Cornerstones or Stumbling Blocks? Deciphering the Rock Tokens in On-Policy Distillation 文章

ArXiv CS.CL2026-06-02NEWSen作者: Yuxuan Jiang, Runchao Li, Shubhashis Roy Dipta, Dawei Li, Zhao Yang

Cornerstones or Stumbling Blocks? Deciphering the Rock Tokens in On-Policy Distillation · 相关技术