NaN problem fix #3
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The problem was the sharpening function in memory.py which had a division with no guarantee that the denominator is not zero. I have tested it by starting to learn from a checkpoint which was very close to outputting NaNs (it always happened in a couple of hundred iterations) and with this fix it worked fine for 300k more iterations. Of course the result of the loss function changed a little at first, so this is not a 100% guarantee, but I'm confident this patch removes one possibility for the NaN problem.
camigord
requested changes
Nov 17, 2017
There was a problem hiding this comment.
Thanks a lot for the fix and sorry for the huge delay... I would like to approve the pull request, but I would like to keep the current README given that the problem may still not be solved. Could you send a new pull request with only the changes in memory.py?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The problem was the sharpening function in memory.py which had a division with no guarantee that the denominator is not zero.
I have tested it by starting to learn from a checkpoint which was very close to outputting NaNs (it always happened in a couple of hundred iterations) and with this fix it worked fine for 300k more iterations.
Of course the result of the loss function changed a little at first, so this is not a 100% guarantee, but I'm confident this patch removes one possibility for the NaN problem.