Contextual Rollout Bandits: A Neural Scheduling Framework for Efficient Reinforcement Learning with Verifiable Rewards
Reinforcement Learning with Verifiable Rewards (RLVR) is an effective paradigm for improving the reasoning capabilities of large language models. However, existing RLVR methods utilize rollouts in an…
Read the full articleYou might also wanna read
Correlation-Aware Contextual Bandits with Surrogate Rewards for LLM Routing
arXiv:2607.09015v1 Announce Type: new Abstract: We study contextual bandit problems with correlated arms and access to surrogate reward sign
When Implausible Tokens Get Reinforced: Tail-Aware Credit Calibration for LLM Reinforcement Learning
arXiv:2607.07976v1 Announce Type: new Abstract: Reinforcement learning (RL) has achieved remarkable success in enhancing the reasoning capab
Adaptive LLM Routing Using Contextual Bandits and Shared Embedding Space
Large Language Models (LLMs) have revolutionized natural language processing, but their varying capabilities and costs pose challenges in pr
arxiv.org·10mo ago
Rethinking Reinforcement Learning for Language Models: The SAO Approach
Single-rollout Asynchronous Optimization (SAO) offers a new path for more stable and effective reinforcement learning in large language mode
Learning More from Less: Reinforcement Learning from Hindsight
arXiv:2607.09042v1 Announce Type: new Abstract: Reinforcement learning (RL) is increasingly used to post-train vision-language-action (VLA)
Understanding Schedule-Free Methods in Nonconvex Optimization: Rate Guarantees and Escaping Saddles
arXiv:2607.09167v1 Announce Type: new Abstract: Schedule-Free methods have attracted growing interest for alleviating the burden of designin
Comments
Sign in to join the conversation.
No comments yet. Be the first.