Reinforcement Learning with Verifiable Rewards
Also known as RLVR
Overview
This field studies how to improve language and multimodal models by rewarding outputs whose correctness or task progress can be checked, for example with unit tests, mathematical answers, formal proofs, or simulated environments. Common approaches use policy-gradient methods such as PPO and GRPO, with outcome rewards supplemented by step-level feedback, rollout exploration, or replay to improve credit assignment and data efficiency. Work spans math and coding benchmarks as well as search, GUI, and other interactive agents. A central tension is that rewards must be scalable yet faithful: imperfect or narrow verifiers can be exploited, and a correct final answer does not by itself establish sound reasoning, visual grounding, or robust generalization.
Latest papers
738 papers69 papers in the last four weeks, down 49% on the four weeks before. 0.5% of all new papers.
X-Coder: Advancing Competitive Programming with Synthetic Tasks, Solutions, and Tests
GDEPO: Group Dual-dynamic and Equal-right Advantage Policy Optimization for Sample-Efficient Reinforcement Learning
ScalePRM: Training Process Reward Models by Scaling Verification Compute Without Ground Truth
Honesty over Accuracy: Trustworthy Language Models through Reinforced Hesitation
Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models
Group Contrastive Policy Optimization: Learning Beyond the Self-Exploration Limit with Golden Answers
Selective Expert Guidance for Effective and Diverse Exploration in Reinforcement Learning of LLMs
Reinforcement Learning with Verifiable yet Noisy Rewards under Imperfect Verifiers
Thinking-Free Policy Initialization Makes Distilled Reasoning Models More Effective and Efficient Reasoners
Clip-Low Increases Entropy and Clip-High Decreases Entropy in Reinforcement Learning of Large Language Models
Improving Sampling Efficiency in RLVR through Adaptive Rollout and Response Reuse
Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
Nested-ReFT: Efficient Reinforcement Learning for Large Language Model Fine-Tuning via Off-Policy Rollouts
Stabilizing Knowledge, Promoting Reasoning: Dual-Token Constraints for RLVR
The Invisible Leash: Why RLVR May or May Not Escape Its Origin
pass@1. Some RLVR models increase average token-level entropy while reducing final-answer entropy: their token choices become more uncertain on average even as their final answers become more concentrated. Together, these findings reveal an ``invisible leash'' in current RLVR practice: it can make existing successes more reliable while losing access to other verified solutions. Empirical support dynamics make this imbalance measurable and motivate exploration strategies that reach underrepresented solutions while retaining prior successes.