Dataset v0.1.2 updates and feedback

Hi participants,

Feel free to leave any questions or feedback regarding v0.1.2 datasets here. We appreciate your participation and understanding!

Please note the following issue in v0.1.2 release:

  1. “ans_full” is empty in two interactions in the multi-turn data (crag-mm-2025/crag-mm-multi-turn-public · Datasets at Hugging Face).
    Please remove these two session_ids that include the affected interactions during local evaluation.
    • validation: 04d98259-27af-41b1-a7be-5798fd1b8e95
    • public_test: 695b4b5c-7c65-4f7b-8968-50fe10482a16

Hi. Please tell @yilun_jin8 and @jyotish to fix the leaderboard. (Details in discussion here). Having a working leaderboard will motivate all participants to work harder and create a better solution for you and Meta. Thank you.

1 Like