marcoabreu commented on a change in pull request #16408: Add MXNet Ops for fast
multihead attention
URL: https://github.com/apache/incubator-mxnet/pull/16408#discussion_r341186884
##########
File path: src/operator/contrib/transformer-inl.h
##########
@@ -34,6 +34,19 @@
namespace mxnet {
namespace op {
+struct InterleavedMatMulParam : public dmlc::Parameter<InterleavedMatMulParam>
{
+ int heads;
+ bool bwd_ignore_zero_init;
+ DMLC_DECLARE_PARAMETER(InterleavedMatMulParam) {
+ DMLC_DECLARE_FIELD(heads)
+ .describe("Set number of heads");
+ DMLC_DECLARE_FIELD(bwd_ignore_zero_init)
+ .describe("Make backward pass ignore AddTo and not init to 0. "
+ " /!\\ Only enable with MXNET_ENABLE_EXEC_ADDTO fonctionality")
Review comment:
Do you envision that to happen in regular use cases? In case you don't want
to raise an exception, we could still print a warning in case the flag is not
set
----------------------------------------------------------------
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
