import torch import torch.nn as nn import torch.nn.functional as F # model architecture class AttentionHead(nn.Module): """a single head of self attention""" def __init__(self, n_embed, head_size, block_size, dropout): super().__init__() self.key = nn.Linear(n_embed, head_size, bias=False) self.query = nn.Linear(n_embed, head_size, bias=False) self.value = nn.Linear(n_embed, head_size, bias=False) self.register_buffer('tril', torch.tril(torch.ones(block_size, block_size))) self.dropout = nn.Dropout(dropout) def forward(self, x): B, T, C = x.shape K = self.key(x) # (B, T, C) Q = self.query(x) # (B, T, C) wei = Q @ K.transpose(-2,-1) * C**-0.5 # (B, T, C) @ (B, H, C) -> (B, T, T) wei = wei.masked_fill(self.tril[:T, :T] == 0, float('-inf')) wei = F.softmax(wei, dim=-1) wei = self.dropout(wei) V = self.value(x) # (B, T, C) out = wei @ V # (B, T, T) @ (B, T, C) -> (B, T, C) return out class MultiHeadAttention(nn.Module): """a multi-head self attention layer""" def __init__(self, n_embed, n_heads, head_size, block_size, dropout): super().__init__() self.heads = nn.ModuleList([AttentionHead(n_embed, head_size, block_size, dropout) for _ in range(n_heads)]) self.fc = nn.Linear(head_size * n_heads, n_embed) self.dropout = nn.Dropout(dropout) def forward(self, x): out = torch.cat([h(x) for h in self.heads], dim=-1) # (B, T, n_heads*C) out = self.fc(out) # (B, T, C) out = self.dropout(out) return out class FeedForward(nn.Module): def __init__(self, n_embed, n_hidden, dropout): super().__init__() self.net = nn.Sequential( nn.Linear(n_embed, n_hidden), nn.ReLU(), nn.Linear(n_hidden, n_embed), nn.Dropout(dropout) ) def forward(self, x): return self.net(x) class Block(nn.Module): def __init__(self, n_embed, n_heads, block_size, dropout): super().__init__() self.sa_heads = MultiHeadAttention(n_embed, n_heads, n_embed // n_heads, block_size, dropout) self.ffwd = FeedForward(n_embed, n_embed*4, dropout) self.ln1 = nn.LayerNorm(n_embed) self.ln2 = nn.LayerNorm(n_embed) def forward(self, x): x = x + self.sa_heads(self.ln1(x)) # [batch_size, block_size, n_embed] x = x + self.ffwd(self.ln2(x)) # [batch_size, block_size, n_embed] return x class NanoGPT(nn.Module): def __init__(self, hyperparameters, device="cpu"): super().__init__() # hyperparameters vocab_size = hyperparameters['vocab_size'] block_size = hyperparameters['block_size'] n_embed = hyperparameters['n_embed'] n_heads = hyperparameters['n_heads'] n_layers = hyperparameters['n_layers'] dropout = hyperparameters['dropout'] self.token_embedding_table = nn.Embedding(vocab_size, n_embed) self.position_embedding_table = nn.Embedding(block_size, n_embed) self.blocks = nn.Sequential(*[Block(n_embed, n_heads, block_size, dropout) for _ in range(n_layers)]) self.ln_f = nn.LayerNorm(n_embed) self.lm_head = nn.Linear(n_embed, vocab_size) self.device = device self.block_size = block_size def forward(self, idx, targets=None): # idx and target are both [batch_size, block_size] B, T = idx.shape tok_emb = self.token_embedding_table(idx) # [batch_size, block_size, n_embed] pos_emb = self.position_embedding_table(torch.arange(T, device=self.device)) # [block_size, n_embed] x = tok_emb + pos_emb # [batch_size, block_size, n_embed] x = self.blocks(x) x = self.ln_f(x) logits = self.lm_head(x) # [batch_size, block_size, vocab_size] if targets is None: loss = None else: B, T, C = logits.shape logits = logits.view(B*T, C) targets = targets.view(B*T) loss = F.cross_entropy(logits, targets) return logits, loss # return 0, 0 def generate(self, idx, max_new_tokens=100): # idx is (B, T) for _ in range(max_new_tokens): # get the last block_size tokens idx_cond = idx[:, -self.block_size:] # (B, T) # get the predictions logits, _ = self(idx_cond) # focus only on the last time step logits = logits[:, -1, :] # becomes (B, C) # apply softmax to get probabilities probs = F.softmax(logits, dim=1) # (B, C) # sample from the distribution idx_next = torch.multinomial(probs, num_samples=1) # (B, 1) # append sampled index to the running sequence idx = torch.cat((idx, idx_next), dim=1) # (B, T+1) return idx