论文解读MM-Conv: A Multimodal Dataset and Benchmark for Context-Aware Grounding in 3D Dialogue跨模态 | 6.5/10