espnet.nets.pytorch_backend.transformer.embedding.ScaledPositionalEncoding

source

class espnet.nets.pytorch_backend.transformer.embedding.ScaledPositionalEncoding(d_model, dropout_rate, max_len=5000)

Bases: PositionalEncoding

Scaled positional encoding module.

See Sec. 3.2 https://arxiv.org/abs/1809.08895

• Parameters:
  • d_model (int) – Embedding dimension.
  • dropout_rate (float) – Dropout rate.
  • max_len (int) – Maximum input length.

Initialize class.

forward(x)

Add positional encoding.

• Parameters:x (torch.Tensor) – Input tensor (batch, time, *).
• Returns: Encoded tensor (batch, time, *).
• Return type: torch.Tensor

reset_parameters()

Reset parameters.