File size: 6,678 Bytes
6e14144
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""Decoder-only transformer with RMSNorm, RoPE, SwiGLU. Educational, modern, single-GPU."""
from __future__ import annotations

import math
from dataclasses import dataclass

import torch
import torch.nn as nn
import torch.nn.functional as F

from config import ModelConfig


class RMSNorm(nn.Module):
    def __init__(self, dim: int, eps: float = 1e-6):
        super().__init__()
        self.weight = nn.Parameter(torch.ones(dim))
        self.eps = eps

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        norm = x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps)
        return self.weight * norm.to(x.dtype)


def build_rope_cache(seq_len: int, head_dim: int, base: float, device, dtype):
    inv_freq = 1.0 / (base ** (torch.arange(0, head_dim, 2, device=device).float() / head_dim))
    t = torch.arange(seq_len, device=device).float()
    freqs = torch.outer(t, inv_freq)
    cos = freqs.cos().to(dtype)
    sin = freqs.sin().to(dtype)
    return cos, sin


def apply_rope(x: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor) -> torch.Tensor:
    # x: (B, H, T, D). Pair adjacent dims and rotate.
    x1, x2 = x[..., 0::2], x[..., 1::2]
    cos = cos[None, None, :x.size(-2), :]
    sin = sin[None, None, :x.size(-2), :]
    rot1 = x1 * cos - x2 * sin
    rot2 = x1 * sin + x2 * cos
    out = torch.stack((rot1, rot2), dim=-1).flatten(-2)
    return out


class CausalSelfAttention(nn.Module):
    def __init__(self, cfg: ModelConfig):
        super().__init__()
        self.n_head = cfg.n_head
        self.head_dim = cfg.head_dim
        self.qkv = nn.Linear(cfg.n_embd, 3 * cfg.n_embd, bias=False)
        self.proj = nn.Linear(cfg.n_embd, cfg.n_embd, bias=False)
        self.dropout = cfg.dropout

    def forward(self, x: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor) -> torch.Tensor:
        B, T, C = x.shape
        qkv = self.qkv(x)
        q, k, v = qkv.chunk(3, dim=-1)
        q = q.view(B, T, self.n_head, self.head_dim).transpose(1, 2)
        k = k.view(B, T, self.n_head, self.head_dim).transpose(1, 2)
        v = v.view(B, T, self.n_head, self.head_dim).transpose(1, 2)

        q = apply_rope(q, cos, sin)
        k = apply_rope(k, cos, sin)

        y = F.scaled_dot_product_attention(
            q, k, v,
            is_causal=True,
            dropout_p=self.dropout if self.training else 0.0,
        )
        y = y.transpose(1, 2).contiguous().view(B, T, C)
        return self.proj(y)


class SwiGLU(nn.Module):
    def __init__(self, cfg: ModelConfig):
        super().__init__()
        hidden = cfg.mlp_mult * cfg.n_embd
        # Round to multiple of 64 for efficiency.
        hidden = ((hidden + 63) // 64) * 64
        self.w1 = nn.Linear(cfg.n_embd, hidden, bias=False)
        self.w3 = nn.Linear(cfg.n_embd, hidden, bias=False)
        self.w2 = nn.Linear(hidden, cfg.n_embd, bias=False)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.w2(F.silu(self.w1(x)) * self.w3(x))


class Block(nn.Module):
    def __init__(self, cfg: ModelConfig):
        super().__init__()
        self.norm1 = RMSNorm(cfg.n_embd)
        self.attn = CausalSelfAttention(cfg)
        self.norm2 = RMSNorm(cfg.n_embd)
        self.mlp = SwiGLU(cfg)

    def forward(self, x, cos, sin):
        x = x + self.attn(self.norm1(x), cos, sin)
        x = x + self.mlp(self.norm2(x))
        return x


class GPT(nn.Module):
    def __init__(self, cfg: ModelConfig):
        super().__init__()
        self.cfg = cfg
        self.tok_emb = nn.Embedding(cfg.vocab_size, cfg.n_embd)
        self.blocks = nn.ModuleList([Block(cfg) for _ in range(cfg.n_layer)])
        self.norm = RMSNorm(cfg.n_embd)
        self.lm_head = nn.Linear(cfg.n_embd, cfg.vocab_size, bias=False)
        if cfg.tie_embeddings:
            self.lm_head.weight = self.tok_emb.weight

        self.apply(self._init_weights)
        # Scale residual projections per GPT-2 init.
        for name, p in self.named_parameters():
            if name.endswith("proj.weight") or name.endswith("w2.weight"):
                nn.init.normal_(p, mean=0.0, std=0.02 / math.sqrt(2 * cfg.n_layer))

        self._rope_cache = None

    def _init_weights(self, m):
        if isinstance(m, nn.Linear):
            nn.init.normal_(m.weight, mean=0.0, std=0.02)
            if m.bias is not None:
                nn.init.zeros_(m.bias)
        elif isinstance(m, nn.Embedding):
            nn.init.normal_(m.weight, mean=0.0, std=0.02)

    def num_params(self, non_embedding: bool = True) -> int:
        n = sum(p.numel() for p in self.parameters())
        if non_embedding and self.cfg.tie_embeddings:
            n -= self.tok_emb.weight.numel()
        return n

    def _rope(self, T: int, device, dtype):
        if (self._rope_cache is None
            or self._rope_cache[0].size(0) < T
            or self._rope_cache[0].device != device
            or self._rope_cache[0].dtype != dtype):
            self._rope_cache = build_rope_cache(
                self.cfg.block_size, self.cfg.head_dim, self.cfg.rope_base, device, dtype,
            )
        cos, sin = self._rope_cache
        return cos[:T], sin[:T]

    def forward(self, idx: torch.Tensor, targets: torch.Tensor | None = None):
        B, T = idx.shape
        assert T <= self.cfg.block_size, f"sequence length {T} > block_size {self.cfg.block_size}"

        x = self.tok_emb(idx)
        cos, sin = self._rope(T, x.device, x.dtype)
        for block in self.blocks:
            x = block(x, cos, sin)
        x = self.norm(x)

        if targets is None:
            logits = self.lm_head(x[:, [-1], :])
            return logits, None

        logits = self.lm_head(x)
        loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1), ignore_index=-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,
                 eos_id: int | None = None):
        for _ in range(max_new_tokens):
            idx_cond = idx if idx.size(1) <= self.cfg.block_size else idx[:, -self.cfg.block_size:]
            logits, _ = self(idx_cond)
            logits = logits[:, -1, :] / max(temperature, 1e-5)
            if top_k is not None:
                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)
            idx = torch.cat((idx, next_id), dim=1)
            if eos_id is not None and (next_id == eos_id).all():
                break
        return idx