pengzhao-intel commented on issue #13697: [MKLDNN] Enable signed int8 support for convolution. URL: https://github.com/apache/incubator-mxnet/pull/13697#issuecomment-450779824 @KellenSunderland Happy New Year! Really thanks for your review and providing amount of very good comments. Regarding the documents, the current quantization flow is very flexible. On the other hand, it's complicated. Definitely, we need systematic documents to answer the FAQ and disclose some technical and architecture details for the advantage users. In addition to the descriptions in the script, we plan to write more info/cases, also compare INT8 and UINT8 results, in the [Model Quantization with Intel® MKL-DNN](https://github.com/apache/incubator-mxnet/tree/master/example/quantization#1) by another PR. Very appriciate for further suggestions and help us improve the documentation. Regarding the API changes, I keep a more open attitude. Becasue the whole quantization flow is under the developments with the `Experimental` lable. Thus, if the API changes or improvements are good for the users or functionality, it will be acceptable. Actually, we will still make the API as stable as possible. Feel free to correct me :) Any other comments either for this PR or the quantization flow are highly welcome 👍
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on 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
