TaoLv commented on a change in pull request #14749: [Mxnet-1397] Support
symbolic api for requantize and dequantize
URL: https://github.com/apache/incubator-mxnet/pull/14749#discussion_r277167425
##########
File path: src/operator/quantization/dequantize.cc
##########
@@ -84,6 +84,10 @@ by keep zero centered for the quantized value:
.set_attr_parser(ParamParser<DequantizeParam>)
.set_num_inputs(3)
.set_num_outputs(1)
+.set_attr<nnvm::FListInputNames>("FListInputNames",
+ [](const NodeAttrs& attrs) {
+ return std::vector<std::string>{"data", "min_range", "max_range"};
Review comment:
If these names will be exposed to front end users, I hope they can align
with other quantization operators. In quantized convolution and quantized FC, I
see they are `min_data` and `max_data`.
----------------------------------------------------------------
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
