MineExplorer: Evaluating MLLM Agent Exploration in Minecraft

MineExplorer is a dataset for evaluating the open-world exploration capabilities of multimodal large language model agents in Minecraft. The dataset was created by researchers including Tianjie Ju, Yueqing Sun, and Zhuosheng Zhang. It was last updated on June 12, 2026.

Use Cases

• Benchmarking agent exploration performance based on the described open-world Minecraft environment.
• Evaluating multimodal reasoning and action generation in dynamic settings based on the dataset's purpose.
• Training or fine-tuning MLLMs for sustained embodied interaction based on the described agent tasks.

Strengths

• Dataset is focused on a specific, high-interest research area: evaluating MLLM agents in open-world environments.
• The description explicitly names the contributing authors and research institution (Meituan-Longcat).
• The dataset page was updated on June 12, 2026, suggesting recent maintenance.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Freshness

Last updated 2026-06-12 15:53:54; freshness should be verified.