Fix a potential random generated related nan issue caused by 0/0 #1
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
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Outdated suggestions cannot be applied.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.
In the update_dict method in https://github.com/cwlkr/torchvahadane/blob/main/torchvahadane/dict_learning.py#L100, when the atomic norm is too small, the corresponding atom is re-initialized with tensor.normal_.
However, if all numbers drawn are negative, the following in-place clamp will yield a zero vector, and in the in-place normalization
of
dictionary[:, k] /= dictionary[:, k].norm()a nan vector will be generated and this will invalidate the whole computation procedure.Herein two layers of protection is added:
(1) make the re-initialized atom non-negative by adding abs()
(2) add a small eps
torch.finfo(torch.float32).epsto the norm so the denominator (norm) will always be non-zero.