A Modified Energy Function is used and the params are defined here. Reference: Online and Linear-Time Attention by Enforcing Monotonic Ali

call model

Compute convolution module. Args: inputs : Input tensor (#batch, time, channels). Returns: tensor: Output te

Take in and process masked source/target sequences. Args: src: the sequence to the encoder (required). tgt: the sequen

Pass the input through the endocder layers in turn. Args: src: the sequnce to the encoder (required). mask: the mask f

Pass the inputs (and mask) through the decoder layer in turn. Args: tgt: the sequence to the decoder (required). memor

Pass the input through the encoder layer. Args: src: the sequence to the encoder layer (required). mask: the mask for

Pass the inputs (and mask) through the decoder layer. Args: tgt: the sequence to the decoder layer (required). memory:

Take in and process masked source sequences. Args: src: the sequence to the encoder (required). src_mask: the additive

Pass the input through the endocder layers in turn. Args: src: the sequnce to the encoder (required). mask: the mask f

Pass the input through the encoder layer. Args: src: the sequence to the encoder layer (required). mask: the mask for

This is where the layer's logic lives.

call function

call function

call function in keras

Attention layer.

Args: attn_inputs (tuple) : it contains 2 params: value, shape: [batch, x_steps, eunits] value_le

Args: attn_inputs (tuple) : it contains 2 params: value, shape: [batch, x_steps, eunits] value_le

Take in and process masked source/target sequences. Args: src: the sequence to the encoder (required). tgt: the seque

Pass the input through the endocder layers in turn. Args: src: the sequnce to the encoder (required). mask: the mask

Pass the inputs (and mask) through the decoder layer in turn. Args: tgt: the sequence to the decoder (required). memo

Pass the input through the encoder layer. Args: src: the sequence to the encoder layer (required). mask: the mask for

Pass the inputs (and mask) through the decoder layer. Args: tgt: the sequence to the decoder layer (required). memory

call function

Runs vanilla LSTM Cell and applies zoneout.

This is where the layer's logic lives.

call function

Compute convolution module. Args: x : Input tensor (#batch, time, channels). Returns: tensor: Output tensor

Take in and process masked source/target sequences. Args: src: the sequence to the encoder (required). tgt: the seque

Pass the input through the endocder layers in turn. Args: src: the sequnce to the encoder (required). mask: the mask

Pass the inputs (and mask) through the decoder layer in turn. Args: tgt: the sequence to the decoder (required). memo