MC-EvoBench: Benchmarking Self-Improvement in Minecraft Agents
MC-EvoBench: Auditing Experience-Driven Self-Improvement in Minecraft Agents
MC-EvoBench evaluates how Minecraft agents convert experience into transferable skills. It identifies gaps between controller and planner-based agents, showing limited capability transfer despite related experience.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- code: Available Artifact link Open source S6
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
equal source exposure, and early execution failures. We construct MC-EvoBench as a controlled Minecraft benchmark with a fixed catalog of 19 registered evolution chains, including 12 primary evaluation chains and seven calibration or probe chains, together with three graph-aligned paired probe families for transfer. The benchmark represents each task as a task-state graph, defines stage-level environment verifiers, a
Open source S3
Code and data are available at https://github.com/xzw-ustc/mc-bench-open-source
Open source S6
Source：arXiv Robotics — research abstracts · arxiv.org