4 papers
StaRPO: Stability-Augmented Reinforcement Policy Optimization
Jinghan Zhang, Fengran Mo, Tharindu Cyril Weerasooriya +5
Reinforcement learning (RL) is effective in enhancing the accuracy of large language models in complex reasoning tasks. Existing RL policy optimization frameworks rely on final-ans…
Sim2Act: Robust Simulation-to-Decision Learning via Adversarial Calibration and Group-Relative Perturbation
Hongyu Cao, Jinghan Zhang, Kunpeng Liu +5
Simulation-to-decision learning enables safe policy training in digital environments without risking real-world deployment, and has become essential in mission-critical domains suc…
Data-Efficient Symbolic Regression via Foundation Model Distillation
Wangyang Ying, Jinghan Zhang, Haoyue Bai +5
Discovering interpretable mathematical equations from observed data (a.k.a. equation discovery or symbolic regression) is a cornerstone of scientific discovery, enabling transparen…
Entropy-based Exploration Conduction for Multi-step Reasoning
Jinghan Zhang, Xiting Wang, Fengran Mo +3
Multi-step processes via large language models (LLMs) have proven effective for solving complex reasoning tasks. However, the depth of exploration of the reasoning procedure can si…