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
|
"""Transformer decoder module."""
from copy import deepcopy
from typing import Optional, Tuple, Type
from torch import nn, Tensor
from text_recognizer.networks.transformer.attention import Attention
from text_recognizer.networks.transformer.mlp import FeedForward
class DecoderBlock(nn.Module):
"""Decoder block."""
def __init__(
self,
self_attn: Attention,
norm: Type[nn.Module],
ff: FeedForward,
cross_attn: Optional[Attention] = None,
) -> None:
super().__init__()
self._layers = ("self_attn", "cross_attn", "ff")
self._blocks = self._build(self_attn, norm, ff, cross_attn)
def _build(
self,
self_attn: Attention,
norm: Type[nn.Module],
ff: FeedForward,
cross_attn: Optional[Attention],
) -> nn.ModuleDict:
return nn.ModuleDict(
{
self.layers[0]: nn.ModuleList([norm, self_attn]),
self.layers[1]: nn.ModuleList([deepcopy(norm), cross_attn]),
self.layers[2]: nn.ModuleList([deepcopy(norm), ff]),
}
)
def _apply(
self,
layer: str,
x: Tensor,
context: Optional[Tensor] = None,
input_mask: Optional[Tensor] = None,
context_mask: Optional[Tensor] = None,
) -> Tensor:
residual = x
norm_fn, layer_fn = self._blocks[layer]
if layer == "self_attn":
out = layer_fn(x=x, input_mask=input_mask)
elif layer == "cross_attn":
out = layer_fn(
x=x, context=context, input_mask=input_mask, context_mask=context_mask
)
else:
out = layer_fn(x)
out += residual
return norm_fn(out)
def forward(
self,
x: Tensor,
context: Optional[Tensor] = None,
input_mask: Optional[Tensor] = None,
context_mask: Optional[Tensor] = None,
) -> Tensor:
"""Applies decoder block on input signals."""
for layer in self._layers:
x = self._apply(
layer=layer,
x=x,
context=context,
input_mask=input_mask,
context_mask=context_mask,
)
return x
class Decoder:
"""Decoder Network."""
def __init__(self, depth: int, block: DecoderBlock) -> None:
self.depth = depth
self.has_pos_emb: bool = block.rotary_embedding is not None
self._block = nn.ModuleList([deepcopy(block) for _ in range(self.depth)])
def forward(
self,
x: Tensor,
context: Optional[Tensor] = None,
input_mask: Optional[Tensor] = None,
context_mask: Optional[Tensor] = None,
) -> Tensor:
"""Applies the network to the signals."""
for block in self._blocks:
x = block(
x=x, context=context, input_mask=input_mask, context_mask=context_mask
)
return x
|
