rnn_cell.RNNCell Class Reference

Inheritance diagram for rnn_cell.RNNCell:

Public Member Functions
def	__init__ (self, name)
def	scope (self, name)
def	apply_over_sequence (self, model, inputs, seq_lengths, initial_states, outputs_with_grads=None)
def	apply (self, model, input_t, seq_lengths, states, timestep)
def	prepare_input (self, model, input_blob)
def	get_state_names (self)

Public Attributes
	name
	recompute_blobs

Detailed Description

Base class for writing recurrent / stateful operations.

One needs to implement 3 methods: _apply, prepare_input and get_state_names.
As a result base class will provice apply_over_sequence method, which
allows you to apply recurrent operations over a sequence of any length.

Definition at line 20 of file rnn_cell.py.

Member Function Documentation

get_state_names()

def rnn_cell.RNNCell.get_state_names

(

self

)

Return the names of the recurrent states.
It's required by apply_over_sequence method in order to allocate
recurrent states for all steps with meaningful names.

Definition at line 111 of file rnn_cell.py.

prepare_input()

def rnn_cell.RNNCell.prepare_input	(		self,
			model,
			input_blob
	)

If some operations in _apply method depend only on the input,
not on recurrent states, they could be computed in advance.

model: CNNModelHelper object new operators would be added to

input_blob: either the whole input sequence with shape
(sequence_length, batch_size, input_dim) or a single input with shape
(1, batch_size, input_dim).

Definition at line 98 of file rnn_cell.py.

---

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

• caffe2/python/rnn_cell.py