List your company
ResearchFormerly WizWand

Video Question Answering on WebVid-QA LLaVA-Hound in-domain (test)

71.5AccuracyMM1.5-Video
35.6244.93554.2563.565Sep 30, 2024
Updated 8mo ago

Evaluation results

MethodLinks
MM1.5-VideoModel Scale=7B, Protocol=SFT2024.0971.5
MM1.5-VideoModel Scale=3B, Protocol=SFT2024.0971.1
MM1.5-VideoModel Scale=1B, Protocol=SFT2024.0968.7
LLAVA-HOUND-SFTModel Scale=7B2024.0966.8
MM1.5-VideoModel Scale=3B, Protocol=Training-free2024.0958.5
MM1.5-VideoModel Scale=7B, Protocol=Training-free2024.0958.5
MM1.5-VideoModel Scale=1B, Protocol=Training-free2024.0955.8
Video-LLaVAModel Scale=7B2024.0942.5
Chat-UniViModel Scale=7B2024.0940.1
Video-ChatGPTModel Scale=7B2024.0938.9
LLaMA-VIDModel Scale=7B2024.0937