File size: 4,758 Bytes
adbcc81 | 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 | """
Milestone 4: Full GPT model.
Architecture:
- Token embedding table
- Learned positional embedding table (will be replaced with RoPE in modernization)
- Stack of transformer Blocks
- Final LayerNorm
- Linear language model head (maps n_embd -> vocab_size)
~10M parameters with the default config.
"""
import torch
import torch.nn as nn
import torch.nn.functional as F
from transformer import Block
class GPT(nn.Module):
def __init__(
self,
vocab_size: int,
n_embd: int = 384,
n_heads: int = 6,
n_layer: int = 6,
block_size: int = 256,
dropout: float = 0.2,
):
super().__init__()
self.block_size = block_size
self.token_emb = nn.Embedding(vocab_size, n_embd)
self.pos_emb = nn.Embedding(block_size, n_embd) # learned positional embeddings
self.blocks = nn.Sequential(*[
Block(n_embd=n_embd, n_heads=n_heads, block_size=block_size, dropout=dropout)
for _ in range(n_layer)
])
self.ln_f = nn.LayerNorm(n_embd) # final layer norm
self.lm_head = nn.Linear(n_embd, vocab_size, bias=False)
# Weight tying: share token embedding and lm_head weights.
# Standard in GPT-2 — reduces params and improves performance.
self.lm_head.weight = self.token_emb.weight
self._init_weights()
def _init_weights(self):
"""Initialize weights following GPT-2 paper."""
for module in self.modules():
if isinstance(module, nn.Linear):
nn.init.normal_(module.weight, mean=0.0, std=0.02)
if module.bias is not None:
nn.init.zeros_(module.bias)
elif isinstance(module, nn.Embedding):
nn.init.normal_(module.weight, mean=0.0, std=0.02)
def forward(self, idx: torch.Tensor, targets: torch.Tensor | None = None):
B, T = idx.shape
assert T <= self.block_size, f"Sequence length {T} exceeds block_size {self.block_size}"
positions = torch.arange(T, device=idx.device) # (T,)
x = self.token_emb(idx) + self.pos_emb(positions) # (B, T, n_embd)
x = self.blocks(x) # (B, T, n_embd)
x = self.ln_f(x) # (B, T, n_embd)
logits = self.lm_head(x) # (B, T, vocab_size)
loss = None
if targets is not None:
# Reshape for cross-entropy: (B*T, vocab_size) vs (B*T,)
loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1))
return logits, loss
@torch.no_grad()
def generate(
self,
idx: torch.Tensor,
max_new_tokens: int,
temperature: float = 1.0,
top_k: int | None = None,
) -> torch.Tensor:
"""Autoregressively generate new tokens.
Args:
idx: (B, T) tensor of seed token ids
max_new_tokens: number of tokens to generate
temperature: >1 = more random, <1 = more focused
top_k: if set, only sample from the top-k logits
"""
for _ in range(max_new_tokens):
# Crop context to block_size
idx_cond = idx[:, -self.block_size:]
logits, _ = self(idx_cond)
logits = logits[:, -1, :] / temperature # (B, vocab_size) — last time step
if top_k is not None:
# Zero out all logits below the top-k
v, _ = torch.topk(logits, min(top_k, logits.size(-1)))
logits[logits < v[:, [-1]]] = float("-inf")
probs = F.softmax(logits, dim=-1)
next_id = torch.multinomial(probs, num_samples=1) # (B, 1)
idx = torch.cat([idx, next_id], dim=1) # (B, T+1)
return idx
# ── Quick model size check ────────────────────────────────────────────────────
if __name__ == "__main__":
from tokenizer import DEVICE, VOCAB_SIZE, BLOCK_SIZE
model = GPT(vocab_size=VOCAB_SIZE, block_size=BLOCK_SIZE).to(DEVICE)
n_params = sum(p.numel() for p in model.parameters())
print(f"Model parameters: {n_params:,} (~{n_params/1e6:.1f}M)")
# Forward pass test
x = torch.zeros((2, 8), dtype=torch.long, device=DEVICE)
logits, loss = model(x, x)
print(f"Logits shape : {logits.shape} (expected [2, 8, {VOCAB_SIZE}])")
print(f"Loss (untrained): {loss.item():.4f} (expected ~{__import__('math').log(VOCAB_SIZE):.2f})")
print("\nMilestone 4 OK: GPT model works.")
|