Add ce_loss metric and TriviaQA/NaturalQuestions tasks#520
Merged
OyvindTafjord merged 9 commits intomainfrom Apr 25, 2024
Merged
Add ce_loss metric and TriviaQA/NaturalQuestions tasks#520OyvindTafjord merged 9 commits intomainfrom
OyvindTafjord merged 9 commits intomainfrom
Conversation
epwalsh
approved these changes
Mar 27, 2024
This file contains hidden or 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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
With @yulinggu-cs we have added a new
ce_lossmetric which just scores the cross-entropy loss of the gold answer to tasks, which should have positive correlation with actual task performance.As examples of using this metric, we added these new tasks:
trivia_qa_wiki_ppl: TrivaQA 8k validation set (Wiki subset, used in Llama-2 paper)natural_qs_open_ppl: NaturalQuestions (nq_open validation subset, 3.6k questions, used in Llama papers)arc_easy_ppl: Sample example of adding a CE-loss version of existing task (this is somewhat inefficient, redoing part of the earlier evaluation, so should be optimized if we end up doing this for existing tasks)We log tasks with this metric to a different tab in wandb.