List your company

Human Preference Evaluation on Long Videos 50 60s

30Win CountStream-R1
21.6823.842628.16May 5, 2026

Evaluation results

MethodLinks
Stream-R1Dimension=Dynamic Reasonableness, Basline=Reward Forcing2026.053031763
Stream-R1Dimension=Visual Quality & Aesthetics, Basline=Reward Forcing2026.052921960
Stream-R1Dimension=Overall Preference, Basline=Reward Forcing2026.052812157
Stream-R1Dimension=Temporal Consistency, Basline=Reward Forcing2026.052512451
Stream-R1Dimension=Text-Video Alignment, Basline=Reward Forcing2026.052291854.1