feat: add EgoLifeQA benchmark - #1498
Open
lorenzozanee wants to merge 2 commits into
Open
Conversation
Integrate EgoLifeQA (CVPR 2025) long-context life-oriented video QA benchmark: 500 4-way multiple_choice questions from lmms-lab/EgoLife (EgoLifeQA_A1_JAKE.json) with 30s egocentric clip retrieval via media_resolver. Closes EvolvingLMMs-Lab#1214.
Integrate EgoLifeQA (CVPR 2025) long-context life-oriented video QA benchmark: 500 4-way multiple_choice questions from lmms-lab/EgoLife (EgoLifeQA_A1_JAKE.json) with 30s egocentric clip retrieval via media_resolver. Closes EvolvingLMMs-Lab#1214.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
lmms_eval/tasks/egolifeqawithmultiple_choiceloglikelihood evaluationquery_timeviamedia_resolverIn scope
lmms_eval/tasks/egolifeqa/_default_template_yaml— dataset config (hf://datasets/lmms-lab/EgoLife,multiple_choice)lmms_eval/tasks/egolifeqa/egolifeqa.yaml— task include wrapperlmms_eval/tasks/egolifeqa/utils.py—doc_to_visual/text/choice/targetwith flexible answer/option handling and video lookuplmms_eval/tasks/egolifeqa/README.md— usage and dataset notesOut of scope
docs/advanced/current_tasks.md(can be added in follow-up)Validation
python -m py_compile lmms_eval/tasks/egolifeqa/utils.py| sample size:N=1| key metrics:compile| result:passuv run ruff check lmms_eval/tasks/egolifeqa/utils.py| sample size:N=1| key metrics:lint| result:passpython /tmp/test_egolifeqa_proof.py| sample size:N=5| key metrics:doc_to_choice/target/text/visual + yaml| result:passRisk / Compatibility
--tasks egolifeqahf://datasets/lmms-lab/EgoLife; videos resolved lazily viaEGOLIFEQA_VIDEO_DIRor HF cacheType of Change
Fixes #1214