fix: Make deduplication in PairClassificationEvaluator stable #1315
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
Samoed
reviewed
Oct 24, 2024
| task_name=self.task_name, | ||
| prompt_type=PromptType.query, |
There was a problem hiding this comment.
PairClassification shouldn't have prompt_type
There was a problem hiding this comment.
I think that it should be removed totally from evaluator. Also can you run some tasks to check scores?
tsirif
force-pushed
the
fix/pairwise_stable_dedup
branch
from
18e0fdd
to
d4c8e72
Compare
There was a problem hiding this comment.
Thanks for the PR - looked it over. I Agree with @Samoed suggestions here. Otherwise, not much to add.
|
@tsirif would love to get this merged in do you have the time to work on it (otherwise it might be worth closing it or allow others to finish it up) |
|
I will do it tomorrow!
Στις Δευ 11 Νοε 2024, 4:21 π.μ. ο χρήστης Kenneth Enevoldsen <
***@***.***> έγραψε:
… @tsirif <https://github.com/tsirif> would love to get this merged in do
you have the time to work on it?
—
Reply to this email directly, view it on GitHub
<#1315 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AA7FKAIR54Y6XD7USWKNBUT2ABZLHAVCNFSM6AAAAABQP5NXCOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINRXGYZTINZWGU>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
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Outdated suggestions cannot be applied.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.
What: Change the way sentence deduplication is made in
PairClassificationEvaluatorHow: Use the same staticmethod as in
RerankingEvaluator:_encode_unique_textsWhy: Compared to
list(set(sentences1 + sentences2)), the function_encode_unique_textsis definitely stable in the order that the deduplicated sentences are generated. This is crucial if the implementation ofmodel.encodeis parallelizing the work via DDP, in which case the splitting of sentences to chunks across DDP workers expects that the same exact list of sentences (order too!) is given to each call ofmodel.encode...Checklist
make test.make lint.