easyrec.blocks

easyrec.blocks.interaction

class easyrec.blocks.interaction.AFM(*args, **kwargs)

Bases: keras.engine.training.Model

Attentional factorization machine layer.

Parameters

• one_hot_feature_columns – List[CategoricalColumn] encodes one hot feature fields, such as sex_id.

• k – Dimension of the second-order weights.

call(inputs, training=None, mask=None)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters

• inputs – Input tensor, or dict/list/tuple of input tensors.

• training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

• mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.

class easyrec.blocks.interaction.FFM(*args, **kwargs)

Bases: keras.engine.training.Model

Field-aware factorization machine layer.

Parameters

• one_hot_feature_columns – List[CategoricalColumn] encodes one hot feature fields, such as sex_id.

• k – Dimension of the second-order weights.

call(inputs, *args, **kwargs)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters

• inputs – Input tensor, or dict/list/tuple of input tensors.

• training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

• mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.

class easyrec.blocks.interaction.FM(*args, **kwargs)

Bases: keras.engine.training.Model

Factorization machine layer using vector w and matrix v.

Parameters

• one_hot_feature_columns – List[CategoricalColumn] encodes one hot feature fields, such as sex_id.

• k – Dimension of the second-order weights.

call(inputs, *args, **kwargs)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters

• inputs – Input tensor, or dict/list/tuple of input tensors.

• training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

• mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.

class easyrec.blocks.interaction.NFM(*args, **kwargs)

Bases: keras.engine.training.Model

Neural factorization machine layer.

Parameters

• one_hot_feature_columns – List[CategoricalColumn] encodes one hot feature fields, such as sex_id.

• k – Dimension of the second-order weights.

• units_list – Dimension of fully connected stack outputs.

• activation – Activation to use.

call(inputs, training=None, mask=None)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters

• inputs – Input tensor, or dict/list/tuple of input tensors.

• training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

• mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.

easyrec.blocks.nn

class easyrec.blocks.nn.DenseBlock(*args, **kwargs)

Bases: keras.engine.training.Model

Multi-perception layer.

Parameters

• units_list – Dimension of fully connected stack outputs.

• activation – Activation to use.

call(inputs, training=None, mask=None)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters

• inputs – Input tensor, or dict/list/tuple of input tensors.

• training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

• mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.

class easyrec.blocks.nn.MultiHeadSelfAttention(*args, **kwargs)

Bases: keras.engine.training.Model

Multi-head self attention layer.

Parameters

• input_dimension – Dimension of input.

• qkv_dimension – Dimension of Query, Key and Value.

• num_heads – Number of heads.

• output_dimension – Dimension of final output.

• use_normalization – Whether to use normalization in Query * Key process.

call(inputs, training=None, mask=None)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters

• inputs – Input tensor, or dict/list/tuple of input tensors.

• training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

• mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.

class easyrec.blocks.nn.ResidualBlock(*args, **kwargs)

Bases: keras.engine.training.Model

Residual layer.

call(inputs, training=None, mask=None)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters

• inputs – Input tensor, or dict/list/tuple of input tensors.

• training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

• mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.

class easyrec.blocks.nn.SelfAttention(*args, **kwargs)

Bases: keras.engine.training.Model

Self attention layer.

Parameters

• input_dimension – Dimension of input.

• qkv_dimension – Dimension of Query, Key and Value.

• use_normalization – Whether to use normalization in Query * Key process.

call(inputs, training=None, mask=None)

Calls the model on new inputs.

In this case call just reapplies all ops in the graph to the new inputs (e.g. build a new computational graph from the provided inputs).

Note: This method should not be called directly. It is only meant to be overridden when subclassing tf.keras.Model. To call a model on an input, always use the __call__ method, i.e. model(inputs), which relies on the underlying call method.

Parameters

• inputs – Input tensor, or dict/list/tuple of input tensors.

• training – Boolean or boolean scalar tensor, indicating whether to run the Network in training mode or inference mode.

• mask – A mask or list of masks. A mask can be either a tensor or None (no mask).

Returns

A tensor if there is a single output, or a list of tensors if there are more than one outputs.