File size: 4,397 Bytes
4754707 | 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 | """v57_fast: drop-in speed-up of BitLMv57 using the Triton BitLinear kernel.
Math identical to model_v57.BitLMv57. Only swaps `BitLinear` → `FastBitLinear`
inside the attention and FFN. Strict ±1 everywhere.
"""
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from model import sign_ste, BinaryEmbedding
from model_v16 import gumbel_hard_attention
from bit_kernel import FastBitLinear
class FastBitFFN(nn.Module):
def __init__(self, d_model, d_ff):
super().__init__()
self.gate = FastBitLinear(d_model, d_ff, binarize_input=True)
self.up = FastBitLinear(d_model, d_ff, binarize_input=True)
self.down = FastBitLinear(d_ff, d_model, binarize_input=True)
def forward(self, x):
return self.down(self.gate(x) * self.up(x))
class FastManyHeadAttention(nn.Module):
def __init__(self, d_model, n_heads, slope_groups=8):
super().__init__()
assert d_model % n_heads == 0
self.d_model = d_model
self.n_heads = n_heads
self.head_dim = d_model // n_heads
self.q_proj = FastBitLinear(d_model, d_model)
self.k_proj = FastBitLinear(d_model, d_model)
self.v_proj = FastBitLinear(d_model, d_model)
self.o_proj = FastBitLinear(d_model, d_model)
heads_per_group = max(1, n_heads // slope_groups)
slopes = []
for g in range(slope_groups):
slopes.extend([1 << g] * heads_per_group)
while len(slopes) < n_heads:
slopes.append(1 << (slope_groups - 1))
self.register_buffer('alibi_slopes_int', torch.tensor(slopes[:n_heads], dtype=torch.long))
def forward(self, x):
B, T, D = x.shape
H, Dh = self.n_heads, self.head_dim
Q = self.q_proj(x).view(B, T, H, Dh).transpose(1, 2)
K = self.k_proj(x).view(B, T, H, Dh).transpose(1, 2)
V = self.v_proj(x).view(B, T, H, Dh).transpose(1, 2)
scores = torch.matmul(Q, K.transpose(-2, -1))
pos = torch.arange(T, device=x.device)
dist = (pos.unsqueeze(0) - pos.unsqueeze(1)).abs()
alibi = self.alibi_slopes_int.view(1, H, 1, 1).to(scores.dtype) \
* dist.view(1, 1, T, T).to(scores.dtype)
scores = scores - alibi
mask = torch.triu(torch.ones(T, T, device=x.device, dtype=torch.bool), diagonal=1)
A = gumbel_hard_attention(scores, mask=mask)
O = torch.matmul(A, V)
O = O.transpose(1, 2).contiguous().view(B, T, D)
return self.o_proj(O)
class FastBlock(nn.Module):
def __init__(self, d_model, n_heads, d_ff):
super().__init__()
self.attn = FastManyHeadAttention(d_model, n_heads)
self.ffn = FastBitFFN(d_model, d_ff)
def forward(self, x):
return sign_ste(x + self.attn(x) + self.ffn(x))
class BitLMv57Fast(nn.Module):
def __init__(self, vocab_size=128, d_model=1024, n_layers=8, n_heads=32,
d_ff=512, max_seq_len=256):
super().__init__()
self.vocab_size = vocab_size
self.d_model = d_model
self.n_layers = n_layers
self.max_seq_len = max_seq_len
self.embed = BinaryEmbedding(vocab_size, d_model)
self.blocks = nn.ModuleList([
FastBlock(d_model, n_heads, d_ff) for _ in range(n_layers)
])
self.out_codebook = nn.Parameter(torch.randn(vocab_size, d_model) * 0.02)
self.logit_scale = nn.Parameter(torch.tensor(1.0 / math.sqrt(d_model)))
self.out_bias = nn.Parameter(torch.zeros(vocab_size))
def forward(self, idx, targets=None):
x = self.embed(idx)
for blk in self.blocks:
x = blk(x)
W_out = sign_ste(self.out_codebook)
scores = torch.matmul(x, W_out.t())
logits = scores * self.logit_scale + self.out_bias
loss = None
if targets is not None:
loss = F.cross_entropy(logits.view(-1, self.vocab_size), targets.view(-1))
return logits, loss
if __name__ == '__main__':
from model_v16 import set_gumbel_tau
set_gumbel_tau(0.5)
m = BitLMv57Fast().cuda()
n = sum(p.numel() for p in m.parameters())
print(f'BitLMv57Fast: {n:,} ({n/1e6:.3f}M)')
x = torch.randint(0, 128, (2, 64), device='cuda')
y = torch.randint(0, 128, (2, 64), device='cuda')
logits, loss = m(x, y)
loss.backward()
print(f'loss={loss.item():.3f}, backward OK')
|