chrishkchris commented on pull request #765: URL: https://github.com/apache/singa/pull/765#issuecomment-657953375
> Hi @chrishkchris , after latest commit, testing softmax is ok. But a side note is that, some test are not correctly label when compiling with DNNL OFF, thus some test failed. is it the case? This PR fix two things: 1. c++ test: add back the kernal without DNNL: (i) if USE_DNNL=ON, use DNNL softmax kernal, if USE_DNNL=OFF, use the old kernal. Both can pass the test cases. The test was failed just because it did not have the kernal for the test 2. fix python test cases but those are not related to DNNL, for example, in opt_test.py: `sgd1.apply(w, w, g)` should be `sgd1.apply(w.name, w, g)` , the first argument is name ---------------------------------------------------------------- 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]
