torchfm package

torchfm.layer

class torchfm.layer.AttentionalFactorizationMachine(embed_dim, attn_size, dropouts)[source]
forward(x)[source]
Parameters

x – Float tensor of size (batch_size, num_fields, embed_dim)

class torchfm.layer.CompressedInteractionNetwork(input_dim, cross_layer_sizes, split_half=True)[source]
forward(x)[source]
Parameters

x – Float tensor of size (batch_size, num_fields, embed_dim)

class torchfm.layer.CrossNetwork(input_dim, num_layers)[source]
forward(x)[source]
Parameters

x – Float tensor of size (batch_size, num_fields, embed_dim)

class torchfm.layer.FactorizationMachine(reduce_sum=True)[source]
forward(x)[source]
Parameters

x – Float tensor of size (batch_size, num_fields, embed_dim)

class torchfm.layer.FeaturesEmbedding(field_dims, embed_dim)[source]
forward(x)[source]
Parameters

x – Long tensor of size (batch_size, num_fields)

class torchfm.layer.FeaturesLinear(field_dims, output_dim=1)[source]
forward(x)[source]
Parameters

x – Long tensor of size (batch_size, num_fields)

class torchfm.layer.FieldAwareFactorizationMachine(field_dims, embed_dim)[source]
forward(x)[source]
Parameters

x – Long tensor of size (batch_size, num_fields)

class torchfm.layer.InnerProductNetwork[source]
forward(x)[source]
Parameters

x – Float tensor of size (batch_size, num_fields, embed_dim)

class torchfm.layer.MultiLayerPerceptron(input_dim, embed_dims, dropout, output_layer=True)[source]
forward(x)[source]
Parameters

x – Float tensor of size (batch_size, num_fields, embed_dim)

class torchfm.layer.OuterProductNetwork(num_fields, embed_dim, kernel_type='mat')[source]
forward(x)[source]
Parameters

x – Float tensor of size (batch_size, num_fields, embed_dim)