6 papers
Actor-Curator: Co-adaptive Curriculum Learning via Policy-Improvement Bandits for RL Post-Training
Zhengyao Gu, Jonathan Light, Raul Astudillo +7
Post-training large foundation models with reinforcement learning typically relies on massive and heterogeneous datasets, making effective curriculum learning both critical and cha…
ConstrainedSQL: Training LLMs for Text2SQL via Constrained Reinforcement Learning
Weiqin Chen, Nhan Huu Pham, Michael Robert Glass +4
Reinforcement learning (RL) has demonstrated significant promise in enhancing the reasoning capabilities of Text2SQL LLMs, especially with advanced algorithms such as GRPO and DAPO…
PAD-TRO: Projection-Augmented Diffusion for Direct Trajectory Optimization
Jushan Chen, Santiago Paternain
Recently, diffusion models have gained popularity and attention in trajectory optimization due to their capability of modeling multi-modal probability distributions. However, addre…
Filtering Learning Histories Enhances In-Context Reinforcement Learning
Weiqin Chen, Xinjie Zhang, Dharmashankar Subramanian +1
Transformer models (TMs) have exhibited remarkable in-context reinforcement learning (ICRL) capabilities, allowing them to generalize to and improve in previously unseen environmen…
Learning Closed-Loop Parametric Nash Equilibria of Multi-Agent Collaborative Field Coverage
Jushan Chen, Santiago Paternain
Multi-agent reinforcement learning is a challenging and active field of research due to the inherent nonstationary property and coupling between agents. A popular approach to model…
DISC: Dynamic Decomposition Improves LLM Inference Scaling
Jonathan Light, Wei Cheng, Benjamin Riviere +6
Inference scaling methods for LLMs often rely on decomposing problems into steps (or groups of tokens), followed by sampling and selecting the best next steps. However, these steps…