List your company
ResearchFormerly WizWand

Video Question Answering on MSVD-QA LLaVA-Hound out-of-domain (test)

67.2AccuracyMM1.5-Video
32.77641.71350.6559.587Sep 30, 2024
Updated 8mo ago

Evaluation results

MethodLinks
MM1.5-VideoModel Scale=7B, Protocol=SFT2024.0967.2
MM1.5-VideoModel Scale=3B, Protocol=SFT2024.0965.2
MM1.5-VideoModel Scale=1B, Protocol=SFT2024.0965
LLAVA-HOUND-SFTModel Scale=7B2024.0962.2
MM1.5-VideoModel Scale=7B, Protocol=Training-free2024.0952.9
MM1.5-VideoModel Scale=3B, Protocol=Training-free2024.0951.1
MM1.5-VideoModel Scale=1B, Protocol=Training-free2024.0949.8
Video-LLaVAModel Scale=7B2024.0939.5
Chat-UniViModel Scale=7B2024.0935.6
Video-ChatGPTModel Scale=7B2024.0934.1
LLaMA-VIDModel Scale=7B2024.0934.1