DivScene: Benchmarking LVLMs for Object Navigation with Diverse Scenes and Objects MMLongBench: Benchmarking Long-Context Vision-Language Models Effectively and Thoroughly
