5 papers
Learning with a Single Rollout via Monte Carlo Pass@k Critic
Fengdi Che, Yang Liu, Lei Yu +4
Estimating token-level advantages in reinforcement learning (RL) for language models remains challenging because scaling up episodic experience collection is expensive. The difficu…
Greed Is Learned: Visible Incentives as Reward-Hacking Triggers
Tong Che, Rui Wu
Deployed agents increasingly act with their reward proxy in view, such as a balance, score, or KPI dashboard. We show that reinforcement learning can make a policy \emph{addicted}…
Constitutional Value Potentials: reading and steering internal priority margins in language models
Tong Che, Rui Wu
A constitution tells a language model what to value, but little tells us whether it does. Adherence is judged from outputs, and output evidence is most fragile on value conflicts,…
Reference Feature Atlases for Mechanistic Auditing of Language Models
Rui Wu, Tong Che
Auditing a new language model usually means relearning and reinterpreting its internal features from scratch. We propose a reference feature atlas: a sparse feature library trained…
LLaMA-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning
Di Zhang, Jianbo Wu, Jingdi Lei +9
This paper presents an advanced mathematical problem-solving framework, LLaMA-Berry, for enhancing the mathematical reasoning ability of Large Language Models (LLMs). The framework…