HF blog·· 22 天前AI 评分34
Hugging Face 论文页：Co-Evolving Harnesses and Models——在策略修正让弱模型在模仿失败处追赶
Paper page - Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails
AI 导读
Hugging Face 上线论文页《Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails》，探讨在策略修正对较弱模型的提升作用。论文指出，在模仿学习失效的场景下，该方法能帮助弱模型追赶。
来源：HF blog · huggingface.co