SHAPER：通过技能演化实现自进化具身智能体

AI 摘要

SHAPER 是一个无需训练的框架，通过环境回滚演化可复用技能和上下文代码工具包，提升具身智能体的性能，同时保持基础模型参数冻结。该框架在 VLABench 和 ESI-Bench 上进行了评估，并与纯执行、监督微调和测试时扩展基线进行了比较。结果表明，在模型训练成本高昂或不可用时，技能和工具包优化是自进化具身智能体的实用途径。

正文节选

Self-Evolving Embodied Agents via Skill-Harness Evolution Abstract SHAPER is a train-free framework that improves embodied agents by evolving reusable skills and a context-code harness around a frozen foundation model through environment rollouts. Embodied agents are increasingly built as systems around foundation models, where performance depends not only on model weights but also on the skills, context, action interfaces, and execution harness surrounding the model. While supervised fine-tunin

发布时间：—

抓取时间：2026-08-13 16:11

来源机构：Hugging Face