.Module):
def __init__(self, ninp, dim_feedforward, dropout):
super(FeedForwardNew, self).__init__()
self.linear1 = nn.Linear(ninp, dim_feedforward)
self.dropout = nn.Dropout(dropout)
self.linear2 = nn.Linear(dim_feedforward, ninp)
self.norm1 = nn.LayerNorm(ninp)
self.norm2 = nn.LayerNorm(ninp)
self.dropout1 = nn.Dropout(dropout)
self.dropout2 = nn.Dropout(dropout)
self.activation = nn.ReLU()
def forward_fn(self, x, branch):
x = x + self.dropout1(branch)
x = self.norm1(x)
branch = self.linear2(self.dropout(self.activation(self.linear1(x))))
x = x + self.dropout2(branch)
x = self.norm2(x)
return x
def forward(self, input_0, input_1):
primals_5 = self.linear1.weight
primals_3 = self.linear1.bias
primals_7 = self.linear2.weight
primals_4 = self.linear2.bias
primals_6 = self.norm1.weight
primals_8 = self.norm1.bias
primals_9 = self.norm2.weight
primals_10 = self.norm2.bias
primals_1 = input_0
primals_2 = input_1
output = call([primals_1, primals_2, primals_3, primals_4,
primals_5, primals_6, primals_7, primals_8, primals_9, primals_10])
return output[0]
| microsoft/Protein-Folding | FeedForward | false | 7,227 | [
"MIT"
] | 1 | f534b2dd1e3f192fbcdadf234f25828c7f458a58 | https://github.com/microsoft/Protein-Folding/tree/f534b2dd1e3f192fbcdadf234f25828c7f458a58 | import torch
from torch import nn
class Model(nn.Module):
def __init__(self, ninp, dim_feedforward, dropout):
super().__init__()
self.linear1 = nn.Linear(ninp, dim_feedforward)
self.dropout = nn.Dropout(dropout)
self.linear2 = nn.Linear(dim_feedforward, ninp)
self.norm1 = nn.LayerNorm(ninp)
self.norm2 = nn.LayerNorm(ninp)
self.dropout1 = nn.Dropout(dropout)
self.dropout2 = nn.Dropout(dropout)
self.activation = nn.ReLU()
def forward_fn(self, x, branch):
x = x + self.dropout1(branch)
x = self.norm1(x)
branch = self.linear2(self.dropout(self.activation(self.linear1(x))))
x = x + self.dropout2(branch)
x = self.norm2(x)
return x
def forward(self, x, branch):
chunk_size = 100000 // x.shape[2]
outputs = []
for i in range(0, x.shape[1], chunk_size):
ed = min(i + chunk_size, x.shape[1])
partial = self.forward_fn(x[:, i:ed], branch[:, i:ed])
outputs.append(partial)
return torch.cat(outputs, dim=1)
def get_inputs():
return [torch.rand([4, 4, 4, 4]), torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4, 0.5]
|
NeuralAccumulatorCell | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/lz/clz2gei2r4v6zp73e5uwuxfqwprorq7wyulpi7djhsfrhn32mgcm.py
# Topologically Sorted Source Nodes: [tanh, sigmoid, W], Original ATen: [aten.tanh, aten.sigmoid, aten.mul]
# Source node to ATen node mapping:
# W => mul
# sigmoid => sigmoid
# tanh => tanh
# Graph fragment:
# %tanh : [num_users=1] = call_function[target=torch.ops.aten.tanh.default](args = (%primals_1,), kwargs = {})
# %sigmoid : [num_users=1] = call_function[target=torch.ops.aten.sigmoid.default](args = (%primals_2,), kwargs = {})
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%tanh, %sigmoid), kwargs = {})
triton_poi_fused_mul_sigmoid_tanh_0 = async_compile.triton('triton_poi_fused_mul_sigmoid_tanh_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[16],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: '*fp32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_sigmoid_tanh_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_sigmoid_tanh_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp2 = tl.load(in_ptr1 + (x0), xmask)
tmp1 = libdevice.tanh(tmp0)
tmp3 = tl.sigmoid(tmp2)
tmp4 = tmp1 * tmp3
tl.store(out_ptr0 + (x0), tmp4, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [tanh, sigmoid, W], Original ATen: [aten.tanh, aten.sigmoid, aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_sigmoid_tanh_0.run(primals_1, primals_2, buf0, 16, grid=grid(16), stream=stream0)
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
# Topologically Sorted Source Nodes: [linear], Original ATen: [aten.mm]
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), reinterpret_tensor(buf0, (4, 4), (1, 4), 0), out=buf1)
del buf0
return (reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 16, 4, 1), 0), primals_1, primals_2, reinterpret_tensor(primals_3, (64, 4), (4, 1), 0), )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4), (4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn import init
from torch.nn.parameter import Parameter
class NeuralAccumulatorCell(nn.Module):
"""A Neural Accumulator (NAC) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.W_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.M_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.register_parameter('W_hat', self.W_hat)
self.register_parameter('M_hat', self.M_hat)
self.register_parameter('bias', None)
self._reset_params()
def _reset_params(self):
init.kaiming_uniform_(self.W_hat)
init.kaiming_uniform_(self.M_hat)
def forward(self, input):
W = torch.tanh(self.W_hat) * torch.sigmoid(self.M_hat)
return F.linear(input, W, self.bias)
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_dim': 4, 'out_dim': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
from torch.nn import init
from torch.nn.parameter import Parameter
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
@triton.jit
def triton_poi_fused_mul_sigmoid_tanh_0(in_ptr0, in_ptr1, out_ptr0, xnumel,
XBLOCK: tl.constexpr):
xnumel = 16
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp2 = tl.load(in_ptr1 + x0, xmask)
tmp1 = libdevice.tanh(tmp0)
tmp3 = tl.sigmoid(tmp2)
tmp4 = tmp1 * tmp3
tl.store(out_ptr0 + x0, tmp4, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4), (4, 1))
assert_size_stride(primals_2, (4, 4), (4, 1))
assert_size_stride(primals_3, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4), (4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_sigmoid_tanh_0[grid(16)](primals_1, primals_2,
buf0, 16, XBLOCK=16, num_warps=1, num_stages=1)
buf1 = empty_strided_cuda((64, 4), (4, 1), torch.float32)
extern_kernels.mm(reinterpret_tensor(primals_3, (64, 4), (4, 1), 0),
reinterpret_tensor(buf0, (4, 4), (1, 4), 0), out=buf1)
del buf0
return reinterpret_tensor(buf1, (4, 4, 4, 4), (64, 16, 4, 1), 0
), primals_1, primals_2, reinterpret_tensor(primals_3, (64, 4), (4,
1), 0)
class NeuralAccumulatorCellNew(nn.Module):
"""A Neural Accumulator (NAC) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.W_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.M_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.register_parameter('W_hat', self.W_hat)
self.register_parameter('M_hat', self.M_hat)
self.register_parameter('bias', None)
self._reset_params()
def _reset_params(self):
init.kaiming_uniform_(self.W_hat)
init.kaiming_uniform_(self.M_hat)
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
def forward(self, input_0):
primals_1 = self.W_hat
primals_2 = self.M_hat
primals_3 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| mikomel/machine-number-sense | NeuralAccumulatorCell | false | 7,228 | [
"MIT"
] | 1 | 173b67e4f25bd8249ba4a41904d4cd4af26bae05 | https://github.com/mikomel/machine-number-sense/tree/173b67e4f25bd8249ba4a41904d4cd4af26bae05 | import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn import init
from torch.nn.parameter import Parameter
class Model(nn.Module):
"""A Neural Accumulator (NAC) cell [1].
Attributes:
in_dim: size of the input sample.
out_dim: size of the output sample.
Sources:
[1]: https://arxiv.org/abs/1808.00508
"""
def __init__(self, in_dim, out_dim):
super().__init__()
self.in_dim = in_dim
self.out_dim = out_dim
self.W_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.M_hat = Parameter(torch.Tensor(out_dim, in_dim))
self.register_parameter('W_hat', self.W_hat)
self.register_parameter('M_hat', self.M_hat)
self.register_parameter('bias', None)
self._reset_params()
def _reset_params(self):
init.kaiming_uniform_(self.W_hat)
init.kaiming_uniform_(self.M_hat)
def forward(self, input):
W = torch.tanh(self.W_hat) * torch.sigmoid(self.M_hat)
return F.linear(input, W, self.bias)
def extra_repr(self):
return 'in_dim={}, out_dim={}'.format(self.in_dim, self.out_dim)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
Conv3x3 | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/v6/cv6oewqqnsshd7he7ylh2kikzu4smtrhj2dmv6nb5csosp7g6vw5.py
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.reflection_pad2d]
# Source node to ATen node mapping:
# out => _unsafe_index, _unsafe_index_1
# Graph fragment:
# %_unsafe_index : [num_users=1] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%primals_1, [None, None, %sub_1, None]), kwargs = {})
# %_unsafe_index_1 : [num_users=2] = call_function[target=torch.ops.aten._unsafe_index.Tensor](args = (%_unsafe_index, [None, None, None, %sub_1]), kwargs = {})
triton_poi_fused_reflection_pad2d_0 = async_compile.triton('triton_poi_fused_reflection_pad2d_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[1024],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_reflection_pad2d_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_reflection_pad2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 576
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 6
x1 = (xindex // 6) % 6
x2 = (xindex // 36)
x3 = xindex
tmp0 = tl.load(in_ptr0 + (15 + ((-1)*(tl_math.abs((-3) + (tl_math.abs((-1) + x0))))) + ((-4)*(tl_math.abs((-3) + (tl_math.abs((-1) + x1))))) + (16*x2)), xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + (x3), tmp0, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/32/c32v7egt4mupqssam3gmac2qgv3ujprjybthsgweflmot256qqw7.py
# Topologically Sorted Source Nodes: [out_1], Original ATen: [aten.convolution]
# Source node to ATen node mapping:
# out_1 => convolution
# Graph fragment:
# %convolution : [num_users=1] = call_function[target=torch.ops.aten.convolution.default](args = (%_unsafe_index_1, %primals_2, %primals_3, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
triton_poi_fused_convolution_1 = async_compile.triton('triton_poi_fused_convolution_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 16) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 6, 6), (144, 36, 6, 1), torch.float32)
# Topologically Sorted Source Nodes: [out], Original ATen: [aten.reflection_pad2d]
stream0 = get_raw_stream(0)
triton_poi_fused_reflection_pad2d_0.run(primals_1, buf0, 576, grid=grid(576), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [out_1], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1, 1), padding=(0, 0), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4, 4), (64, 16, 4, 1))
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [out_1], Original ATen: [aten.convolution]
triton_poi_fused_convolution_1.run(buf2, primals_3, 256, grid=grid(256), stream=stream0)
del primals_3
return (buf2, primals_2, buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 3, 3), (36, 9, 3, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
class Conv3x3(nn.Module):
"""Layer to pad and convolve input
"""
def __init__(self, in_channels, out_channels, use_refl=True):
super(Conv3x3, self).__init__()
if use_refl:
self.pad = nn.ReflectionPad2d(1)
else:
self.pad = nn.ZeroPad2d(1)
self.conv = nn.Conv2d(int(in_channels), int(out_channels), 3)
def forward(self, x):
out = self.pad(x)
out = self.conv(out)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'in_channels': 4, 'out_channels': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import math as tl_math
import torch.nn as nn
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_reflection_pad2d_0(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 576
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex % 6
x1 = xindex // 6 % 6
x2 = xindex // 36
x3 = xindex
tmp0 = tl.load(in_ptr0 + (15 + -1 * tl_math.abs(-3 + tl_math.abs(-1 +
x0)) + -4 * tl_math.abs(-3 + tl_math.abs(-1 + x1)) + 16 * x2),
xmask, eviction_policy='evict_last')
tl.store(out_ptr0 + x3, tmp0, xmask)
@triton.jit
def triton_poi_fused_convolution_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl
.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 16 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 3, 3), (36, 9, 3, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 6, 6), (144, 36, 6, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_reflection_pad2d_0[grid(576)](primals_1, buf0, 576,
XBLOCK=128, num_warps=4, num_stages=1)
del primals_1
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1, 1),
padding=(0, 0), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 4, 4), (64, 16, 4, 1))
buf2 = buf1
del buf1
triton_poi_fused_convolution_1[grid(256)](buf2, primals_3, 256,
XBLOCK=256, num_warps=4, num_stages=1)
del primals_3
return buf2, primals_2, buf0
class Conv3x3New(nn.Module):
"""Layer to pad and convolve input
"""
def __init__(self, in_channels, out_channels, use_refl=True):
super(Conv3x3New, self).__init__()
if use_refl:
self.pad = nn.ReflectionPad2d(1)
else:
self.pad = nn.ZeroPad2d(1)
self.conv = nn.Conv2d(int(in_channels), int(out_channels), 3)
def forward(self, input_0):
primals_2 = self.conv.weight
primals_3 = self.conv.bias
primals_1 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| minjabenho/image2pcl | Conv3x3 | false | 7,229 | [
"Apache-2.0"
] | 1 | 7e696ee48edae30814d32f32e605ad6cf8bf702c | https://github.com/minjabenho/image2pcl/tree/7e696ee48edae30814d32f32e605ad6cf8bf702c | import torch
import torch.nn as nn
class Model(nn.Module):
"""Layer to pad and convolve input
"""
def __init__(self, in_channels, out_channels, use_refl=True):
super().__init__()
if use_refl:
self.pad = nn.ReflectionPad2d(1)
else:
self.pad = nn.ZeroPad2d(1)
self.conv = nn.Conv2d(int(in_channels), int(out_channels), 3)
def forward(self, x):
out = self.pad(x)
out = self.conv(out)
return out
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4]
|
fadein_layer | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/ai/cai2poorsup7nq2otw37m3ij7xgd7zz7yh67qrptmbrb5helnvpb.py
# Topologically Sorted Source Nodes: [mul, mul_1, add], Original ATen: [aten.mul, aten.add]
# Source node to ATen node mapping:
# add => add
# mul => mul
# mul_1 => mul_1
# Graph fragment:
# %mul : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%select, 1.0), kwargs = {})
# %mul_1 : [num_users=1] = call_function[target=torch.ops.aten.mul.Tensor](args = (%select_1, 0.0), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mul, %mul_1), kwargs = {})
triton_poi_fused_add_mul_0 = async_compile.triton('triton_poi_fused_add_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[64],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp3 = tl.load(in_ptr0 + (64 + x0), xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = 0.0
tmp5 = tmp3 * tmp4
tmp6 = tmp2 + tmp5
tl.store(out_ptr0 + (x0), tmp6, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul, mul_1, add], Original ATen: [aten.mul, aten.add]
stream0 = get_raw_stream(0)
triton_poi_fused_add_mul_0.run(arg0_1, buf0, 64, grid=grid(64), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| from _paritybench_helpers import _mock_config
import torch
import torch.nn as nn
import torch.utils.data
class fadein_layer(nn.Module):
def __init__(self, config):
super(fadein_layer, self).__init__()
self.alpha = 0.0
def update_alpha(self, delta):
self.alpha = self.alpha + delta
self.alpha = max(0, min(self.alpha, 1.0))
def set_alpha(self, value):
self.alpha = max(0, min(value, 1.0))
def forward(self, x):
return torch.add(x[0].mul(1.0 - self.alpha), x[1].mul(self.alpha))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'config': _mock_config()}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr
):
xnumel = 64
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp3 = tl.load(in_ptr0 + (64 + x0), xmask)
tmp1 = 1.0
tmp2 = tmp0 * tmp1
tmp4 = 0.0
tmp5 = tmp3 * tmp4
tmp6 = tmp2 + tmp5
tl.store(out_ptr0 + x0, tmp6, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4), (16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_mul_0[grid(64)](arg0_1, buf0, 64, XBLOCK=64,
num_warps=1, num_stages=1)
del arg0_1
return buf0,
class fadein_layerNew(nn.Module):
def __init__(self, config):
super(fadein_layerNew, self).__init__()
self.alpha = 0.0
def update_alpha(self, delta):
self.alpha = self.alpha + delta
self.alpha = max(0, min(self.alpha, 1.0))
def set_alpha(self, value):
self.alpha = max(0, min(value, 1.0))
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mingo-x/pggan-pytorch | fadein_layer | false | 7,230 | [
"MIT"
] | 1 | a1dde73cd4df52476fe7c948d81fa9caea8070a5 | https://github.com/mingo-x/pggan-pytorch/tree/a1dde73cd4df52476fe7c948d81fa9caea8070a5 | from _paritybench_helpers import _mock_config
import torch
import torch.nn as nn
import torch.utils.data
class Model(nn.Module):
def __init__(self, config):
super().__init__()
self.alpha = 0.0
def update_alpha(self, delta):
self.alpha = self.alpha + delta
self.alpha = max(0, min(self.alpha, 1.0))
def set_alpha(self, value):
self.alpha = max(0, min(value, 1.0))
def forward(self, x):
return torch.add(x[0].mul(1.0 - self.alpha), x[1].mul(self.alpha))
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
pixelwise_norm_layer | # AOT ID: ['0_inference']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/lg/clgum5lgqb2igszc5bgmtzzepcvym6hp73vttrcwjtey34a6skhy.py
# Topologically Sorted Source Nodes: [pow_1, mean, add, pow_2, truediv], Original ATen: [aten.pow, aten.mean, aten.add, aten.div]
# Source node to ATen node mapping:
# add => add
# mean => mean
# pow_1 => pow_1
# pow_2 => pow_2
# truediv => div
# Graph fragment:
# %pow_1 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%arg0_1, 2), kwargs = {})
# %mean : [num_users=1] = call_function[target=torch.ops.aten.mean.dim](args = (%pow_1, [1], True), kwargs = {})
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%mean, 1e-08), kwargs = {})
# %pow_2 : [num_users=1] = call_function[target=torch.ops.aten.pow.Tensor_Scalar](args = (%add, 0.5), kwargs = {})
# %div : [num_users=1] = call_function[target=torch.ops.aten.div.Tensor](args = (%arg0_1, %pow_2), kwargs = {})
triton_poi_fused_add_div_mean_pow_0 = async_compile.triton('triton_poi_fused_add_div_mean_pow_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mean_pow_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 5, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_div_mean_pow_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = (xindex // 64)
tmp0 = tl.load(in_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x0 + (64*x2)), xmask, eviction_policy='evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = 4.0
tmp13 = tmp11 / tmp12
tmp14 = 1e-08
tmp15 = tmp13 + tmp14
tmp16 = libdevice.sqrt(tmp15)
tmp17 = tmp0 / tmp16
tl.store(out_ptr0 + (x3), tmp17, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [pow_1, mean, add, pow_2, truediv], Original ATen: [aten.pow, aten.mean, aten.add, aten.div]
stream0 = get_raw_stream(0)
triton_poi_fused_add_div_mean_pow_0.run(arg0_1, buf0, 256, grid=grid(256), stream=stream0)
del arg0_1
return (buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
arg0_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
fn = lambda: call([arg0_1])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
import torch.utils.data
class pixelwise_norm_layer(nn.Module):
def __init__(self):
super(pixelwise_norm_layer, self).__init__()
self.eps = 1e-08
def forward(self, x):
return x / (torch.mean(x ** 2, dim=1, keepdim=True) + self.eps) ** 0.5
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {}]
| import torch
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
from torch._inductor.runtime.triton_helpers import libdevice
import torch.nn as nn
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_add_div_mean_pow_0(in_ptr0, out_ptr0, xnumel, XBLOCK:
tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x0 = xindex % 16
x2 = xindex // 64
tmp0 = tl.load(in_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + (x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp3 = tl.load(in_ptr0 + (16 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp6 = tl.load(in_ptr0 + (32 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp9 = tl.load(in_ptr0 + (48 + x0 + 64 * x2), xmask, eviction_policy=
'evict_last')
tmp2 = tmp1 * tmp1
tmp4 = tmp3 * tmp3
tmp5 = tmp2 + tmp4
tmp7 = tmp6 * tmp6
tmp8 = tmp5 + tmp7
tmp10 = tmp9 * tmp9
tmp11 = tmp8 + tmp10
tmp12 = 4.0
tmp13 = tmp11 / tmp12
tmp14 = 1e-08
tmp15 = tmp13 + tmp14
tmp16 = libdevice.sqrt(tmp15)
tmp17 = tmp0 / tmp16
tl.store(out_ptr0 + x3, tmp17, xmask)
def call(args):
arg0_1, = args
args.clear()
assert_size_stride(arg0_1, (4, 4, 4, 4), (64, 16, 4, 1))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_add_div_mean_pow_0[grid(256)](arg0_1, buf0, 256,
XBLOCK=128, num_warps=4, num_stages=1)
del arg0_1
return buf0,
class pixelwise_norm_layerNew(nn.Module):
def __init__(self):
super(pixelwise_norm_layerNew, self).__init__()
self.eps = 1e-08
def forward(self, input_0):
arg0_1 = input_0
output = call([arg0_1])
return output[0]
| mingo-x/pggan-pytorch | pixelwise_norm_layer | false | 7,231 | [
"MIT"
] | 1 | a1dde73cd4df52476fe7c948d81fa9caea8070a5 | https://github.com/mingo-x/pggan-pytorch/tree/a1dde73cd4df52476fe7c948d81fa9caea8070a5 | import torch
import torch.nn as nn
import torch.utils.data
class Model(nn.Module):
def __init__(self):
super().__init__()
self.eps = 1e-08
def forward(self, x):
return x / (torch.mean(x ** 2, dim=1, keepdim=True) + self.eps) ** 0.5
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return []
|
equalized_conv2d | # AOT ID: ['0_forward']
from ctypes import c_void_p, c_long, c_int
import torch
import math
import random
import os
import tempfile
from math import inf, nan
from torch._inductor.hooks import run_intermediate_hooks
from torch._inductor.utils import maybe_profile
from torch._inductor.codegen.memory_planning import _align as align
from torch import device, empty_strided
from torch._inductor.async_compile import AsyncCompile
from torch._inductor.select_algorithm import extern_kernels
from torch._inductor.codegen.multi_kernel import MultiKernelCall
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
aten = torch.ops.aten
inductor_ops = torch.ops.inductor
_quantized = torch.ops._quantized
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
empty_strided_xpu = torch._C._dynamo.guards._empty_strided_xpu
reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor
alloc_from_pool = torch.ops.inductor._alloc_from_pool
async_compile = AsyncCompile()
# kernel path: runs/run_shard_4/inductor_cache/mr/cmrzofxtfa5fe3ax4o3n5qvgpvhbgcrspjauzarmp4t443npav4h.py
# Topologically Sorted Source Nodes: [mul], Original ATen: [aten.mul]
# Source node to ATen node mapping:
# mul => mul
# Graph fragment:
# %mul : [num_users=2] = call_function[target=torch.ops.aten.mul.Tensor](args = (%primals_1, 0.1767766952966369), kwargs = {})
triton_poi_fused_mul_0 = async_compile.triton('triton_poi_fused_mul_0', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[256],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_mul_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + (x0), xmask)
tmp1 = 0.1767766952966369
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + (x0), tmp2, xmask)
''', device_str='cuda')
# kernel path: runs/run_shard_4/inductor_cache/ng/cng33o2pb2zyuwp4i2wm3nk4pftenhhzjucbrpjfmm3hxihqatqj.py
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
# Source node to ATen node mapping:
# add => add
# Graph fragment:
# %add : [num_users=1] = call_function[target=torch.ops.aten.add.Tensor](args = (%convolution, %expand), kwargs = {})
triton_poi_fused_add_1 = async_compile.triton('triton_poi_fused_add_1', '''
import triton
import triton.language as tl
from triton.compiler.compiler import AttrsDescriptor
from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties
@triton_heuristics.pointwise(
size_hints=[2048],
filename=__file__,
triton_meta={'signature': {0: '*fp32', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=80, major=8, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=108), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]},
inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_1', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': 'A9C866B4A14FD3277824029365D703C2427B2E685E54EC9B3EF4ADC8D1EEAC1D', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False},
min_elem_per_thread=0
)
@triton.jit
def triton_poi_fused_add_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr):
xnumel = 1296
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = (xindex // 81) % 4
tmp0 = tl.load(in_out_ptr0 + (x3), xmask)
tmp1 = tl.load(in_ptr0 + (x1), xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + (x3), tmp2, xmask)
''', device_str='cuda')
async_compile.wait(globals())
del async_compile
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4, ), (1, ))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
# Topologically Sorted Source Nodes: [mul], Original ATen: [aten.mul]
stream0 = get_raw_stream(0)
triton_poi_fused_mul_0.run(primals_1, buf0, 256, grid=grid(256), stream=stream0)
del primals_1
# Topologically Sorted Source Nodes: [x], Original ATen: [aten.convolution]
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1, 1), padding=(4, 4), dilation=(1, 1), transposed=False, output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 9, 9), (324, 81, 9, 1))
buf2 = buf1; del buf1 # reuse
# Topologically Sorted Source Nodes: [add], Original ATen: [aten.add]
triton_poi_fused_add_1.run(buf2, primals_3, 1296, grid=grid(1296), stream=stream0)
del primals_3
return (buf2, primals_2, buf0, )
def benchmark_compiled_module(times=10, repeat=10):
from torch._dynamo.testing import rand_strided
from torch._inductor.utils import print_performance
primals_1 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_2 = rand_strided((4, 4, 4, 4), (64, 16, 4, 1), device='cuda:0', dtype=torch.float32)
primals_3 = rand_strided((4, ), (1, ), device='cuda:0', dtype=torch.float32)
fn = lambda: call([primals_1, primals_2, primals_3])
return print_performance(fn, times=times, repeat=repeat)
if __name__ == "__main__":
from torch._inductor.wrapper_benchmark import compiled_module_main
compiled_module_main('None', benchmark_compiled_module)
| import torch
import torch.nn as nn
from torch.nn.init import normal
import torch.utils.data
def _calculate_fan_in_and_fan_out(tensor):
dimensions = tensor.ndimension()
if dimensions < 2:
raise ValueError(
'Fan in and fan out can not be computed for tensor with less than 2 dimensions'
)
if dimensions == 2:
fan_in = tensor.size(1)
fan_out = tensor.size(0)
else:
num_input_fmaps = tensor.size(1)
num_output_fmaps = tensor.size(0)
receptive_field_size = 1
if tensor.dim() > 2:
receptive_field_size = tensor[0][0].numel()
fan_in = num_input_fmaps * receptive_field_size
fan_out = num_output_fmaps * receptive_field_size
return fan_in, fan_out
class equalized_conv2d(nn.Module):
def __init__(self, c_in, c_out, k_size, stride, pad, initializer=
'kaiming', bias=False, a=0.0):
super(equalized_conv2d, self).__init__()
self.conv = nn.Conv2d(c_in, c_out, k_size, stride, pad, bias=False)
if initializer == 'kaiming':
normal(self.conv.weight)
fan_in, _ = _calculate_fan_in_and_fan_out(self.conv.weight)
gain = (2.0 / (1.0 + a ** 2)) ** 0.5
self.scale = gain / fan_in ** 0.5
self.bias = torch.nn.Parameter(torch.FloatTensor(c_out).fill_(0))
def forward(self, x):
x = self.conv(x.mul(self.scale))
return x + self.bias.view(1, -1, 1, 1).expand_as(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [[], {'c_in': 4, 'c_out': 4, 'k_size': 4, 'stride': 1, 'pad': 4}]
| import torch
from torch._inductor.select_algorithm import extern_kernels
import triton
import triton.language as tl
from torch._inductor.runtime.triton_heuristics import grid
from torch._C import _cuda_getCurrentRawStream as get_raw_stream
import torch.nn as nn
from torch.nn.init import normal
import torch.utils.data
assert_size_stride = torch._C._dynamo.guards.assert_size_stride
empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda
@triton.jit
def triton_poi_fused_mul_0(in_ptr0, out_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 256
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x0 = xindex
tmp0 = tl.load(in_ptr0 + x0, xmask)
tmp1 = 0.1767766952966369
tmp2 = tmp0 * tmp1
tl.store(out_ptr0 + x0, tmp2, xmask)
@triton.jit
def triton_poi_fused_add_1(in_out_ptr0, in_ptr0, xnumel, XBLOCK: tl.constexpr):
xnumel = 1296
xoffset = tl.program_id(0) * XBLOCK
xindex = xoffset + tl.arange(0, XBLOCK)[:]
xmask = xindex < xnumel
x3 = xindex
x1 = xindex // 81 % 4
tmp0 = tl.load(in_out_ptr0 + x3, xmask)
tmp1 = tl.load(in_ptr0 + x1, xmask, eviction_policy='evict_last')
tmp2 = tmp0 + tmp1
tl.store(in_out_ptr0 + x3, tmp2, xmask)
def call(args):
primals_1, primals_2, primals_3 = args
args.clear()
assert_size_stride(primals_1, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_2, (4, 4, 4, 4), (64, 16, 4, 1))
assert_size_stride(primals_3, (4,), (1,))
with torch.cuda._DeviceGuard(0):
torch.cuda.set_device(0)
buf0 = empty_strided_cuda((4, 4, 4, 4), (64, 16, 4, 1), torch.float32)
get_raw_stream(0)
triton_poi_fused_mul_0[grid(256)](primals_1, buf0, 256, XBLOCK=128,
num_warps=4, num_stages=1)
del primals_1
buf1 = extern_kernels.convolution(buf0, primals_2, stride=(1, 1),
padding=(4, 4), dilation=(1, 1), transposed=False,
output_padding=(0, 0), groups=1, bias=None)
assert_size_stride(buf1, (4, 4, 9, 9), (324, 81, 9, 1))
buf2 = buf1
del buf1
triton_poi_fused_add_1[grid(1296)](buf2, primals_3, 1296, XBLOCK=
256, num_warps=4, num_stages=1)
del primals_3
return buf2, primals_2, buf0
def _calculate_fan_in_and_fan_out(tensor):
dimensions = tensor.ndimension()
if dimensions < 2:
raise ValueError(
'Fan in and fan out can not be computed for tensor with less than 2 dimensions'
)
if dimensions == 2:
fan_in = tensor.size(1)
fan_out = tensor.size(0)
else:
num_input_fmaps = tensor.size(1)
num_output_fmaps = tensor.size(0)
receptive_field_size = 1
if tensor.dim() > 2:
receptive_field_size = tensor[0][0].numel()
fan_in = num_input_fmaps * receptive_field_size
fan_out = num_output_fmaps * receptive_field_size
return fan_in, fan_out
class equalized_conv2dNew(nn.Module):
def __init__(self, c_in, c_out, k_size, stride, pad, initializer=
'kaiming', bias=False, a=0.0):
super(equalized_conv2dNew, self).__init__()
self.conv = nn.Conv2d(c_in, c_out, k_size, stride, pad, bias=False)
if initializer == 'kaiming':
normal(self.conv.weight)
fan_in, _ = _calculate_fan_in_and_fan_out(self.conv.weight)
gain = (2.0 / (1.0 + a ** 2)) ** 0.5
self.scale = gain / fan_in ** 0.5
self.bias = torch.nn.Parameter(torch.FloatTensor(c_out).fill_(0))
def forward(self, input_0):
primals_3 = self.bias
primals_1 = self.conv.weight
primals_2 = input_0
output = call([primals_1, primals_2, primals_3])
return output[0]
| mingo-x/pggan-pytorch | equalized_conv2d | false | 7,232 | [
"MIT"
] | 1 | a1dde73cd4df52476fe7c948d81fa9caea8070a5 | https://github.com/mingo-x/pggan-pytorch/tree/a1dde73cd4df52476fe7c948d81fa9caea8070a5 | import torch
import torch.nn as nn
from torch.nn.init import normal
import torch.utils.data
def _calculate_fan_in_and_fan_out(tensor):
dimensions = tensor.ndimension()
if dimensions < 2:
raise ValueError(
'Fan in and fan out can not be computed for tensor with less than 2 dimensions'
)
if dimensions == 2:
fan_in = tensor.size(1)
fan_out = tensor.size(0)
else:
num_input_fmaps = tensor.size(1)
num_output_fmaps = tensor.size(0)
receptive_field_size = 1
if tensor.dim() > 2:
receptive_field_size = tensor[0][0].numel()
fan_in = num_input_fmaps * receptive_field_size
fan_out = num_output_fmaps * receptive_field_size
return fan_in, fan_out
class Model(nn.Module):
def __init__(self, c_in, c_out, k_size, stride, pad, initializer=
'kaiming', bias=False, a=0.0):
super().__init__()
self.conv = nn.Conv2d(c_in, c_out, k_size, stride, pad, bias=False)
if initializer == 'kaiming':
normal(self.conv.weight)
fan_in, _ = _calculate_fan_in_and_fan_out(self.conv.weight)
gain = (2.0 / (1.0 + a ** 2)) ** 0.5
self.scale = gain / fan_in ** 0.5
self.bias = torch.nn.Parameter(torch.FloatTensor(c_out).fill_(0))
def forward(self, x):
x = self.conv(x.mul(self.scale))
return x + self.bias.view(1, -1, 1, 1).expand_as(x)
def get_inputs():
return [torch.rand([4, 4, 4, 4])]
def get_init_inputs():
return [4, 4, 4, 1, 4]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.