RNN descriptor for backward propagation. More...
#include <dnnl.hpp>
Public Member Functions | |
desc (prop_kind aprop_kind, algorithm activation, rnn_direction direction, const memory::desc &src_layer_desc, const memory::desc &src_iter_desc, const memory::desc &weights_layer_desc, const memory::desc &weights_iter_desc, const memory::desc &bias_desc, const memory::desc &dst_layer_desc, const memory::desc &dst_iter_desc, const memory::desc &diff_src_layer_desc, const memory::desc &diff_src_iter_desc, const memory::desc &diff_weights_layer_desc, const memory::desc &diff_weights_iter_desc, const memory::desc &diff_bias_desc, const memory::desc &diff_dst_layer_desc, const memory::desc &diff_dst_iter_desc, rnn_flags flags=rnn_flags::undef, float alpha=0.0f, float beta=0.0f) | |
Initializes an RNN descriptor for backward propagation using prop_kind , activation , direction , and memory descriptors. More... | |
RNN descriptor for backward propagation.
|
inline |
Initializes an RNN descriptor for backward propagation using prop_kind
, activation
, direction
, and memory descriptors.
alpha
, beta
and flags
are parameters to the RNN descriptor. If activation
is eltwise_relu, alpha
represents the negative slope. beta
and flags
are currently ignored.
format_kind
.src_iter_desc
(simultaneously with diff_src_iter_desc
), bias_desc
(simultaneously with diff_bias_desc
), and dst_iter_desc
(simultaneously with diff_src_iter_desc
) are allowed point to a zero memory descriptor, which would indicate that the RNN primitive should not use them and consider them to be zero values.