HiL-ResRL: A Model-Agnostic Finetuning Adapter via Human-in-the-loop Residual Reinforcement Learning

[system]

---

This is a companion discussion topic for the original entry at https://arxiv.org/abs/2606.22860