TY  - RPRT
TI  - Direct Self-Evolving Optimization: Evolving LLMs without Challenger Training
AU  - Yuyang Deng
AU  - Yu Wang
AU  - Jiayun Wang
PY  - 2026
UR  - https://arxiv.org/abs/2609.34279
ID  - 2609.34279
ER  -