Qianchu Liu
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
Recursive Reward Improvement via Adversarial Meta-Prompt Evolution
According to the abstract excerpt, the authors propose RRI, a method for self-improvement loops on open-ended tasks. An adversary writes a deliberately useless answer meant to game a model-written rubric, so that weak rubrics can be caught before a solver trains on them.