List your company

Open-ended Video Question Answering on NExTQA (val)

28.9AccuracyFrameOracle
14.02817.88921.7525.611Oct 4, 2025

Evaluation results

MethodLinks
FrameOracleBackbone=VideoLLaMA3-7B, Frames=64→13.92025.1028.9
FrameOracleBackbone=LLaVA-Video-7B, Frames=64→13.92025.1028.8
FrameOracleBackbone=VideoLLaMA3-7B, Frames=16→10.42025.1028.3
FrameOracleBackbone=Qwen3-VL-8B, Frames=128→27.82025.1028.1
FrameOracleBackbone=LLaVA-Video-7B, Frames=16→10.42025.1027.8
VideoLLaMA3-7BFrames=162025.1027.8
LLaVA-Video-7BFrames=162025.1027.3
FrameOracleBackbone=Qwen3-VL-8B, Frames=32→20.92025.1026.6
FrameOracleBackbone=Qwen2.5-VL-3B, Frames=128→27.82025.1026
Qwen3-VL-8BFrames=322025.1026
FrameOracleBackbone=Qwen2.5-VL-3B, Frames=32→20.92025.1025.6
Qwen2.5-VL-3BFrames=322025.1025.1
FrameOracleBackbone=LLaVA-OneVision-7B, Frames=64→13.92025.1016.5
FrameOracleBackbone=LLaVA-OneVision-7B, Frames=16→10.42025.1016.1
LLaVA-OneVision-7BFrames=162025.1014.6