File size: 9,130 Bytes
82f9e44 |
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 |
import torch
import torch.nn as nn
import torch.nn.functional as F
import math
import numpy as np
class InputEmbeddings(nn.Module):
def __init__(self, d_model: int, vocab_size: int):
super().__init__()
self.d_model = d_model
self.vocab_size = vocab_size
self.embedding = nn.Embedding(vocab_size, d_model)
def forward(self, x):
return self.embedding(x) * math.sqrt(self.d_model)
class PositionalEncoding(nn.Module):
def __init__(self, d_model: int, seq_len: int, dropout):
super().__init__()
self.d_model = d_model
self.seq_len = seq_len
self.dropout = nn.Dropout(dropout)
# create matrix pe of (seq_length , d_model)
pe = torch.zeros(seq_len, d_model)
# create a vector of shape (seq_length ,1 )
# formula = pos / 10000 ** (2i / d_model)
pos = torch.arange(0, seq_len, dtype=torch.float).unsqueeze(1)
div = torch.exp(torch.arange(0, d_model, 2).float() * (-math.log(10000.0)/d_model))
# apply sin for evens
pe[:, 0::2] = torch.sin(pos * div)
# apply cos for odds
pe[:, 1::2] = torch.cos(pos * div)
# changing pe (seq_len, d_model) -> (1, seq_len, d_model)
pe = pe.unsqueeze(0)
self.register_buffer('pe', pe)
def forward(self, x):
x = x + (self.pe[:, :x.shape[1], :]).requires_grad_(False)
return self.dropout(x)
class LayerNormalization(nn.Module):
def __init__(self, eps=10**-6):
super().__init__()
self.eps = eps
self.alpha = nn.Parameter(torch.ones(1)) # multiplier
self.bias = nn.Parameter(torch.zeros(1))
def forward(self, x):
mean = x.mean(dim=-1, keepdim=True)
std = x.std(dim=-1, keepdim=True)
return self.alpha * ((x - mean) / (std + self.eps)) + self.bias
class FeedForwardBlock(nn.Module):
def __init__(self, d_model: int, d_ff: int, dropout):
super().__init__()
self.dropout = nn.Dropout(dropout)
# fully connected NN -> input:d_model(512) inner:d_ff(2048) output:d_model(512)
self.linear_1 = nn.Linear(d_model, d_ff) # w1 & b1
self.linear_2 = nn.Linear(d_ff, d_model) # w2 & b2
def forward(self, x):
# (Batch,Seq_len,d_model) --> (Batch,Seq_len,d_ff) --> (Batch,Seq__len,d_model)
return self.linear_2(self.dropout(torch.relu(self.linear_1(x))))
class MultiHeadAttentionBlock(nn.Module):
def __init__(self, d_model: int, h: int, dropout: float):
super().__init__()
self.d_model = d_model
self.h = h
# d_model must divisible by number of attention heads
assert d_model % h == 0, "d_model must divisible by h"
self.d_k = d_model//h
self.w_q = nn.Linear(d_model, d_model)
self.w_k = nn.Linear(d_model, d_model)
self.w_v = nn.Linear(d_model, d_model)
self.w_o = nn.Linear(d_model, d_model)
self.dropout = nn.Dropout(dropout)
@staticmethod
def attention(query,key,value,dropout:nn.Dropout):
# d_k = query.shape[-1]
# #(Batch, h, Seq_len, d_k) --> (Batch, h, Seq_len, Seq_len)
# attention_scores = (query @ key.transpose(-2,-1)) / math.sqrt(d_k)
# if mask is not None:
# attention_scores.masked_fill_(mask==0,-1e9)
# attention_scores = attention_scores.softmax(dim = -1) # (Batch, h, Seq_len, Seq_len)
# if dropout is not None:
# attention_scores = dropout(attention_scores)
# return (attention_scores @ value) , attention_scores
attn_output = F.scaled_dot_product_attention(
query, key, value,
attn_mask=None,
dropout_p=dropout.p if dropout is not None else 0.0,
is_causal=True # If this is for decoder/causal models, set True
)
return attn_output, None
def forward(self, q, k, v):
query = self.w_q(q) # (Batch, Seq_len, d_model) --> (Batch, Seq_len, d_model)
key = self.w_k(k) # (Batch, Seq_len, d_model) --> (Batch, Seq_len, d_model)
value = self.w_v(v) # (Batch, Seq_len, d_model) --> (Batch, Seq_len, d_model)
# (Batch, Seq_len, d_model) --> (Batch, Seq_len, d_model, h, d_k) --> (Batch, h, Seq_len, d_k)
query = query.view(query.shape[0], query.shape[1], self.h , self.d_k).transpose(1,2)
key = key.view(key.shape[0], key.shape[1], self.h , self.d_k).transpose(1,2)
value = value.view(value.shape[0], value.shape[1], self.h , self.d_k).transpose(1,2)
x,attention_scores = MultiHeadAttentionBlock.attention(query,key,value,self.dropout)
# print("Attention scores : ",attention_scores)
# print_attention(attention_scores)
# print("Attention shape : ",attention_scores.shape)
# (Batch, h, Seq_len, d_k) --> (Batch, Seq_len, h, Seq_len) --> (Batch, Seq_len, d_model)
x = x.transpose(1,2).contiguous().view(x.shape[0], -1, self.h*self.d_k)
# (Batch, Seq_len, d_model) --> (Batch, Seq_len, d_model)
return self.w_o(x)
class ResidualConnection(nn.Module):
def __init__(self, dropout:float):
super().__init__()
self.dropout = nn.Dropout(dropout)
self.norm = LayerNormalization()
def forward(self, x, sublayer):
return x + self.dropout(sublayer(self.norm(x)))
class DecoderBlock(nn.Module):
def __init__(self, self_attention_block:MultiHeadAttentionBlock,feed_forward_block:FeedForwardBlock, dropout:float ):
super().__init__()
self.self_attention_block = self_attention_block
self.feed_forward_block = feed_forward_block
self.residual_connections = nn.ModuleList((ResidualConnection(dropout) for _ in range(2)))
def forward(self, x):
x = self.residual_connections[0](x, lambda x: self.self_attention_block(x, x, x))
x = self.residual_connections[1](x, self.feed_forward_block)
return x
class Decoder(nn.Module):
def __init__(self, layers:nn.ModuleList):
super().__init__()
self.layers = layers
self.norm = LayerNormalization()
def forward(self, x):
for layer in self.layers:
x = layer(x)
return self.norm(x)
class ProjectionLayer(nn.Module):
def __init__(self, d_model:int, vocab_size :int ):
super().__init__()
self.proj = nn.Linear(d_model,vocab_size)
def forward(self, x):
#(Batch, Seq_len, d_model) --> (Batch, Seq_len, vocab_size)
return torch.log_softmax(self.proj(x),dim=-1)
class GPT(nn.Module):
def __init__(self,decoder:Decoder, tgt_embed:InputEmbeddings, tgt_pos:PositionalEncoding, projection_layer:ProjectionLayer):
super().__init__()
self.decoder = decoder
self.tgt_embed = tgt_embed
self.tgt_pos = tgt_pos
self.projection_layer = projection_layer
def decode(self, tgt):
tgt = self.tgt_embed(tgt)
tgt = self.tgt_pos(tgt)
return self.decoder(tgt)
def project(self, x):
return self.projection_layer(x)
def build_gpt(vocab_size:int, seq_len:int, d_model:int = 512, N:int = 6, h:int = 8,d_ff:int = 2048, dropout:float = 0.3) -> GPT:
#create embedding layer
tgt_embed = InputEmbeddings(d_model,vocab_size)
#create positional encoding layer
tgt_pos = PositionalEncoding(d_model, seq_len, dropout)
decoder_blocks = []
for _ in range(N):
decoder_self_attention_block = MultiHeadAttentionBlock(d_model, h, dropout)
decoder_feed_forward_block = FeedForwardBlock(d_model, d_ff, dropout)
decoder_block = DecoderBlock(decoder_self_attention_block, decoder_feed_forward_block, dropout)
decoder_blocks.append(decoder_block)
decoder = Decoder(nn.ModuleList(decoder_blocks))
#create the projection layer
projection_layer = ProjectionLayer(d_model,vocab_size)
#create the transformer
gpt = GPT(decoder,tgt_embed,tgt_pos,projection_layer)
for p in gpt.parameters():
if p.dim() > 1:
nn.init.xavier_uniform_(p)
return gpt
def colorize(value):
"""Return colored string based on value between 0 and 1."""
# Map to 0-255 for brightness
brightness = int(232 + value * 23) # 232 to 255 grayscale in ANSI
return f"\033[48;5;{brightness}m {value:.2f} \033[0m"
def print_attention(attention_scores):
batch_size, num_heads, seq_len, _ = attention_scores.shape
for head in range(num_heads):
print(f"\nAttention Head {head + 1}:\n" + "-" * 30)
attn = attention_scores[0, head] # Take batch 0
attn = (attn - attn.min()) / (attn.max() - attn.min() + 1e-8) # Normalize
for i in range(seq_len):
for j in range(seq_len):
print(colorize(attn[i, j].item()), end=" ")
print() # Newline after each row
if __name__ == "__main__":
gpt = build_gpt(10000, 350)
print(gpt) |