Caffe2 - Python API
A deep learning, cross platform ML framework
Public Member Functions | Public Attributes | List of all members
caffe2.python.rnn_cell.AttentionCell Class Reference
Inheritance diagram for caffe2.python.rnn_cell.AttentionCell:
caffe2.python.rnn_cell.RNNCell caffe2.python.rnn_cell.LSTMWithAttentionCell caffe2.python.rnn_cell.MILSTMWithAttentionCell

Public Member Functions

def __init__ (self, encoder_output_dim, encoder_outputs, encoder_lengths, decoder_cell, decoder_state_dim, attention_type, weighted_encoder_outputs, attention_memory_optimization, kwargs)
def get_attention_weights (self)
def prepare_input (self, model, input_blob)
def build_initial_coverage (self, model)
def get_state_names (self)
def get_output_dim (self)
def get_output_state_index (self)
- Public Member Functions inherited from caffe2.python.rnn_cell.RNNCell
def __init__ (self, name=None, forward_only=False, initializer=None)
def initializer (self)
def initializer (self, value)
def scope (self, name)
def apply_over_sequence (self, model, inputs, seq_lengths=None, initial_states=None, outputs_with_grads=None)
def apply (self, model, input_t, seq_lengths, states, timestep)
def apply_override (self, model, input_t, seq_lengths, timestep, extra_inputs=None)
def prepare_input (self, model, input_blob)
def get_output_state_index (self)
def get_state_names (self)
def get_state_names_override (self)
def get_output_dim (self)

Public Attributes

- Public Attributes inherited from caffe2.python.rnn_cell.RNNCell

Detailed Description

Definition at line 1122 of file

Member Function Documentation

def caffe2.python.rnn_cell.AttentionCell.build_initial_coverage (   self,
initial_coverage is always zeros of shape [encoder_length],
which shape must be determined programmatically dureing network

This method also sets self.coverage_weights, a separate transform
of encoder_outputs which is used to determine coverage contribution
tp attention.

Definition at line 1301 of file

The documentation for this class was generated from the following file: