CogVLM2 icon indicating copy to clipboard operation
CogVLM2 copied to clipboard
verified publisher icon THUDM

Is there a plan to release the TQA dataset used to train CogVLM2-Video?

Open peiliu0408 opened this issue 1 year ago • 3 comments

as mentioned as title, Is there a plan to release the TQA dataset used to train CogVLM2-Video?

peiliu0408 avatar Sep 02 '24 11:09 peiliu0408

It is possible that this data will be released in the future, but the timing is not guaranteed. You can use CogVLM2-Video to generate your own TQA data.

huangshiyu13 avatar Sep 02 '24 12:09 huangshiyu13

Is there a video training script?

YoungjaeDev avatar Sep 03 '24 01:09 YoungjaeDev

Is there a video training script?

You can have a try with this fine-tuning code: https://github.com/modelscope/ms-swift/blob/main/docs/source_en/Multi-Modal/cogvlm2-video-best-practice.md

huangshiyu13 avatar Sep 03 '24 02:09 huangshiyu13