Hugging Face Papers ↗ · 2026-11-13
自我改进记忆实现持续递归自我改进
Media reports Hugging Face Papers Source language: Chinese
论文摘要(作者报告,未经本站复测):Harness 优化通常是在固定的策略模型和基准上进行研究的。
Full text
Continual recursive self-improvement lets harness optimizers reuse and revise prior edits across changing benchmarks or policy models by storing rationales, predicted and observed outcomes, and evaluation contexts in a self-improvement memory, improving final scores over continuation baselines.