List your company
ResearchFormerly WizWand

Video Question Answering on SSV2-QA LLaVA-Hound out-of-domain (test)

37.9AccuracyMM1.5-Video
18.6623.65528.6533.645Sep 30, 2024
Updated 8mo ago

Evaluation results

MethodLinks
MM1.5-VideoModel Scale=7B, Protocol=SFT2024.0937.9
LLAVA-HOUND-SFTModel Scale=7B2024.0935.4
MM1.5-VideoModel Scale=3B, Protocol=SFT2024.0935.2
MM1.5-VideoModel Scale=1B, Protocol=SFT2024.0934
MM1.5-VideoModel Scale=7B, Protocol=Training-free2024.0930.4
MM1.5-VideoModel Scale=3B, Protocol=Training-free2024.0928.2
MM1.5-VideoModel Scale=1B, Protocol=Training-free2024.0927.2
Video-LLaVAModel Scale=7B2024.0924.3
LLaMA-VIDModel Scale=7B2024.0922.2
Chat-UniViModel Scale=7B2024.0920.6
Video-ChatGPTModel Scale=7B2024.0919.4