31 lines
954 B
Python
31 lines
954 B
Python
import numpy as np
|
|
import torch
|
|
from torch import nn
|
|
|
|
|
|
class PositionalEncoding(nn.Module):
|
|
|
|
def __init__(self, d_model, dropout=0.1, max_len=5000, batch_first=False):
|
|
super().__init__()
|
|
self.batch_first = batch_first
|
|
|
|
self.dropout = nn.Dropout(p=dropout)
|
|
|
|
pe = torch.zeros(max_len, d_model)
|
|
position = torch.arange(0, max_len, dtype=torch.float).unsqueeze(1)
|
|
div_term = torch.exp(torch.arange(
|
|
0, d_model, 2).float() * (-np.log(10000.0) / d_model))
|
|
pe[:, 0::2] = torch.sin(position * div_term)
|
|
pe[:, 1::2] = torch.cos(position * div_term)
|
|
pe = pe.unsqueeze(0).transpose(0, 1)
|
|
|
|
self.register_buffer("pe", pe)
|
|
|
|
def forward(self, x):
|
|
# not used in the final model
|
|
if self.batch_first:
|
|
x = x + self.pe.permute(1, 0, 2)[:, : x.shape[1], :]
|
|
else:
|
|
x = x + self.pe[: x.shape[0], :]
|
|
return self.dropout(x)
|