[GitHub] neodooth commented on issue #3058: how to output loss

[gi...@git.apache.org]

neodooth commented on issue #3058: how to output loss
URL: https://github.com/apache/incubator-mxnet/issues/3058#issuecomment-322703436
 
 
   @LeonJWH 
   Hi, the links should be around the commit 83dc0b4d46383ef63b2731f12663aa947d96c2ad.
   https://github.com/apache/incubator-mxnet/blob/83dc0b4d46383ef63b2731f12663aa947d96c2ad/example/image-classification/train_model.py#L82
   https://github.com/apache/incubator-mxnet/blob/83dc0b4d46383ef63b2731f12663aa947d96c2ad/example/rcnn/rcnn/metric.py#L7
   https://github.com/apache/incubator-mxnet/blob/83dc0b4d46383ef63b2731f12663aa947d96c2ad/example/rcnn/tools/train_rpn.py#L78
   
   I haven't been using mxnet for quite a while and it looks like some major updates has taken place... but the ideas should still be the same, those metrics (for computing accuracy or loss) will be called when an iter is finished, and add some print there should do. I used to infer speed by reading the default logs and compute the intervals of the lines and iter numbers, but if you really want to print the speed, you might need to modify the code beneath the ```fit``` method; there must be a ```for``` or ```while``` somewhere to launch each iteration.
   
   The metrics thing is from my experience of running the rcnn example, I don't know if it's the same if you're running something else.
 
----------------------------------------------------------------
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:
users@infra.apache.org


With regards,
Apache Git Services