TY  - RPRT
TI  - A Hessian-Free Actor-Critic Algorithm for Bi-Level Reinforcement Learning with Applications to LLM Fine-Tuning
AU  - Sihan Zeng
AU  - Sujay Bhatt
AU  - Sumitra Ganesh
AU  - Alec Koppel
PY  - 2026
UR  - https://arxiv.org/abs/2601.16399
ID  - 2601.16399
ER  -