7 papers
Beyond Solvability: Task Learnability as a Static Prior for LLM RL Post-Training
Ting Zhou, Zhenqing Ling, Daoyuan Chen +4
Reinforcement learning (RL) has become a central post-training paradigm for eliciting reasoning capabilities in large language models, yet uniform task sampling allocates compute w…
CDR-Bench: Evaluating Faithful Execution of Compositional, Order-Sensitive Data Refinement Recipes
Yuchen Huang, Xiang Li, Zhenqing Ling +5
Data refinement involves executing multi-step recipes over evolving text states, where both composition and execution order of processing operators determine the outcome. While exi…
GEOALIGN: Geometric Rollout Curation for Robust LLM Reinforcement Learning
Ting Zhou, Zhenqing Ling, Yiyang Zhao +2
Online reinforcement learning is widely used to align large language models (LLMs) with reward signals, yet training can be unstable under noisy or misspecified rewards. We identif…
BOTS: A Unified Framework for Bayesian Online Task Selection in LLM Reinforcement Finetuning
Qianli Shen, Daoyuan Chen, Yilun Huang +4
Reinforcement finetuning (RFT) is a key technique for aligning Large Language Models (LLMs) with human preferences and enhancing reasoning, yet its effectiveness is highly sensitiv…
MindGYM: What Matters in Question Synthesis for Thinking-Centric Fine-Tuning?
Zhe Xu, Daoyuan Chen, Zhenqing Ling +2
Large foundation models face challenges in acquiring transferable, structured thinking abilities, especially when supervised with rigid templates or crowd-annotated instruction dat…
Diversity as a Reward: Fine-Tuning LLMs on a Mixture of Domain-Undetermined Data
Zhenqing Ling, Daoyuan Chen, Liuyi Yao +3
Fine-tuning large language models (LLMs) using diverse datasets is crucial for enhancing their overall performance across various domains. In practical scenarios, existing methods…