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
|
import torch
import torch.nn as nn
import transformers
def ltr_mask(seq_len: int) -> torch.Tensor:
mask = torch.ones((seq_len, seq_len), dtype=torch.bool)
return torch.tril(mask, diagonal=-1)
def rtl_mask(seq_len: int) -> torch.Tensor:
return ltr_mask(seq_len).T
def add_attn_hooks(model: transformers.BertModel, text_direction: str) -> None:
"""
Forces bidirectional `model` into a unidirectional one based on `direction`.
Adds hooks to `model`'s self-attention blocks, in-place.
Args:
model: only implemented for BERT models right now
text_direction: one of "ltr" or "rtl"
"""
assert text_direction.lower() in ("ltr", "rtl")
mask_func = ltr_mask if text_direction.lower() == "ltr" else rtl_mask
model.register_buffer("attn_mask", mask_func(model.config.max_position_embeddings).to(model.device))
def attn_hook(attn_module: nn.Module, args: tuple, kwargs: dict):
"""
Assuming https://github.com/huggingface/transformers/blob/33868a057c02f0368ba63bd1edb746be38fe3d90/src/transformers/models/bert/modeling_bert.py#L515
so no `kwargs` and `attention_mask` is second positional arg.
Uses nonlocal `model.attn_mask` to save memory.
"""
assert not kwargs
args = list(args)
assert args[1].size()[-2:] == model.attn_mask.size(), f"{args[1].size()=} {model.attn_mask.size()=}"
args[1] = model.attn_mask
return tuple(args), kwargs
for name, module in model.named_modules():
if isinstance(module, transformers.models.bert.modeling_bert.BertSelfAttention):
module._forward_hooks.clear() # in case we run multiple times
module.register_forward_pre_hook(attn_hook, with_kwargs=True)
|
