chrishkchris commented on issue #554: [bug]batchnorm running mean var not updated URL: https://github.com/apache/singa/pull/554#issuecomment-546778720 Yes, I see why the running mean is not updated. This is because running_mean = running_mean * bnh.factor + mean * (1 - bnh.factor); It generates a new tensor to be stored in the variable running mean, since it is NOT in place operation. this is why the original running mean is not updated So if we use in place operation the Original running mean tensor can be updated.
---------------------------------------------------------------- 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: us...@infra.apache.org With regards, Apache Git Services