You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for the great work. The paper mentions MMC4 during pre-training. As for the supervised finetuning stage, I'm curious if the authors have considered using the MIMIC-IT dataset from Otter which is an interleaved image text dataset. Would adding interleaved image text during supervised finetuning potentially further boost the overall performance?
The text was updated successfully, but these errors were encountered:
Thanks for the great work. The paper mentions MMC4 during pre-training. As for the supervised finetuning stage, I'm curious if the authors have considered using the MIMIC-IT dataset from Otter which is an interleaved image text dataset. Would adding interleaved image text during supervised finetuning potentially further boost the overall performance?
The text was updated successfully, but these errors were encountered: