make add_unk optional and don't use it for ner
#2839
Merged
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Outdated suggestions cannot be applied.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.
The
<unk>token is a common practice to have a mechanism to add labels that are not common enough for training. This mainly relates to classification and basically simulates an "others" category.However, for NER, we have a different mechanism with the
O-tag, hence I don't think<unk>has any practical use.Using
<unk>for NER can lead to down-weighting the overall score when the NER-model is not trained long enough (random apperances of<unk>still happen. See (#2762 and #2761) and can break pipelines, that do not expect entities marked asunk(#2224).I propose making the
<unk>optional, per defaultTrue, but allow disableling it for all ner-tasks. As well as disableling it at all ner-related examples.