File size: 3,138 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 | """v3 variant: parallel attention+FFN residual so the 3-way sum is always odd (no ties).
Rationale: the v2 block is
x = sign(x + attn(x)) # values {-2, 0, 2}, 0 -> +1 (bias)
x = sign(x + ffn(x)) # values {-2, 0, 2}, 0 -> +1 (bias)
The sign-on-zero bias pushes every residual toward +1, compounds across 8 layers.
v3 block:
a = attn(x); f = ffn(x)
x_out = sign(x + a + f) # values {-3, -1, 1, 3}, never 0, no bias
Same ±1 invariant but strictly unbiased at the residual.
"""
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from model import (
sign_ste, sign_ste_clipped, BitLinearRaw, BitLinear,
BiAttention, BitFFN, BinaryEmbedding,
)
class BitBlockV3(nn.Module):
def __init__(self, d_model, n_heads, d_ff):
super().__init__()
self.attn = BiAttention(d_model, n_heads)
self.ffn = BitFFN(d_model, d_ff)
def forward(self, x):
a = self.attn(x)
f = self.ffn(x)
return sign_ste(x + a + f)
class BitLMv3(nn.Module):
def __init__(self, vocab_size=128, d_model=256, n_layers=8, n_heads=8, 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([
BitBlockV3(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
@torch.no_grad()
def generate(self, idx, max_new_tokens=200, temperature=1.0, top_k=None):
self.eval()
for _ in range(max_new_tokens):
idx_cond = idx[:, -self.max_seq_len:]
logits, _ = self(idx_cond)
logits = logits[:, -1, :] / max(temperature, 1e-5)
if top_k is not None:
v, _ = torch.topk(logits, top_k)
logits[logits < v[:, [-1]]] = -float('inf')
probs = F.softmax(logits, dim=-1)
nxt = torch.multinomial(probs, num_samples=1)
idx = torch.cat([idx, nxt], dim=1)
return idx
if __name__ == '__main__':
m = BitLMv3()
n = sum(p.numel() for p in m.parameters())
print(f"v3 params: {n:,} ({n/1e6:.2f}M)")
x = torch.randint(0, 128, (2, 64))
y = torch.randint(0, 128, (2, 64))
logits, loss = m(x, y)
print("logits:", logits.shape, "loss:", loss.item())
loss.backward()
print("backward OK")
|