Fix Tokenization + misc fixes #354
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
Collaborator
hynky1999
commented
Oct 9, 2024
hynky1999
commented
- Fixes the info metric_aggregated default constructor. This was causing errror when one used multiple different metrics for one task (e.g pmi and token norm)
- Fixes pair tokenization:
- In case on pairwise tokenization the continuation no longer has added special tokens + it's simplified
- The non-pairwise tokenization now uses last token as continuation in case the last token gets merged
- Add pair tokenization to VLLM
- Changes multiprocessing to multiprocess which allow to use lambda functions and partial applications in task configs (before this would fail when dataset loading process was > 1 and multiprocessing was used)
…r non-pairwise, add pairwise to vllm, use multiprocess fordataset loading
clefourrier
reviewed
Oct 10, 2024
src/lighteval/models/model_config.py
Outdated
| @@ -226,6 +226,7 @@ class VLLMModelConfig: | |||
| multichoice_continuations_start_space: bool = ( | |||
| True # whether to add a space at the start of each continuation in multichoice generation | |||
| ) | |||
| pair_wise_tokenization: bool = False | |||
There was a problem hiding this comment.
pairwise is one single word -> pairwise_tokenization
clefourrier
reviewed
Oct 10, 2024
clefourrier
reviewed
Oct 10, 2024
clefourrier
reviewed
Oct 10, 2024
NathanHB
reviewed
Oct 10, 2024
Comment on lines
187
to
188
| self.tok_encode(context, add_special_tokens=self.add_special_tokens), | ||
| self.tok_encode(continuation, add_special_tokens=self.add_special_tokens), |
There was a problem hiding this comment.
mhh why do we want the special tokens for the continuation too ?
There was a problem hiding this comment.
Thanks for the catch I re-wrote it slightly based on discussion on slack
NathanHB
reviewed
Oct 10, 2024
NathanHB
approved these changes
Oct 10, 2024
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.