bgawrych commented on code in PR #21002:
URL: https://github.com/apache/incubator-mxnet/pull/21002#discussion_r921954676
##########
src/operator/nn/dnnl/dnnl_fully_connected-inl.h:
##########
@@ -100,7 +94,7 @@ class FCInputIndex {
const bool has_bias = !full_param.default_param.no_bias;
const bool quantized = dnnl_param.quantized;
const bool sum_input_quantized =
- quantized && dnnl_param.with_sum && !dnnl_param.enable_float_output;
+ quantized && dnnl_param.with_sum &&
!dnnl_param.enabled_float_output.has_value();
Review Comment:
it is enough as we expect to have not quantized sum input with these
parameters - currently enabled_float_output if has any value it can only be
float32 - and even if in future bfloat will be added situation should be the
same as for floats
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
