List your company

Hallucination Mitigation on VRIPT-HAL

52.1F1 ScoreVideoChat-R1
23.530.92538.3545.775Dec 21, 2025

Evaluation results

MethodLinks
VideoChat-R1Mitigation Method=SmartSight2025.1252.13.7
Qwen2.5-VL-7BMitigation Method=SmartSight2025.12502.1
Gemini 1.5 ProMitigation Method=Base (Greedy Decoding)2025.1249.3—
VideoChat-R1Mitigation Method=Base (Greedy Decoding)2025.1248.4—
Qwen2.5-VL-7BMitigation Method=Base (Greedy Decoding)2025.1247.9—
Claude 3.5 SonnetMitigation Method=Base (Greedy Decoding)2025.1244.6—
InternVL3-8BMitigation Method=SmartSight2025.1234.13.7
LLaVA-OneVision-7BMitigation Method=SmartSight2025.1231.45.6
InternVL3-8BMitigation Method=Base (Greedy Decoding)2025.1230.4—
LLaVA-OneVision-7BMitigation Method=Base (Greedy Decoding)2025.1225.8—
Video-LLaVAMitigation Method=SmartSight2025.1225.71.1
Video-LLaVAMitigation Method=Base (Greedy Decoding)2025.1224.6—