This repository has been archived by the owner on Nov 22, 2022. It is now read-only.
Fix "calling median on empty tensor" issue in MR #1322
Closed
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
Summary: Language model metrics reporter occassionally throws "median on empty tensor", e.g: f182865817 why? We used to check the cross entropy tensor != 0 before calling mean/median on it, to exclude the calculation of the padding positions which is just 0, which possibly results in an empty tensor. This diff checks the tensor posisions against pad_index instead of using 0.0:a batch can have close to 0 CE, or the predict word index happened to be pad_index. Also a failed safe is added if the tensor[target != self.pad_index] is indeed empty. Differential Revision: D21081502 fbshipit-source-id: 685705897bd350b1ae3af6fae1e84725e045bb70
facebook-github-bot
added
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D21081502 |
|
This pull request has been merged in 7db6094. |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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.
Summary:
Language model metrics reporter occassionally throws "median on empty tensor", e.g: f182865817
why?
We used to check the cross entropy tensor != 0 before calling mean/median on it, to exclude the calculation of the padding positions which is just 0, which possibly results in an empty tensor.
This diff checks the tensor posisions against pad_index instead of using 0.0:a batch can have close to 0 CE, or the predict word index happened to be pad_index.
Also a failed safe is added if the tensor[target != self.pad_index] is indeed empty.
Differential Revision: D21081502