List your company
ResearchFormerly WizWand

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

59.3AccuracyMM1.5-Video
23.62832.88942.1551.411Sep 30, 2024
Updated 8mo ago

Evaluation results

MethodLinks
MM1.5-VideoModel Scale=7B, Protocol=SFT2024.0959.3
MM1.5-VideoModel Scale=3B, Protocol=SFT2024.0957.2
MM1.5-VideoModel Scale=1B, Protocol=SFT2024.0955.3
LLAVA-HOUND-SFTModel Scale=7B2024.0952.6
MM1.5-VideoModel Scale=7B, Protocol=Training-free2024.0948.1
MM1.5-VideoModel Scale=3B, Protocol=Training-free2024.0946
MM1.5-VideoModel Scale=1B, Protocol=Training-free2024.0943.3
Video-LLaVAModel Scale=7B2024.0930.8
Chat-UniViModel Scale=7B2024.0925.9
Video-ChatGPTModel Scale=7B2024.0925.7
LLaMA-VIDModel Scale=7B2024.0925