HarmVideoBench: Benchmarking Harmful Video Understanding in Large Multimodal Models figure
AlphaXiv 中文概览(可滚动查看)