朱橹

icon indicating an email [email protected]

icon company University of Electronic Science and Technology of China icon location student of University of Electronic Science and Technology of China

Results 4 issues of 朱橹

Update basic_utils.py

增加注释

Update basic_utils.py

增加注释

How to fine-tune based on the weights of LanguageBind/Video-LLaVA-7B？

In the original finetune.sh script, it was divided into model_name_or_path and pretrain_mm_mlp_adapter, representing the paths to the language model and the projector, respectively. However, in LanguageBind/Video-LLaVA-7B, the weights of all...

MLVU performance does not match with the paper llava-video

1 comment

In VIDEO INSTRUCTION TUNING WITH SYNTHETIC DATA , the performance of LLaVA-Video-7B on MLVU is 70.8.However i can only get 67.0. I do not know what went wrong and could...