The paper introduces $ ext{RLR}^3$, a novel framework that extends verifiable rewards in Reinforcement Learning to handle partially verifiable, multi-criteria vision-language tasks by integrating robu…

The paper introduces a verifier-fuzzing framework to detect and analyze failure modes in Reinforcement Learning with Verifiable Rewards (RLVR) where bugs in the reward verifier can be exploited by the…

The paper demonstrates that using Reinforcement Learning from Verifiable Rewards (RLVR) significantly improves small language models' functional correctness in code generation, particularly when combi…

This paper introduces a novel backdoor attack (ACB) against Reinforcement Learning with Verifiable Rewards (RLVR), demonstrating that poisoning the training data can implant a backdoor that significan…

The paper introduces REFT, a novel method that diversifies rollouts by sampling the first token after the reasoning marker, significantly improving performance in Reinforcement Learning with Verifiabl…

The paper proposes Credit-Attenuated Privileged Feedback (CAPF), a training-time mechanism that uses verifier-side information to guide LLM search agents, significantly improving their performance on…

This paper introduces Isospectral Optimization (ISO), a new optimization framework for reinforcement learning with verifiable rewards (RLVR), which inherits the base model's weight spectra while optim…

The paper introduces Atomic Decomposition and Recombination (ADR), a novel framework that generates genuinely novel and challenging verifiable code tasks, significantly improving the scalability of Re…

IRDS introduces a novel data selection method that uses a verifier-coupled sparse autoencoder framework to efficiently select high-quality Reinforcement Learning with Verifiable Rewards (RLVR) trainin…

EchoRL proposes a lightweight module to exploit valuable learning signals from advantage-degenerated rollouts in Reinforcement Learning with Verifiable Rewards (RLVR), significantly improving LLM post…

The paper introduces Off-Context GRPO, a method for reinforcement learning with verifiable rewards that uses guided rollouts and importance-corrected objectives to improve reasoning in large language…

The paper analyzes the performance of an annealed softmax policy in a Bayesian bandit setting, proving that under specific prior conditions, it achieves near-optimal regret rates by effectively sampli…

The paper introduces a reinforcement learning framework, inspired by AlphaZero, to automate and improve the proof search process within the Tamarin protocol analysis tool, resulting in shorter and mor…

This paper investigates the importance of including reasoning traces in the responses of Large Language Models during both training and inference for Neural Machine Translation, and shows that it posi…

CARE-RL introduces a framework combining protocol-aware reward generation and capability-aware optimization to effectively mitigate cross-domain conflicts in multi-domain reinforcement learning for LL…

The paper introduces Cross-Model Entropy (CME), a novel label-free reward signal that uses an independent verifier model to assess the quality of a generator's output, significantly improving LLM perf…

This paper proposes a new imitation learning algorithm called DistIL that uses distributional feedback to improve policy improvement and regret guarantees.