ciyongch commented on a change in pull request #17187: [MKLDNN] Support channel wise quantization
for FullyConnected
URL: https://github.com/apache/incubator-mxnet/pull/17187#discussion_r361887846
##########
File path: python/mxnet/contrib/quantization.py
##########
@@ -109,6 +110,9 @@ def _quantize_symbol(sym, ctx, excluded_symbols=None, excluded_operators=None,
The quantized destination type for input data.
quantize_mode: str
The mode that quantization pass to apply.
+ quantize_granularity: str
+ The granularity of quantization, currently supports 'tensor-wise' and 'channel-wise'
Review comment:
I thought `tensor` and `channel` is not that descriptive for the argument. There's also
`per-axis`/`per-channel` and `per-tensor` naming, which is similar to `channel-wise` and `tensor-wise`.
Better to keep this?
----------------------------------------------------------------
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:
users@infra.apache.org
With regards,
Apache Git Services
|