1
MineExplorer: Evaluating Open-World Exploration of MLLM Agents in Minecraft
用Minecraft沙盒世界,给多模态大模型智能体的开放探索能力来场硬核测试
arXiv:2605.30931v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) have shown strong capabilities in perception, reasoning, and …