8 papers
Agents' Last Exam
Yiyou Sun, Xinyang Han, Weichen Zhang +306
Recent AI systems have achieved strong results on a wide range of benchmarks, yet these gains have not translated into economically meaningful deployment across many professional d…
PhysProver: Advancing Automatic Theorem Proving for Physics
Hanning Zhang, Ruida Wang, Rui Pan +3
The combination of verifiable languages and LLMs has significantly influenced both the mathematical and computer science communities because it provides a rigorous foundation for t…
OpenGenAlign: A Preference Dataset and Benchmark for Trustworthy Reward Modeling in Open-Ended, Long-Context Generation
Hanning Zhang, Juntong Song, Juno Zhu +3
Reward Modeling is critical in evaluating and improving the generation of Large Language Models (LLMs). While numerous recent works have shown its feasibility in improving safety,…
Entropy-Regularized Process Reward Model
Hanning Zhang, Pengcheng Wang, Shizhe Diao +6
Large language models (LLMs) have shown promise in performing complex multi-step reasoning, yet they continue to struggle with mathematical reasoning, often making systematic error…
DuaShepherd: Integrating Stepwise Correctness and Potential Rewards for Mathematical Reasoning
Yuanhao Wu, Juntong Song, Hanning Zhang +2
In this paper, we propose DuaShepherd, a novel reward modeling framework that integrates two complementary reward signals, correctness and potential, to enhance the mathematical re…
Understanding Overadaptation in Supervised Fine-Tuning: The Role of Ensemble Methods
Yifan Hao, Xingyuan Pan, Hanning Zhang +3
Supervised fine-tuning (SFT) on domain-specific data is the dominant approach for adapting foundation models to specialized tasks. However, it has been observed that SFT models ten…