stephenrawls commented on a change in pull request #14208: Add support for fast
variable-length LSTM
URL: https://github.com/apache/incubator-mxnet/pull/14208#discussion_r280663388
##########
File path: src/operator/rnn.cc
##########
@@ -106,25 +128,33 @@ static bool RNNShape(const nnvm::NodeAttrs& attrs,
out_shape->push_back(cellStateShape);
}
}
+
return true;
}
static bool RNNType(const nnvm::NodeAttrs& attrs,
std::vector<int> *in_type,
std::vector<int> *out_type) {
const RNNParam& param_ = nnvm::get<RNNParam>(attrs.parsed);
- if (param_.mode == rnn_enum::kLstm) {
- CHECK_EQ(in_type->size(), 4U);
- } else {
- CHECK_EQ(in_type->size(), 3U);
- }
+
+ CHECK_EQ(in_type->size(), GetNumInputArguments(param_));
+
+ size_t seq_len_input_idx = rnn_enum::kSequenceLength;
+ if (param_.mode != rnn_enum::kLstm)
+ --seq_len_input_idx;
Review comment:
Done. I think I got all of them.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
