Valentin Churavy
Valentin Churavy
I agree that it is a bit weird and it looks like you are hitting a local optima over and over again. Is your dataset balanced and are you using...
What was your batch size if I may ask. Also which version of MXNet proper are you using and which optimizer? -V On Mon, 17 Apr 2017 at 21:08 Petter...
@pluskid Do you think that we are hitting a particular cornercase here or is that just a side effect of using SGD based methods? As far as I understand it...
@nowozin I am working on this and the non functional PR is at #16
Yes with NDArray it should be possible, I just have been quite busy the last two weeks and haven't gotten any coding done. Getting this to work is on the...
I made progress today, but it still has some way to go, but help is always welcome.
1. Yes that is the branch I have been working on 2. The branch has a small example in there and the best thing to do is to execute that...
I hope that I can make another push for this during the coming week. There are a few issues I have a handle on and one that I am not...
Depends on #39 and see my work in #16. The ability to define custom loss functions requires layers written in python and is what inspired me to work on it....
CustomOp support is a longer project see #166 . If anybody is interested in having this helping with #173 is a good place to start.