text-video-alignment-evaluation

【效率与协作】Evaluate text-to-video alignment through fine-grained semantic understanding via multi-agent question generation and knowledge-augmented answering. Generate 12,000 atomic yes/no questions from 2,000 prompts across 10 evaluation categories, achieving 58.47 correlation with human judgment.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准