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
|
"""Feedforward layer in transformer.
Stolen from lucidrains:
https://github.com/lucidrains/x-transformers/blob/main/x_transformers/x_transformers.py
"""
from typing import Optional
import torch.nn.functional as F
from torch import Tensor, nn
class GEGLU(nn.Module):
def __init__(self, dim_in: int, dim_out: int) -> None:
super().__init__()
self.fc = nn.Linear(dim_in, dim_out * 2)
def forward(self, x: Tensor) -> Tensor:
x, gate = self.fc(x).chunk(2, dim=-1)
return x * F.gelu(gate)
class FeedForward(nn.Module):
def __init__(
self,
dim: int,
dim_out: Optional[int] = None,
expansion_factor: int = 4,
glu: bool = True,
dropout_rate: float = 0.0,
) -> None:
super().__init__()
inner_dim = dim * expansion_factor
dim_out = dim_out if dim_out is not None else dim
in_projection = (
nn.Sequential(nn.Linear(dim, inner_dim), nn.GELU())
if not glu
else GEGLU(dim, inner_dim)
)
self.mlp = nn.Sequential(
in_projection, nn.Dropout(dropout_rate), nn.Linear(inner_dim, dim_out)
)
def forward(self, x: Tensor) -> Tensor:
return self.mlp(x)
|
