DrugEncoder.LSTM

Click here to view source code.

class LSTM(num_embedding: int = 37, embedding_dim: int = 768, ft_dim: int = 768, dropout: float = 0.3,
bidirectional: bool = True, num_layers: int = 2)

LSTM can be used to encode SMILES of drugs.

PARAMETERS:

  • num_embedding (int, optional) - Number of embeddings. (default: 37)

  • embedding_dim (int, optional) - Dimension of embeddings. (default: 768)

  • ft_dim (int, optional) - Dimension of encoded SMILES. (default: 768)

  • dropout (float, optional) - Dropout rate of LSTM. (default: 0.3)

  • bidirectional (bool, optional) - Parameter bidirectional of torch.nn.LSTM. (default: True)

  • num_layers (int, optional) - Parameter num_layers of torch.nn.LSTM. (default: 2)

SHAPES:

  • input: Preprocessed SMILES [batch_size, seq_len]

  • output: Encoded SMILES [batch_size, ft_dim, seq_len]

forward(f: torch.Tensor)

  • f (torch.Tensor) - The input of LSTM.