File size: 21,406 Bytes
c41b22c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e74c01b
c41b22c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e74c01b
c41b22c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
50550c9
c41b22c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e74c01b
 
 
 
 
 
 
 
 
 
c41b22c
 
 
 
 
 
 
 
 
e74c01b
 
c41b22c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
# Modified from Flux
#
# Copyright 2024 Black Forest Labs

# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at

#     http://www.apache.org/licenses/LICENSE-2.0

# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.

import math  # noqa: I001
from dataclasses import dataclass
from functools import partial

import torch
import torch.nn.functional as F
from einops import rearrange
# from liger_kernel.ops.rms_norm import LigerRMSNormFunction
from torch import Tensor, nn


try:
    import flash_attn
    from flash_attn.flash_attn_interface import (
        _flash_attn_forward,
        flash_attn_varlen_func,
    )
except ImportError:
    flash_attn = None
    flash_attn_varlen_func = None
    _flash_attn_forward = None


MEMORY_LAYOUT = {
    "flash": (
        lambda x: x.view(x.shape[0] * x.shape[1], *x.shape[2:]),
        lambda x: x,
    ),
    "torch": (
        lambda x: x.transpose(1, 2),
        lambda x: x.transpose(1, 2),
    ),
    "vanilla": (
        lambda x: x.transpose(1, 2),
        lambda x: x.transpose(1, 2),
    ),
}


def attention(
    q,
    k,
    v,
    mode="torch",
    drop_rate=0,
    attn_mask=None,
    causal=False,
    cu_seqlens_q=None,
    cu_seqlens_kv=None,
    max_seqlen_q=None,
    max_seqlen_kv=None,
    batch_size=1,
):
    """
    Perform QKV self attention.

    Args:
        q (torch.Tensor): Query tensor with shape [b, s, a, d], where a is the number of heads.
        k (torch.Tensor): Key tensor with shape [b, s1, a, d]
        v (torch.Tensor): Value tensor with shape [b, s1, a, d]
        mode (str): Attention mode. Choose from 'self_flash', 'cross_flash', 'torch', and 'vanilla'.
        drop_rate (float): Dropout rate in attention map. (default: 0)
        attn_mask (torch.Tensor): Attention mask with shape [b, s1] (cross_attn), or [b, a, s, s1] (torch or vanilla).
            (default: None)
        causal (bool): Whether to use causal attention. (default: False)
        cu_seqlens_q (torch.Tensor): dtype torch.int32. The cumulative sequence lengths of the sequences in the batch,
            used to index into q.
        cu_seqlens_kv (torch.Tensor): dtype torch.int32. The cumulative sequence lengths of the sequences in the batch,
            used to index into kv.
        max_seqlen_q (int): The maximum sequence length in the batch of q.
        max_seqlen_kv (int): The maximum sequence length in the batch of k and v.

    Returns:
        torch.Tensor: Output tensor after self attention with shape [b, s, ad]
    """
    pre_attn_layout, post_attn_layout = MEMORY_LAYOUT[mode]
    q = pre_attn_layout(q)
    k = pre_attn_layout(k)
    v = pre_attn_layout(v)

    if mode == "torch":
        if attn_mask is not None and attn_mask.dtype != torch.bool:
            attn_mask = attn_mask.to(q.dtype)
        x = F.scaled_dot_product_attention(
            q, k, v, attn_mask=attn_mask, dropout_p=drop_rate, is_causal=causal
        )
    elif mode == "flash":
        assert flash_attn_varlen_func is not None
        x: torch.Tensor = flash_attn_varlen_func(
            q,
            k,
            v,
            cu_seqlens_q,
            cu_seqlens_kv,
            max_seqlen_q,
            max_seqlen_kv,
        )  # type: ignore
        # x with shape [(bxs), a, d]
        x = x.view(batch_size, max_seqlen_q, x.shape[-2], x.shape[-1])  # type: ignore # reshape x to [b, s, a, d]
    elif mode == "vanilla":
        scale_factor = 1 / math.sqrt(q.size(-1))

        b, a, s, _ = q.shape
        s1 = k.size(2)
        attn_bias = torch.zeros(b, a, s, s1, dtype=q.dtype, device=q.device)
        if causal:
            # Only applied to self attention
            assert attn_mask is None, (
                "Causal mask and attn_mask cannot be used together"
            )
            temp_mask = torch.ones(b, a, s, s, dtype=torch.bool, device=q.device).tril(
                diagonal=0
            )
            attn_bias.masked_fill_(temp_mask.logical_not(), float("-inf"))
            attn_bias.to(q.dtype)

        if attn_mask is not None:
            if attn_mask.dtype == torch.bool:
                attn_bias.masked_fill_(attn_mask.logical_not(), float("-inf"))
            else:
                attn_bias += attn_mask

        # TODO: Maybe force q and k to be float32 to avoid numerical overflow
        attn = (q @ k.transpose(-2, -1)) * scale_factor
        attn += attn_bias
        attn = attn.softmax(dim=-1)
        attn = torch.dropout(attn, p=drop_rate, train=True)
        x = attn @ v
    else:
        raise NotImplementedError(f"Unsupported attention mode: {mode}")

    x = post_attn_layout(x)
    b, s, a, d = x.shape
    out = x.reshape(b, s, -1)
    return out


def apply_gate(x, gate=None, tanh=False):
    """AI is creating summary for apply_gate

    Args:
        x (torch.Tensor): input tensor.
        gate (torch.Tensor, optional): gate tensor. Defaults to None.
        tanh (bool, optional): whether to use tanh function. Defaults to False.

    Returns:
        torch.Tensor: the output tensor after apply gate.
    """
    if gate is None:
        return x
    if tanh:
        return x * gate.unsqueeze(1).tanh()
    else:
        return x * gate.unsqueeze(1)


class MLP(nn.Module):
    """MLP as used in Vision Transformer, MLP-Mixer and related networks"""

    def __init__(
        self,
        in_channels,
        hidden_channels=None,
        out_features=None,
        act_layer=nn.GELU,
        norm_layer=None,
        bias=True,
        drop=0.0,
        use_conv=False,
        device=None,
        dtype=None,
    ):
        super().__init__()
        out_features = out_features or in_channels
        hidden_channels = hidden_channels or in_channels
        bias = (bias, bias)
        drop_probs = (drop, drop)
        linear_layer = partial(nn.Conv2d, kernel_size=1) if use_conv else nn.Linear

        self.fc1 = linear_layer(
            in_channels, hidden_channels, bias=bias[0], device=device, dtype=dtype
        )
        self.act = act_layer()
        self.drop1 = nn.Dropout(drop_probs[0])
        self.norm = (
            norm_layer(hidden_channels, device=device, dtype=dtype)
            if norm_layer is not None
            else nn.Identity()
        )
        self.fc2 = linear_layer(
            hidden_channels, out_features, bias=bias[1], device=device, dtype=dtype
        )
        self.drop2 = nn.Dropout(drop_probs[1])

    def forward(self, x):
        x = self.fc1(x)
        x = self.act(x)
        x = self.drop1(x)
        x = self.norm(x)
        x = self.fc2(x)
        x = self.drop2(x)
        return x


class TextProjection(nn.Module):
    """
    Projects text embeddings. Also handles dropout for classifier-free guidance.

    Adapted from https://github.com/PixArt-alpha/PixArt-alpha/blob/master/diffusion/model/nets/PixArt_blocks.py
    """

    def __init__(self, in_channels, hidden_size, act_layer, dtype=None, device=None):
        factory_kwargs = {"dtype": dtype, "device": device}
        super().__init__()
        self.linear_1 = nn.Linear(
            in_features=in_channels,
            out_features=hidden_size,
            bias=True,
            **factory_kwargs,
        )
        self.act_1 = act_layer()
        self.linear_2 = nn.Linear(
            in_features=hidden_size,
            out_features=hidden_size,
            bias=True,
            **factory_kwargs,
        )

    def forward(self, caption):
        hidden_states = self.linear_1(caption)
        hidden_states = self.act_1(hidden_states)
        hidden_states = self.linear_2(hidden_states)
        return hidden_states


class TimestepEmbedder(nn.Module):
    """
    Embeds scalar timesteps into vector representations.
    """

    def __init__(
        self,
        hidden_size,
        act_layer,
        frequency_embedding_size=256,
        max_period=10000,
        out_size=None,
        dtype=None,
        device=None,
    ):
        factory_kwargs = {"dtype": dtype, "device": device}
        super().__init__()
        self.frequency_embedding_size = frequency_embedding_size
        self.max_period = max_period
        if out_size is None:
            out_size = hidden_size

        self.mlp = nn.Sequential(
            nn.Linear(
                frequency_embedding_size, hidden_size, bias=True, **factory_kwargs
            ),
            act_layer(),
            nn.Linear(hidden_size, out_size, bias=True, **factory_kwargs),
        )
        nn.init.normal_(self.mlp[0].weight, std=0.02)  # type: ignore
        nn.init.normal_(self.mlp[2].weight, std=0.02)  # type: ignore

    @staticmethod
    def timestep_embedding(t, dim, max_period=10000):
        """
        Create sinusoidal timestep embeddings.

        Args:
            t (torch.Tensor): a 1-D Tensor of N indices, one per batch element. These may be fractional.
            dim (int): the dimension of the output.
            max_period (int): controls the minimum frequency of the embeddings.

        Returns:
            embedding (torch.Tensor): An (N, D) Tensor of positional embeddings.

        .. ref_link: https://github.com/openai/glide-text2im/blob/main/glide_text2im/nn.py
        """
        half = dim // 2
        freqs = torch.exp(
            -math.log(max_period)
            * torch.arange(start=0, end=half, dtype=torch.float32)
            / half
        ).to(device=t.device)
        args = t[:, None].float() * freqs[None]
        embedding = torch.cat([torch.cos(args), torch.sin(args)], dim=-1)
        if dim % 2:
            embedding = torch.cat(
                [embedding, torch.zeros_like(embedding[:, :1])], dim=-1
            )
        return embedding

    def forward(self, t):
        t_freq = self.timestep_embedding(
            t, self.frequency_embedding_size, self.max_period
        ).type(self.mlp[0].weight.dtype)  # type: ignore
        t_emb = self.mlp(t_freq)
        return t_emb


class EmbedND(nn.Module):
    def __init__(self, dim: int, theta: int, axes_dim: list[int]):
        super().__init__()
        self.dim = dim
        self.theta = theta
        self.axes_dim = axes_dim

    def forward(self, ids: Tensor) -> Tensor:
        n_axes = ids.shape[-1]
        emb = torch.cat(
            [rope(ids[..., i], self.axes_dim[i], self.theta) for i in range(n_axes)],
            dim=-3,
        )

        return emb.unsqueeze(1)


class MLPEmbedder(nn.Module):
    def __init__(self, in_dim: int, hidden_dim: int):
        super().__init__()
        self.in_layer = nn.Linear(in_dim, hidden_dim, bias=True)
        self.silu = nn.SiLU()
        self.out_layer = nn.Linear(hidden_dim, hidden_dim, bias=True)

    def forward(self, x: Tensor) -> Tensor:
        return self.out_layer(self.silu(self.in_layer(x)))


def rope(pos, dim: int, theta: int):
    assert dim % 2 == 0
    scale = torch.arange(0, dim, 2, dtype=torch.float64, device=pos.device) / dim
    omega = 1.0 / (theta**scale)
    out = torch.einsum("...n,d->...nd", pos, omega)
    out = torch.stack(
        [torch.cos(out), -torch.sin(out), torch.sin(out), torch.cos(out)], dim=-1
    )
    out = rearrange(out, "b n d (i j) -> b n d i j", i=2, j=2)
    return out.float()


def attention_after_rope(q, k, v, pe):
    q, k = apply_rope(q, k, pe)

    from .attention import attention

    x = attention(q, k, v, mode="torch")
    return x


@torch.compile(mode="max-autotune-no-cudagraphs", dynamic=True)
def apply_rope(xq, xk, freqs_cis):
    # 将 num_heads 和 seq_len 的维度交换回原函数的处理顺序
    xq = xq.transpose(1, 2)  # [batch, num_heads, seq_len, head_dim]
    xk = xk.transpose(1, 2)

    # 将 head_dim 拆分为复数部分(实部和虚部)
    xq_ = xq.float().reshape(*xq.shape[:-1], -1, 1, 2)
    xk_ = xk.float().reshape(*xk.shape[:-1], -1, 1, 2)

    # 应用旋转位置编码(复数乘法)
    xq_out = freqs_cis[..., 0] * xq_[..., 0] + freqs_cis[..., 1] * xq_[..., 1]
    xk_out = freqs_cis[..., 0] * xk_[..., 0] + freqs_cis[..., 1] * xk_[..., 1]

    # 恢复张量形状并转置回目标维度顺序
    xq_out = xq_out.reshape(*xq.shape).type_as(xq).transpose(1, 2)
    xk_out = xk_out.reshape(*xk.shape).type_as(xk).transpose(1, 2)

    return xq_out, xk_out


@torch.compile(mode="max-autotune-no-cudagraphs", dynamic=True)
def scale_add_residual(
    x: torch.Tensor, scale: torch.Tensor, residual: torch.Tensor
) -> torch.Tensor:
    return x * scale + residual


@torch.compile(mode="max-autotune-no-cudagraphs", dynamic=True)
def layernorm_and_scale_shift(
    x: torch.Tensor, scale: torch.Tensor, shift: torch.Tensor
) -> torch.Tensor:
    return torch.nn.functional.layer_norm(x, (x.size(-1),)) * (scale + 1) + shift


class SelfAttention(nn.Module):
    def __init__(self, dim: int, num_heads: int = 8, qkv_bias: bool = False):
        super().__init__()
        self.num_heads = num_heads
        head_dim = dim // num_heads

        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
        self.norm = QKNorm(head_dim)
        self.proj = nn.Linear(dim, dim)

    def forward(self, x: Tensor, pe: Tensor) -> Tensor:
        qkv = self.qkv(x)
        q, k, v = rearrange(qkv, "B L (K H D) -> K B L H D", K=3, H=self.num_heads)
        q, k = self.norm(q, k, v)
        x = attention_after_rope(q, k, v, pe=pe)
        x = self.proj(x)
        return x


@dataclass
class ModulationOut:
    shift: Tensor
    scale: Tensor
    gate: Tensor


class RMSNorm(torch.nn.Module):
    def __init__(self, dim: int):
        super().__init__()
        self.scale = nn.Parameter(torch.ones(dim))

    # @staticmethod
    # def rms_norm_fast(x, weight, eps):
    #     return LigerRMSNormFunction.apply(
    #         x,
    #         weight,
    #         eps,
    #         0.0,
    #         "gemma",
    #         True,
    #     )

    @staticmethod
    def rms_norm(x, weight, eps):
        x_dtype = x.dtype
        x = x.float()
        rrms = torch.rsqrt(torch.mean(x**2, dim=-1, keepdim=True) + eps)
        return (x * rrms).to(dtype=x_dtype) * weight

    def forward(self, x: Tensor):
        # return self.rms_norm_fast(x, self.scale, 1e-6)
        return self.rms_norm(x, self.scale, 1e-6)


class QKNorm(torch.nn.Module):
    def __init__(self, dim: int):
        super().__init__()
        self.query_norm = RMSNorm(dim)
        self.key_norm = RMSNorm(dim)

    def forward(self, q: Tensor, k: Tensor, v: Tensor) -> tuple[Tensor, Tensor]:
        q = self.query_norm(q)
        k = self.key_norm(k)
        return q.to(v), k.to(v)


class Modulation(nn.Module):
    def __init__(self, dim: int, double: bool):
        super().__init__()
        self.is_double = double
        self.multiplier = 6 if double else 3
        self.lin = nn.Linear(dim, self.multiplier * dim, bias=True)

    def forward(self, vec: Tensor) -> tuple[ModulationOut, ModulationOut | None]:
        out = self.lin(nn.functional.silu(vec))[:, None, :].chunk(
            self.multiplier, dim=-1
        )

        return (
            ModulationOut(*out[:3]),
            ModulationOut(*out[3:]) if self.is_double else None,
        )


class DoubleStreamBlock(nn.Module):
    def __init__(
        self, hidden_size: int, num_heads: int, mlp_ratio: float, qkv_bias: bool = False
    ):
        super().__init__()

        mlp_hidden_dim = int(hidden_size * mlp_ratio)
        self.num_heads = num_heads
        self.hidden_size = hidden_size
        self.img_mod = Modulation(hidden_size, double=True)
        self.img_norm1 = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6)
        self.img_attn = SelfAttention(
            dim=hidden_size, num_heads=num_heads, qkv_bias=qkv_bias
        )

        self.img_norm2 = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6)
        self.img_mlp = nn.Sequential(
            nn.Linear(hidden_size, mlp_hidden_dim, bias=True),
            nn.GELU(approximate="tanh"),
            nn.Linear(mlp_hidden_dim, hidden_size, bias=True),
        )

        self.txt_mod = Modulation(hidden_size, double=True)
        self.txt_norm1 = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6)
        self.txt_attn = SelfAttention(
            dim=hidden_size, num_heads=num_heads, qkv_bias=qkv_bias
        )

        self.txt_norm2 = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6)
        self.txt_mlp = nn.Sequential(
            nn.Linear(hidden_size, mlp_hidden_dim, bias=True),
            nn.GELU(approximate="tanh"),
            nn.Linear(mlp_hidden_dim, hidden_size, bias=True),
        )

    def forward(
        self, img: Tensor, txt: Tensor, vec: Tensor, pe: Tensor
    ) -> tuple[Tensor, Tensor]:
        img_mod1, img_mod2 = self.img_mod(vec)
        txt_mod1, txt_mod2 = self.txt_mod(vec)

        # prepare image for attention
        img_modulated = self.img_norm1(img)
        img_modulated = (1 + img_mod1.scale) * img_modulated + img_mod1.shift
        img_qkv = self.img_attn.qkv(img_modulated)
        img_q, img_k, img_v = rearrange(
            img_qkv, "B L (K H D) -> K B L H D", K=3, H=self.num_heads
        )
        img_q, img_k = self.img_attn.norm(img_q, img_k, img_v)

        # prepare txt for attention
        txt_modulated = self.txt_norm1(txt)
        txt_modulated = (1 + txt_mod1.scale) * txt_modulated + txt_mod1.shift
        txt_qkv = self.txt_attn.qkv(txt_modulated)
        txt_q, txt_k, txt_v = rearrange(
            txt_qkv, "B L (K H D) -> K B L H D", K=3, H=self.num_heads
        )
        txt_q, txt_k = self.txt_attn.norm(txt_q, txt_k, txt_v)

        # run actual attention
        q = torch.cat((txt_q, img_q), dim=1)
        k = torch.cat((txt_k, img_k), dim=1)
        v = torch.cat((txt_v, img_v), dim=1)

        attn = attention_after_rope(q, k, v, pe=pe)
        txt_attn, img_attn = attn[:, : txt.shape[1]], attn[:, txt.shape[1] :]

        # calculate the img bloks
        img = img + img_mod1.gate * self.img_attn.proj(img_attn)
        img_mlp = self.img_mlp(
            (1 + img_mod2.scale) * self.img_norm2(img) + img_mod2.shift
        )
        img = scale_add_residual(img_mlp, img_mod2.gate, img)

        # calculate the txt bloks
        txt = txt + txt_mod1.gate * self.txt_attn.proj(txt_attn)
        txt_mlp = self.txt_mlp(
            (1 + txt_mod2.scale) * self.txt_norm2(txt) + txt_mod2.shift
        )
        txt = scale_add_residual(txt_mlp, txt_mod2.gate, txt)
        return img, txt


class SingleStreamBlock(nn.Module):
    """
    A DiT block with parallel linear layers as described in
    https://arxiv.org/abs/2302.05442 and adapted modulation interface.
    """

    def __init__(
        self,
        hidden_size: int,
        num_heads: int,
        mlp_ratio: float = 4.0,
        qk_scale: float | None = None,
    ):
        super().__init__()
        self.hidden_dim = hidden_size
        self.num_heads = num_heads
        head_dim = hidden_size // num_heads
        self.scale = qk_scale or head_dim**-0.5

        self.mlp_hidden_dim = int(hidden_size * mlp_ratio)
        # qkv and mlp_in
        self.linear1 = nn.Linear(hidden_size, hidden_size * 3 + self.mlp_hidden_dim)
        # proj and mlp_out
        self.linear2 = nn.Linear(hidden_size + self.mlp_hidden_dim, hidden_size)

        self.norm = QKNorm(head_dim)

        self.hidden_size = hidden_size
        self.pre_norm = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6)

        self.mlp_act = nn.GELU(approximate="tanh")
        self.modulation = Modulation(hidden_size, double=False)

    def forward(self, x: Tensor, vec: Tensor, pe: Tensor) -> Tensor:
        mod, _ = self.modulation(vec)
        x_mod = (1 + mod.scale) * self.pre_norm(x) + mod.shift
        qkv, mlp = torch.split(
            self.linear1(x_mod), [3 * self.hidden_size, self.mlp_hidden_dim], dim=-1
        )

        q, k, v = rearrange(qkv, "B L (K H D) -> K B L H D", K=3, H=self.num_heads)
        q, k = self.norm(q, k, v)

        # compute attention
        attn = attention_after_rope(q, k, v, pe=pe)
        # compute activation in mlp stream, cat again and run second linear layer
        output = self.linear2(torch.cat((attn, self.mlp_act(mlp)), 2))
        return scale_add_residual(output, mod.gate, x)


class LastLayer(nn.Module):
    def __init__(self, hidden_size: int, patch_size: int, out_channels: int):
        super().__init__()
        self.norm_final = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6)
        self.linear = nn.Linear(
            hidden_size, patch_size * patch_size * out_channels, bias=True
        )
        self.adaLN_modulation = nn.Sequential(
            nn.SiLU(), nn.Linear(hidden_size, 2 * hidden_size, bias=True)
        )

    def forward(self, x: Tensor, vec: Tensor) -> Tensor:
        shift, scale = self.adaLN_modulation(vec).chunk(2, dim=1)
        x = (1 + scale[:, None, :]) * self.norm_final(x) + shift[:, None, :]
        x = self.linear(x)
        return x