17 papers
TRACE: A Unified Rollout Budget Allocation Framework for Efficient Agentic Reinforcement Learning
Heming Zou, Qi Wang, Yun Qu +9
Reinforcement learning with verifiable rewards (RLVR) is a promising approach for enhancing reasoning and agentic behavior in large language models. However, rollout-intensive poli…
RLVR without Ineffective Samples: Group Prioritized Off-Policy Optimization for LLM Reasoning
Yixiu Mao, Yun Qu, Qi Wang +2
Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful paradigm for enhancing the reasoning capabilities of large language models (LLMs). However, its effe…
Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex
Yun Qu, Qi Wang, Yixiu Mao +11
Reinforcement learning with verifiable rewards (RLVR) has become a standard approach for large language models (LLMs) post-training to incentivize reasoning capacity. Among existin…
Dynamics-Predictive Sampling for Active RL Finetuning of Large Reasoning Models
Yixiu Mao, Yun Qu, Qi Wang +2
Reinforcement learning (RL) finetuning has become a key technique for enhancing the reasoning abilities of large language models (LLMs). However, its effectiveness critically depen…
Small Generalizable Prompt Predictive Models Can Steer Efficient RL Post-Training of Large Reasoning Models
Yun Qu, Qi Wang, Yixiu Mao +8
Reinforcement learning enhances the reasoning capabilities of large language models but often involves high computational costs due to rollout-intensive optimization. Online prompt…
Adaptive Neighborhood-Constrained Q Learning for Offline Reinforcement Learning
Yixiu Mao, Yun Qu, Qi Wang +1
Offline reinforcement learning (RL) suffers from extrapolation errors induced by out-of-distribution (OOD) actions. To address this, offline RL algorithms typically impose constrai…
Utility-Diversity Aware Online Batch Selection for LLM Supervised Fine-tuning
Heming Zou, Yixiu Mao, Yun Qu +2
Supervised fine-tuning (SFT) is a commonly used technique to adapt large language models (LLMs) to downstream tasks. In practice, SFT on a full dataset is computationally expensive…
VAO: Validation-Aligned Optimization for Cross-Task Generative Auto-Bidding
Yiqin Lv, Zhiyu Mou, Miao Xu +9
Generative auto-bidding has demonstrated strong performance in online advertising, yet it often suffers from data scarcity in small-scale settings with limited advertiser participa…
Enhancing Generative Auto-bidding with Offline Reward Evaluation and Policy Search
Zhiyu Mou, Yiqin Lv, Miao Xu +9
Auto-bidding is a critical tool for advertisers to improve advertising performance. Recent progress has demonstrated that AI-Generated Bidding (AIGB), which learns a conditional ge…
Can Prompt Difficulty be Online Predicted for Accelerating RL Finetuning of Reasoning Models?
Yun Qu, Qi Wang, Yixiu Mao +3
Recent advances have witnessed the effectiveness of reinforcement learning (RL) finetuning in enhancing the reasoning capabilities of large language models (LLMs). The optimization…
Fast and Robust: Task Sampling with Posterior and Diversity Synergies for Adaptive Decision-Makers in Randomized Environments
Yun Qu, Qi Cheems Wang, Yixiu Mao +2
Task robust adaptation is a long-standing pursuit in sequential decision-making. Some risk-averse strategies, e.g., the conditional value-at-risk principle, are incorporated in dom…
Latent Reward: LLM-Empowered Credit Assignment in Episodic Reinforcement Learning
Yun Qu, Yuhang Jiang, Boyuan Wang +4
Reinforcement learning (RL) often encounters delayed and sparse feedback in real-world applications, even with only episodic rewards. Previous approaches have made some progress in…
Model Predictive Task Sampling for Efficient and Robust Adaptation
Qi Wang, Zehao Xiao, Yixiu Mao +4
Foundation models have revolutionized general-purpose problem-solving, offering rapid task adaptation through pretraining, meta-training, and finetuning. Recent crucial advances in…
Doubly Mild Generalization for Offline Reinforcement Learning
Yixiu Mao, Qi Wang, Yun Qu +2
Offline Reinforcement Learning (RL) suffers from the extrapolation error and value overestimation. From a generalization perspective, this issue can be attributed to the over-gener…
Offline Reinforcement Learning with OOD State Correction and OOD Action Suppression
Yixiu Mao, Qi Wang, Chen Chen +2
In offline reinforcement learning (RL), addressing the out-of-distribution (OOD) action issue has been a focus, but we argue that there exists an OOD state issue that also impairs…
Stop Wandering, Find the Keys: LLMs Discriminate Key States for Efficient Multi-Agent Exploration
Yun Qu, Boyuan Wang, Yuhang Jiang +7
With expansive state-action spaces, efficient multi-agent exploration remains a longstanding challenge in reinforcement learning. Although pursuing novelty, diversity, or uncertain…
Robust Fast Adaptation from Adversarially Explicit Task Distribution Generation
Cheems Wang, Yiqin Lv, Yixiu Mao +3
Meta-learning is a practical learning paradigm to transfer skills across tasks from a few examples. Nevertheless, the existence of task distribution shifts tends to weaken meta-lea…