MolecularDiffusion.modules.layers.common¶

Classes¶

`MLP`	Multi-layer Perceptron.
`SinusoidalPositionEmbedding`	Positional embedding based on sine and cosine functions, proposed in Attention Is All You Need.
`SinusoidsEmbeddingNew`

Module Contents¶

class MolecularDiffusion.modules.layers.common.MLP(input_dim, hidden_dims, short_cut=False, batch_norm=False, activation='relu', dropout=0)¶

Bases: torch.nn.Module

Multi-layer Perceptron. Note there is no batch normalization, activation or dropout in the last layer.

Parameters:

input_dim (int) – input dimension
hidden_dim (list of int) – hidden dimensions
short_cut (bool, optional) – use short cut or not
batch_norm (bool, optional) – apply batch normalization or not
activation (str or function, optional) – activation function
dropout (float, optional) – dropout rate

forward(input)¶

dims¶

layers¶

short_cut = False¶

class MolecularDiffusion.modules.layers.common.SinusoidalPositionEmbedding(output_dim)¶

Bases: torch.nn.Module

Positional embedding based on sine and cosine functions, proposed in Attention Is All You Need.

Parameters:: output_dim (int) – output dimension

forward(input)¶

class MolecularDiffusion.modules.layers.common.SinusoidsEmbeddingNew(max_res=15.0, min_res=15.0 / 2000.0, div_factor=4)¶

Bases: torch.nn.Module

forward(x)¶

dim¶

frequencies¶

n_frequencies¶