AI.info

Kang An

Explore Kang An on AI.info.

• onPanda: Efficient Annotation of On-Policy Alignment Data for LLMs and Agents via Token-Level Correction