Reinforcement Learning with Verifiable Rewards

Also known as RLVR

Ask about this topic

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

CardsList

692 papers137 papers in the last four weeks, up 111% on the four weeks before. 0.8% of all new papers.

Feb 24, 2026

Overconfident Errors Need Stronger Correction: Asymmetric Confidence Penalties for Reinforcement Learning

Reinforcement Learning with Verifiable Rewards (RLVR) has become the leading paradigm for enhancing reasoning in Large Language Models (LLMs). However, standard RLVR algorithms suffer from a well-documented pathology: while they improve Pass@1 accuracy through sharpened sampling, they simultaneously narrow the model's reasoning boundary and reduce generation diversity. We identify a root cause that existing methods overlook: the uniform penalization of errors. Current approaches -- whether data-filtering methods that select prompts by difficulty, or advantage normalization schemes -- treat all incorrect rollouts within a group identically. We show that this uniformity allows overconfident errors (incorrect reasoning paths that the RL process has spuriously reinforced) to persist and monopolize probability mass, ultimately suppressing valid exploratory trajectories. To address this, we propose the Asymmetric Confidence-aware Error Penalty (ACE). ACE introduces a per-rollout confidence shift metric, c_i = log(pi_theta(y_i|x) / pi_ref(y_i|x)), to dynamically modulate negative advantages. Theoretically, we demonstrate that ACE's gradient can be decomposed into the gradient of a selective regularizer restricted to overconfident errors, plus a well-characterized residual that partially moderates the regularizer's strength. We conduct extensive experiments fine-tuning Qwen2.5-Math-7B, Qwen3-8B-Base, and Llama-3.1-8B-Instruct on the DAPO-Math-17K dataset using GRPO and DAPO within the VERL framework. Evaluated on MATH-500 and AIME 2025, ACE composes seamlessly with existing methods and consistently improves the full Pass@k spectrum across all three model families and benchmarks.

Yuanda Xu, Hejian Sang, Zhengze Zhou +2

LinkedIn Corporation, CA, USA

RL for Language Model ReasoningReinforcement Learning with Verifiable Rewards
AskPDF
Feb 24, 2026

Buffer Matters: Unleashing the Power of Off-Policy Reinforcement Learning in Large Language Model Reasoning

Traditional on-policy Reinforcement Learning with Verifiable Rewards (RLVR) frameworks suffer from experience waste and reward homogeneity, which directly hinders learning efficiency on difficult samples during large language models post-training. In this paper, we introduce Batch Adaptation Policy Optimization (BAPO), an off-policy RLVR framework to improve the data efficiency in large language models post-training. It dynamically selects training batches by re-evaluating historically difficult samples and reusing high-quality ones, while holding a lower bound guarantee for policy improvement. Extensive experiments further demonstrate that BAPO achieves an average 12.5% improvement over GRPO across mathematics, planning, and visual reasoning tasks. Crucially, BAPO successfully resolves 40.7% of problems that base models consistently fail to solve.

Xu Wan, Yansheng Wang, Wenqi Huang +1

Zhejiang University · Bytedance Seed Robotics · Peking University +1

RL for Language Model ReasoningRL Post-TrainingReinforcement Learning with Verifiable RewardsOff-Policy RLExperience Replay
AskPDF
Feb 23, 2026

ReSyn: Autonomously Scaling Synthetic Environments for Reasoning Models

Reinforcement learning with verifiable rewards (RLVR) has emerged as a promising approach for training reasoning language models (RLMs) by leveraging supervision from verifiers. Although verifier implementation is easier than solution annotation for many tasks, existing synthetic data generation methods remain largely solution-centric, while verifier-based methods rely on a few hand-crafted procedural environments. In this work, we scale RLVR by introducing ReSyn, a pipeline that generates diverse reasoning environments equipped with instance generators and verifiers, covering tasks such as constraint satisfaction, algorithmic puzzles, and spatial reasoning. A Qwen2.5-7B-Instruct model trained with RL on ReSyn data achieves consistent gains across reasoning benchmarks and out-of-domain math benchmarks, including a 27% relative improvement on the challenging BBEH benchmark. Ablations show that verifier-based supervision and increased task diversity both contribute significantly, providing empirical evidence that generating reasoning environments at scale can enhance reasoning abilities in RLMs

Andre He, Nathaniel Weir, Kaj Bostrom +4

Carnegie Mellon University · Amazon Web Services

Synthetic Data Generation for Language ModelsRL for Language Model ReasoningReinforcement Learning with Verifiable Rewards
AskPDF
Feb 23, 2026

DSDR: Dual-Scale Diversity Regularization for Exploration in LLM Reasoning

Reinforcement learning with verifiers (RLVR) is a central paradigm for improving large language model (LLM) reasoning, yet existing methods often suffer from limited exploration. Policies tend to collapse onto a few reasoning patterns and prematurely stop deep exploration, while conventional entropy regularization introduces only local stochasticity and fails to induce meaningful path-level diversity, leading to weak and unstable learning signals in group-based policy optimization. We propose DSDR, a Dual-Scale Diversity Regularization reinforcement learning framework that decomposes diversity in LLM reasoning into global and coupling components. Globally, DSDR promotes diversity among correct reasoning trajectories to explore distinct solution modes. Locally, it applies a length-invariant, token-level entropy regularization restricted to correct trajectories, preventing entropy collapse within each mode while preserving correctness. The two scales are coupled through a global-to-local allocation mechanism that emphasizes local regularization for more distinctive correct trajectories. We provide theoretical support showing that DSDR preserves optimal correctness under bounded regularization, sustains informative learning signals in group-based optimization, and yields a principled global-to-local coupling rule. Experiments on multiple reasoning benchmarks demonstrate consistent improvements in accuracy and pass@k, highlighting the importance of dual-scale diversity for deep exploration in RLVR. Code is available at https://github.com/SUSTechBruce/DSDR.

Zhongwei Wan, Yun Shen, Zhihao Dou +9

The Ohio State University, USA · Case Western Reserve University, USA · The Chinese University of Hong Kong, Hong Kong +4

Reinforcement LearningRL for Language Model ReasoningReinforcement Learning with Verifiable Rewards
AskPDF
Feb 22, 2026

How to Allocate, How to Learn? Dynamic Rollout Allocation and Advantage Modulation for Policy Optimization

Reinforcement Learning with Verifiable Rewards (RLVR) has proven effective for Large Language Model (LLM) reasoning, yet current methods face key challenges in resource allocation and policy optimization dynamics: (i) uniform rollout allocation ignores gradient variance heterogeneity across problems, and (ii) the softmax policy structure causes gradient attenuation for high-confidence correct actions, while excessive gradient updates may destabilize training. Therefore, we propose DynaMO, a theoretically-grounded dual-pronged optimization framework. At the sequence level, we prove that uniform allocation is suboptimal and derive variance-minimizing allocation from the first principle, establishing Bernoulli variance as a computable proxy for gradient informativeness. At the token level, we develop gradient-aware advantage modulation grounded in theoretical analysis of gradient magnitude bounds. Our framework compensates for gradient attenuation of high-confidence correct actions while utilizing entropy changes as computable indicators to stabilize excessive update magnitudes. Extensive experiments conducted on a diverse range of mathematical reasoning benchmarks demonstrate consistent improvements over strong RLVR baselines. Our implementation is available at: https://github.com/GithubX-F/DynaMO-RL.

Yangyi Fang, Jiaye Lin, Xiaoliang Fu +6

Meituan · Tsinghua University · Fudan University +1

Reinforcement LearningRL for Language Model ReasoningReinforcement Learning with Verifiable Rewards
AskPDF
Feb 22, 2026

Controllable Exploration in Hybrid-Policy RLVR for Multi-Modal Reasoning

Reinforcement Learning with verifiable rewards (RLVR) has emerged as a primary learning paradigm for enhancing the reasoning capabilities of multi-modal large language models (MLLMs). However, during RL training, the enormous state space of MLLM and sparse rewards often leads to entropy collapse, policy degradation, or over-exploitation of suboptimal behaviors. This necessitates an exploration strategy that maintains productive stochasticity while avoiding the drawbacks of uncontrolled random sampling, yielding inefficient exploration. In this paper, we propose CalibRL, a hybrid-policy RLVR framework that supports controllable exploration with expert guidance, enabled by two key mechanisms. First, a distribution-aware advantage weighting scales updates by group rareness to calibrate the distribution, therefore preserving exploration. Meanwhile, the asymmetric activation function (LeakyReLU) leverages the expert knowledge as a calibration baseline to moderate overconfident updates while preserving their corrective direction. CalibRL increases policy entropy in a guided manner and clarifies the target distribution by estimating the on-policy distribution through online sampling. Updates are driven by these informative behaviors, avoiding convergence to erroneous patterns. Importantly, these designs help alleviate the distributional mismatch between the model's policy and expert trajectories, thereby achieving a more stable balance between exploration and exploitation. Extensive experiments across eight benchmarks, including both in-domain and out-of-domain settings, demonstrate consistent improvements, validating the effectiveness of our controllable hybrid-policy RLVR training. Code is available at https://github.com/zhh6425/CalibRL.

Zhuoxu Huang, Mengxi Jia, Hao Sun +2

Aberystwyth University · Institute of Artificial Intelligence (TeleAl), China Telecom · Tsinghua University

RL ExplorationRL for VLMsReinforcement Learning with Verifiable RewardsMultimodal Reasoning
AskPDF
Feb 20, 2026

Gradient Regularization Mitigates Reward Hacking in Reinforcement Learning from Human Feedback and Verifiable Rewards

Reinforcement Learning from Human Feedback (RLHF) or Verifiable Rewards (RLVR) are two key steps in the post-training of modern Language Models (LMs). A common problem is reward hacking, where the policy may exploit inaccuracies of the reward and learn an unintended behavior. Most previous works address this by limiting the policy update with a Kullback-Leibler (KL) penalty towards a reference model. We propose a different framing: Train the LM in a way that biases policy updates towards regions in which the reward is more accurate. First, we derive a theoretical connection between the accuracy of a reward model and the flatness of an optimum at convergence. Gradient regularization (GR) can then be used to bias training to flatter regions and thereby maintain reward model accuracy. We confirm these results by showing that the gradient norm and reward accuracy are empirically correlated in RLHF. We then empirically show that Reference Resets of the KL penalty find flatter regions with a higher reward accuracy. We further improve on this by proposing to use explicit GR with an efficient finite-difference estimate. Empirically, GR performs better than a KL penalty across a diverse set of RL experiments with LMs. GR achieves a higher GPT-judged win-rate in RLHF, avoids overly focusing on the format in rule-based math rewards, and prevents hacking the judge in LLM-as-a-Judge math tasks.

Johannes Ackermann, Michael Noukhovitch, Takashi Ishida +1

The University of Tokyo · Mila · RIKEN AIP

Reward HackingReinforcement Learning with Verifiable RewardsPolicy LearningRL from Human FeedbackRegularized RL
AskPDF
Feb 19, 2026

MASPO: Unifying Gradient Utilization, Probability Mass, and Signal Reliability for Robust and Sample-Efficient LLM Reasoning

Existing Reinforcement Learning with Verifiable Rewards (RLVR) algorithms, such as GRPO, rely on rigid, uniform, and symmetric trust region mechanisms that are fundamentally misaligned with the complex optimization dynamics of Large Language Models (LLMs). In this paper, we identify three critical challenges in these methods: (1) inefficient gradient utilization caused by the binary cutoff of hard clipping, (2) insensitive probability mass arising from uniform ratio constraints that ignore the token distribution, and (3) asymmetric signal reliability stemming from the disparate credit assignment ambiguity between positive and negative samples. To bridge these gaps, we propose Mass-Adaptive Soft Policy Optimization (MASPO), a unified framework designed to harmonize these three dimensions. MASPO integrates a differentiable soft Gaussian gating to maximize gradient utility, a mass-adaptive limiter to balance exploration across the probability spectrum, and an asymmetric risk controller to align update magnitudes with signal confidence. Extensive evaluations demonstrate that MASPO serves as a robust, all-in-one RLVR solution, significantly outperforming baselines. Our code is available at: https://github.com/FlyTune/MASPO-RL.

Xiaoliang Fu, Jiaye Lin, Yangyi Fang +7

Meituan · Fudan University · Tsinghua University +2

RL for Language ModelsRL for Language Model ReasoningReinforcement Learning with Verifiable RewardsRobust RLSample-Efficient RL
AskPDF
Feb 18, 2026

DeepVision-103K: A Visually Diverse, Broad-Coverage, and Verifiable Mathematical Dataset for Multimodal Reasoning

Reinforcement Learning with Verifiable Rewards (RLVR) has been shown effective in enhancing the visual reflection and reasoning capabilities of Large Multimodal Models (LMMs). However, existing datasets are predominantly derived from either small-scale manual construction or recombination of prior resources, which limits data diversity and coverage, thereby constraining further gains in model performance. To this end, we introduce DeepVision-103K, a comprehensive dataset for RLVR training that covers diverse K12 mathematical topics, extensive knowledge points, and rich visual elements. Models trained on DeepVision achieve strong performance on multimodal mathematical benchmarks, and generalize effectively to general multimodal reasoning tasks. Further analysis reveals enhanced visual perception, reflection and reasoning capabilities in trained models, validating DeepVision's effectiveness for advancing multimodal reasoning. Data: this url.

Haoxiang Sun, Lizhen Xu, Bing Zhao +5

Alibaba Group · Shanghai Jiao Tong University

Multimodal ReasoningMathematical ReasoningReinforcement Learning with Verifiable Rewards
AskPDF
Feb 17, 2026

The Obfuscation Atlas: Mapping Where Honesty Emerges in RLVR with Deception Probes

Training against white-box deception detectors has been proposed as a way to make AI systems honest. However, such training risks models learning to obfuscate their deception to evade the detector. Prior work has studied obfuscation only in artificial settings where models were directly rewarded for harmful output. We construct a realistic coding environment where reward hacking via hardcoding test cases naturally occurs, and show that obfuscation emerges in this setting. We introduce a taxonomy of possible outcomes when training against a deception detector. The model either remains honest, or becomes deceptive via two possible obfuscation strategies. (i) Obfuscated activations: the model outputs deceptive text while modifying its internal representations to no longer trigger the detector. (ii) Obfuscated policy: the model outputs deceptive text that evades the detector, typically by including a justification for the reward hack. Empirically, obfuscated activations arise from representation drift during RL, with or without a detector penalty. The detector penalty only incentivizes obfuscated policies; we theoretically show this is expected for policy gradient methods. Sufficiently high KL regularization and detector penalty can yield honest policies, establishing white-box deception detectors as viable training signals for tasks prone to reward hacking.

Mohammad Taufeeque, Stefan Heimersheim, Adam Gleave +1

FAR.AI.

Deception in Language ModelsLLM HonestyReinforcement Learning with Verifiable Rewards
AskPDF
Feb 16, 2026

On the Emergence of Implicit Curriculum in RLVR Learning Dynamics

Reinforcement learning with verifiable rewards (RLVR) has been a main driver of recent breakthroughs in large reasoning models. Yet it remains a mystery how rewards based solely on final outcomes can help overcome the long-horizon barrier to extended reasoning. To understand this, we develop a theory of the training dynamics of RLVR for transformers on compositional reasoning tasks. Our theory shows that mixed-difficulty training naturally induces an implicit curriculum: without any explicit schedule, easier problems become learnable first and shape the frontier for harder ones, creating a learning progression from easy to hard during optimization. The effectiveness of this curriculum is governed by the smoothness of the difficulty spectrum. When the spectrum is smooth, training dynamics enter a well-behaved relay regime, in which persistent gradient signals on easier problems make slightly harder ones tractable and keep training at the edge of competence. When the spectrum contains abrupt discontinuities, training undergoes grokking-type phase transitions with prolonged plateaus before progress recurs. As a technical contribution, our analysis develops and adapts techniques from Fourier analysis on finite groups to our setting. We validate the predicted mechanisms empirically via controlled synthetic experiments and real-model RLVR runs.

Yu Huang, Zixin Wen, Yuejie Chi +4

Department of Statistics and Data Science, Wharton School, University of Pennsylvania · Machine Learning Department, Carnegie Mellon University · Department of Statistics and Data Science, Yale University +1

RL for Language Model ReasoningReinforcement Learning with Verifiable RewardsCurriculum LearningNeural Network Training Dynamics
AskPDF
Feb 13, 2026

To Mix or To Merge: Toward Multi-Domain Reinforcement Learning for Large Language Models

Reinforcement Learning with Verifiable Rewards (RLVR) plays a key role in stimulating the explicit reasoning capability of Large Language Models (LLMs). We can achieve expert-level performance in some specific domains via RLVR, such as coding or math. When a general multi-domain expert-level model is required, we need to carefully consider the collaboration of RLVR across different domains. The current state-of-the-art models mainly employ two different training paradigms for multi-domain RLVR: mixed multi-task RLVR and separate RLVR followed by model merging. However, most of the works did not provide a detailed comparison and analysis about these paradigms. To this end, we choose multiple commonly used high-level tasks (e.g., math, coding, science, instruction following, and agent) as our target domains and design extensive qualitative and quantitative experiments using open-source datasets. We find the RLVR across domains exhibits small mutual interferences, and reasoning-intensive domains have mutually synergistic effects. Furthermore, we analyze the internal mechanisms from the perspectives of information constraints, model prediction behavior and self-verification. Our homepage is at https://github.com/Mosi-AI/M2RL.

Haoqing Wang, Xiang Long, Ziheng Li +3

Samsung Research, Beijing, China · Peking University

RL for Language ModelsGradient InterferenceReinforcement Learning with Verifiable RewardsMulti-Task RLModel Merging
AskPDF
Feb 12, 2026

Composition-RL: Compose Your Verifiable Prompts for Reinforcement Learning of Large Language Models

Large-scale verifiable prompts underpin the success of Reinforcement Learning with Verifiable Rewards (RLVR), but they contain many uninformative examples and are costly to expand further. Recent studies focus on better exploiting limited training data by prioritizing hard prompts whose rollout pass rate is 0. However, easy prompts with a pass rate of 1 also become increasingly prevalent as training progresses, thereby reducing the effective data size. To mitigate this, we propose Composition-RL, an offline data-augmentation approach that composes existing prompts before RL to address the pass-rate-1 bottleneck without modifying the RL optimizer. The composed dataset is generated and verified once from the fixed source pool and then used for RL training. Extensive experiments across model sizes from 4B to 30B show that Composition-RL consistently improves reasoning capability over RL trained on the original dataset. Performance can be further boosted with a curriculum variant of Composition-RL that gradually increases compositional depth over training. Additionally, Composition-RL enables more effective cross-domain RL by composing prompts drawn from different domains. All our code, data, and models are publicly available to facilitate future research.

Xin Xu, Clive Bai, Kai Yang +8

HY, Tencent · The Hong Kong University of Science and Technology · The University of Hong Kong

RL for Language Model ReasoningReinforcement Learning with Verifiable RewardsData AugmentationPrompt Engineering
AskPDF
Feb 10, 2026

Reward Modeling for Reinforcement Learning-Based LLM Reasoning: Design, Challenges, and Evaluation

Large Language Models (LLMs) demonstrate transformative potential, yet their reasoning remains inconsistent and unreliable. Reinforcement learning (RL)-based fine-tuning is a key mechanism for improvement, but its effectiveness is fundamentally governed by reward design. Despite its importance, the relationship between reward modeling and core LLM challenges--such as evaluation bias, hallucination, distribution shift, and efficient learning--remains poorly understood. This work argues that reward modeling is not merely an implementation detail but a central architect of reasoning alignment, shaping what models learn, how they generalize, and whether their outputs can be trusted. We introduce Reasoning-Aligned Reinforcement Learning (RARL), a reasoning-centric taxonomic perspective that organizes diverse reward paradigms for multi-step reasoning. Within this perspective, we present a taxonomy of reward mechanisms, analyze reward hacking as a pervasive failure mode, and examine how reward signals unify challenges ranging from inference-time scaling to hallucination mitigation. We further critically evaluate existing benchmarks, highlighting vulnerabilities such as data contamination and reward misalignment, and outline directions for more robust evaluation. By integrating fragmented research threads and clarifying the interplay between reward design and fundamental reasoning capabilities, this work provides a foundational roadmap for building reasoning models that are robust, verifiable, and trustworthy.

Pei-Chi Pan, Yingbin Liang, Sen Lin

Department of Computer Science University of Houston · The Ohio State University · University of Houston

Reward ModelingReward HackingProcess Reward ModelsReasoning EvaluationRL for Language Model ReasoningReinforcement Learning with Verifiable Rewards
AskPDF
Feb 8, 2026

Process Rewards for Outcome-Guided Reasoning Steps in LLM Reinforcement Learning

Mathematical reasoning in large language models has improved substantially with reinforcement learning using verifiable rewards, where final answers can be checked automatically and converted into reliable training signals. Most such pipelines optimize outcome correctness only, which yields sparse feedback for long, multi-step solutions and offers limited guidance on intermediate reasoning errors. Recent work therefore introduces process reward models (PRMs) to score intermediate steps and provide denser supervision. In practice, PRM scores are often imperfectly aligned with final correctness and can reward locally fluent reasoning that still ends in an incorrect answer. When optimized as absolute rewards, such signals can amplify fluent failure modes and induce reward hacking. We propose PROGRS, a framework that leverages PRMs while keeping outcome correctness dominant. PROGRS treats process rewards as relative preferences within outcome groups rather than absolute targets. We introduce outcome-conditioned centering, which shifts PRM scores of incorrect trajectories to have zero mean within each prompt group. It removes systematic bias while preserving informative rankings. PROGRS combines a frozen quantile-regression PRM with a multi-scale coherence evaluator. We integrate the resulting centered process bonus into Group Relative Policy Optimization (GRPO) without auxiliary objectives or additional trainable components. Across MATH-500, AMC, AIME, MinervaMath, and OlympiadBench, PROGRS consistently improves Pass@1 over outcome-only baselines and achieves stronger performance with fewer rollouts. These results show that outcome-conditioned centering enables safe and effective use of process rewards for mathematical reasoning.

Mohammad Rezaei, Jens Lehmann, Sahar Vahdati

Faculty of Computer Science, Technical University of Dresden, Dresden, Germany · Amazon Science · Technical University of Dresden, Dresden, Germany +1

LLM Mathematical ReasoningReinforcement LearningProcess Reward ModelsReinforcement Learning with Verifiable Rewards
AskPDF
Feb 6, 2026

F-GRPO: Don't Let Your Policy Learn the Obvious and Forget the Rare

Reinforcement Learning with Verifiable Rewards (RLVR) is commonly based on group sampling to estimate advantages and stabilize policy updates. In practice, computational limits often rule out very large groups, so training proceeds with finite rollout sets that can reinforce only the correct behavior they expose. At practical group sizes, updates can miss rare-correct trajectories while still containing mixed rewards, concentrating probability on more common sampled solutions. We derive the probability of such prompt-local tail-miss events as a function of group size, showing non-monotonic behavior, and in the categorical abstraction characterize how unsampled-correct mass can shrink even as total correct mass grows. Motivated by this analysis, we propose a difficulty-aware scaling coefficient, inspired by Focal loss, that down-weights updates on high-success sampled groups. Empirically, categorical simulation illustrates the same effect in the categorical setting, Maze provides a single-solution test, and LLM experiments include a representative GRPO group-size sweep together with fixed-NN transfer across GRPO, DAPO, and CISPO. On Qwen2.5-7B at N=8N{=}8, our method improves average math pass@256 from 64.1 →\rightarrow 70.3 (GRPO), 69.3 →\rightarrow 72.5 (DAPO), and 73.2 →\rightarrow 76.8 (CISPO); OOD pass@256 also improves in all three cases, without increasing group size or computational cost.

Daniil Plyusov, Alexey Gorbatovski, Boris Shaposhnikov +4

T-Tech · Saint Petersburg Electrotechnical University “LETI”

Reinforcement LearningRL for Language Model ReasoningGroup Relative Policy OptimizationReinforcement Learning with Verifiable Rewards
AskPDF
Feb 5, 2026

Unveiling Implicit Advantage Symmetry: Why GRPO Struggles with Exploration and Difficulty Adaptation

Reinforcement Learning with Verifiable Rewards (RLVR), particularly GRPO, has become the standard for eliciting LLM reasoning. However, its efficiency in exploration and difficulty adaptation remains an open challenge. In this work, we identify an implicit advantage symmetry inherent in Group Relative Advantage Estimation (GRAE) as a structural property that provides a new perspective for understanding these bottlenecks. This symmetry induces two critical limitations: (i) at the group level, strict symmetry in weights between correct and incorrect trajectories leaves unsampled action logits unchanged, thereby hindering the exploration of novel correct solution. (ii) at the sample level, the algorithm implicitly prioritizes medium-difficulty samples, remaining agnostic to the non-stationary demands of difficulty focus. Through controlled experiments, we reveal that this symmetric property is sub-optimal, yielding two pivotal insights: (i) asymmetrically down-weighting the advantages of correct trajectories encourages essential exploration but risks instability; (ii) learning efficiency can be boosted by a curriculum-like transition-prioritizing simpler samples initially before gradually shifting to complex ones. Motivated by these findings, we propose Asymmetric GRAE (A-GRAE), which dynamically modulates exploration incentives and sample-difficulty focus. Experiments across seven benchmarks demonstrate that A-GRAE consistently improves GRPO and its variants across both LLMs and MLLMs. Code is available at https://github.com/HKU-HealthAI/A-GRAE

Zhiqi Yu, Zhangquan Chen, Mengting Liu +2

University of Hong Kong · Tsinghua University · Sun Yat-sen University

RL for Language Model ReasoningGroup Relative Policy OptimizationRL ExplorationReinforcement Learning with Verifiable Rewards
AskPDF
Feb 4, 2026

CodeScaler: Scaling Code LLM Training and Test-Time Inference via Reward Models

Reinforcement Learning from Verifiable Rewards (RLVR) has driven recent progress in code large language models by leveraging execution-based feedback from unit tests, but its scalability is fundamentally constrained by the availability and reliability of high-quality test cases. We propose CodeScaler, a reward model designed to scale both reinforcement learning training and test-time inference for code generation. CodeScaler is trained on carefully curated preference data derived from verified code problems and incorporates syntax-aware code extraction and validity-preserving reward shaping to ensure stable and robust optimization. Across four coding benchmarks, CodeScaler consistently outperforms execution-based RL by +1.55 points on Qwen3-8B-Base and +4.23 points on Qwen3-14B-Base. By further scaling to 44K problems with additional synthetic data, CodeScaler yields +14.64 points improvement over the base model without requiring any test cases. At inference time, CodeScaler serves as an effective test-time scaling method, achieving performance comparable to unit test approaches while providing a 10-fold reduction in latency. Moreover, CodeScaler surpasses existing reward models on RM-Bench not only in the code domain (+3.3 points), but also in general and reasoning domains (+2.7 points on average).

Xiao Zhu, Xinyu Zhou, Boyu Zhu +5

LARK, HKUST(GZ) · Kuaishou Technology · UCL +3

RL for Code GenerationReward ModelingTest-Time ScalingReinforcement Learning with Verifiable Rewards
AskPDF
Jan 30, 2026

HeaPA: Difficulty-Aware Heap Sampling and On-Policy Query Augmentation for LLM Reinforcement Learning

RLVR has become a standard recipe for training LLMs on reasoning tasks with verifiable outcomes, but when rollout generation dominates the cost, efficiency hinges on which prompts are sampled and when. In practice, prompt pools are often static or only weakly coupled to policy progress, so uniform sampling fails to track the moving capability frontier and wastes rollouts on regions that are already solved or still unreachable. Prior methods improve efficiency via filtering, curricula, adaptive rollout allocation, or teacher guidance, but they often assume a fixed pool, which does not support stable on-policy pool growth, or they introduce additional teacher cost and latency. In this work, we propose HeaPA (Heap Sampling and On-Policy Query Augmentation), which maintains a bounded, evolving pool, tracks the frontier with heap-based boundary sampling, grows the pool via on-policy augmentation under lightweight asynchronous validation, and stabilizes correlated queries via topology-aware pool statistics re-estimation and controlled reinsertion. Across two training corpora, two training recipes, and seven benchmarks, HeaPA consistently improves accuracy and reaches target performance with fewer computations at comparable wall-clock time. Analyses attribute the gains to frontier-focused sampling and on-policy pool growth, with more pronounced improvements at mid-to-large model scales. Our training code is publicly available at https://github.com/horizon-llm/HeaPA.

Weiqi Wang, Xin Liu, Binxuan Huang +13

Stores Foundational AI, Amazon Inc., Palo Alto, CA, USA · Department of Computer Science and Engineering, HKUST, Hong Kong SAR, China

RL for Language ModelsRL for Language Model ReasoningAdaptive SamplingReinforcement Learning with Verifiable RewardsRL Post-TrainingSample-Efficient RL
AskPDF
Jan 16, 2026

Spurious Rewards Paradox: Mechanistically Understanding How RLVR Activates Memorization Shortcuts in LLMs

Reinforcement Learning with Verifiable Rewards (RLVR) is highly effective for enhancing LLM reasoning, yet recent evidence shows models like Qwen 2.5 achieve significant gains even with spurious or incorrect rewards. We investigate this phenomenon and identify a "Perplexity Paradox": spurious RLVR triggers a divergence where answer-token perplexity drops while prompt-side coherence degrades, suggesting the model is bypassing reasoning in favor of memorization. Using Path Patching, Logit Lens, JSD analysis, and Neural Differential Equations, we uncover a hidden Anchor-Adapter circuit that facilitates this shortcut. We localize a Functional Anchor in the middle layers (L18-20) that triggers the retrieval of memorized solutions, followed by Structural Adapters in later layers (L21+) that transform representations to accommodate the shortcut signal. Finally, we demonstrate that scaling specific MLP keys within this circuit allows for bidirectional causal steering-artificially amplifying or suppressing contamination-driven performance. Our results provide a mechanistic roadmap for identifying and mitigating data contamination in RLVR-tuned models. Code is available at https://github.com/idwts/How-RLVR-Activates-Memorization-Shortcuts.

Lecheng Yan, Ruizhe Li, Guanhua Chen +5

Southern University of Science and Technology · University of Science and Technology of China · University of Aberdeen +5

Reward HackingMemorization in Language ModelsLLM InterpretabilityRL for Language Model ReasoningReinforcement Learning with Verifiable Rewards
AskPDF
Jan 11, 2026

X-Coder: Advancing Competitive Programming with Synthetic Tasks, Solutions, and Tests

Competitive programming remains challenging for code LLMs. Despite recent progress, many training pipelines still depend on scarce real-world data, raising concerns about scalability and near-duplicate benchmark contamination. In this paper, we examine whether synthetic training artifacts can support the complete SFT-to-RL cycle for competitive programming: no real-world tasks, solutions, or test cases are directly used for post-training. To this end, we synthesize tasks, verified solutions, and reliable test cases that serve as reward signals for reinforcement learning. To improve reward reliability, we introduce a dual-verification strategy that reduces noise in both selected solutions and test outputs. Using this high-quality data, we train the X-Coder model series. X-Coder-14B achieves 67.5% avg@8 on LiveCodeBench v5 and 63.4% on v6, outperforming its base model by over 40 points. Further analysis provides practical insights into synthetic post-training, highlighting the value of diverse tasks, verified long-CoT supervision, and code-centric reinforcement. Our data and models are released at https://github.com/JieWu02/X-Coder

Jie Wu, Haoling Li, Xin Zhang +8

Tsinghua University · Microsoft

RL for Code GenerationSynthetic Data GenerationCompetitive ProgrammingReinforcement Learning with Verifiable RewardsLLM Code Generation
AskPDF
Jan 11, 2026

GDEPO: Group Dual-dynamic and Equal-right Advantage Policy Optimization for Sample-Efficient Reinforcement Learning

Automated Theorem Proving (ATP) is a fundamental AI challenge, requiring machine-verifiable proofs in formal languages like Lean to evaluate reasoning. Reinforcement Learning (RL), especially Group Relative Policy Optimization (GRPO), is a mainstream approach. However, GRPO faces two critical issues in ATP: its relative advantage estimation may conflict with binary verifier feedback under composite rewards, and its static sampling strategy discards batches without valid proofs, causing data waste. Repeated training on mastered problems also wastes computation. To address this, we propose Group Dual-dynamic and Equal-right Advantage Policy Optimization (GDEPO), incorporating three core mechanisms: 1) dynamic additional sampling, resampling ineffective batches until a valid proof is found; 2) equal-right advantage, decoupling the advantage function's sign (correctness) from its magnitude (auxiliary rewards) to ensure stable policy updates while increasing effective computation; and 3) dynamic additional iterations, applying extra gradient steps to initially failed but eventually successful samples to accelerate learning. Experiments on MinF2F-test, MathOlympiadBench, and PutnamBench confirm GDEPO's effectiveness, with ablation studies validating its synergistic components. The method enhances data utilization and optimization efficiency, offering a novel training paradigm for ATP.

Zhengqing Yan, Fan Guo, Xinyang Liu +5

State Key Laboratory of Engines, Tianjin University, 300354 Tianjin, China. · Artificial Intelligence Center, Cylingo Group, 100080 Beijing, China.

Group Relative Policy OptimizationReinforcement Learning with Verifiable RewardsSample-Efficient RLAutomated Theorem Proving
AskPDF
Dec 2, 2025

ScalePRM: Training Process Reward Models by Scaling Verification Compute Without Ground Truth

Training process reward models (PRMs) requires step-level correctness labels, obtained either through expensive human annotation or by relying on ground-truth answers, limiting the ability to scale process-level supervision. We propose ScalePRM, which scales verification compute as an alternative: given a problem and a candidate solution, we generate multiple independent verifications of each reasoning step and aggregate their judgments to produce synthetic step-level labels without ground truth. We explore two representative inference-time scaling strategies, parallel scaling through self-consistency and sequential scaling through meta-critique, and train generative PRMs on the resulting synthetic data. On ProcessBench, a benchmark for identifying erroneous steps in mathematical reasoning, PRMs trained on step-level self-consistency data achieve 67.5 F1, surpassing reference-guided training with ground-truth access (66.4 F1) and GPT-4o as a critic (61.9 F1). When deployed as reward signals in RL training with Qwen2.5-Math-7B, our best PRM achieves 47.4% average accuracy across six mathematical reasoning benchmarks, outperforming ground-truth-based RLVR (43.9%). We also identify and address reward exploitation patterns unique to generative PRM-based RL. Our results demonstrate that scaling verification compute is a viable alternative to ground-truth supervision for training process reward models.

Salman Rahman, Sruthi Gorantla, Arpit Gupta +3

2UCLA · Work done while as an intern at Amazon AGI. · 1Amazon AGI

Process Reward ModelsReinforcement Learning with Verifiable RewardsLLM Mathematical Reasoning
AskPDF
Nov 14, 2025

Honesty over Accuracy: Trustworthy Language Models through Reinforced Hesitation

Modern language models fail a fundamental requirement of trustworthy intelligence: knowing when not to answer. Despite achieving impressive accuracy on benchmarks, these models produce confident hallucinations, even when wrong answers carry catastrophic consequences. Our evaluations on GSM8K, MedQA and GPQA show frontier models almost never abstain despite explicit warnings of severe penalties, suggesting that prompts cannot override training that rewards any answer over no answer. As a remedy, we propose Reinforced Hesitation (RH): a modification to Reinforcement Learning from Verifiable Rewards (RLVR) to use ternary rewards (+1 correct, 0 abstention, -λλ error) instead of binary. Controlled experiments on logic puzzles reveal that varying λλ produces distinct models along a Pareto frontier, where each training penalty yields the optimal model for its corresponding risk regime: low penalties produce aggressive answerers, high penalties conservative abstainers. The same frontier holds on MATH Levels 4--5 and on medical QA, where it transfers to an unseen dataset. We then introduce two inference strategies that exploit trained abstention as a coordination signal: cascading routes queries through models with decreasing risk tolerance, while self-cascading re-queries the same model on abstention. Both outperform majority voting with lower computational cost. These results establish abstention as a first-class training objective that transforms ``I don't know'' from failure into a coordination signal, enabling models to earn trust through calibrated honesty about their limits.

Mohamad Amin Mohamadi, Tianhao Wang, Zhiyuan Li

Toyota Technological Institute at Chicago · University of California, San Diego

LLM AbstentionLLM Hallucination MitigationLLM ReliabilityLanguage ModelingReinforcement Learning with Verifiable RewardsRefusal Behavior in Language Models+3
AskPDF
Oct 24, 2025

Beyond Reasoning Gains: Mitigating General-Capability Forgetting in Large Reasoning Models

Reinforcement learning with verifiable rewards (RLVR) has delivered impressive gains in mathematical and multimodal reasoning and has become a standard post-training paradigm for contemporary language and vision-language models. However, the RLVR recipe introduces a significant risk of capability regression, in which models forget foundational skills after prolonged training without employing regularization strategies. We empirically confirm this concern, observing that open-source reasoning models suffer performance degradation on core capabilities such as perception and faithfulness. While imposing regularization terms like KL divergence can help prevent deviation from the base model, these terms are computed on the current task and therefore do not guarantee preservation of broader knowledge. Meanwhile, commonly used experience replay across heterogeneous domains makes it nontrivial to decide how much training emphasis each objective should receive. To address this, we propose RECAP-a replay strategy with dynamic objective reweighting for general knowledge preservation. Our reweighting mechanism adapts online using short-horizon signals of convergence and instability, shifting the post-training focus away from saturated objectives and toward underperforming or volatile ones. Our method is end-to-end and readily applicable to existing RLVR pipelines without training additional models or heavy tuning. Extensive experiments on benchmarks using Qwen2.5-VL-3B and Qwen2.5-VL-7B demonstrate the effectiveness of our method, which not only preserves general capabilities but also improves reasoning by enabling more flexible trade-offs among in-task rewards.

Hoang Phan, Xianjun Yang, Yuanshun Yao +6

Meta Superintelligence Labs · New York University · Johns Hopkins University

RL Post-TrainingCatastrophic ForgettingAdaptive Loss WeightingReinforcement Learning with Verifiable RewardsExperience ReplayReasoning for VLMs
AskPDF
Oct 9, 2025

Group Contrastive Policy Optimization: Learning Beyond the Self-Exploration Limit with Golden Answers

Reinforcement learning with verifiable rewards (RLVR) has become a powerful paradigm for improving the reasoning abilities of large language models, yet group-based methods lose their learning signal when every sampled response receives zero reward. We propose Group Contrastive Policy Optimization (GCPO), a hybrid objective that preserves standard on-policy optimization for informative rollout groups and activates a proximal contrastive objective only when exploration completely fails. The auxiliary objective pairs an offline golden reasoning trajectory with the current policy's failed responses, thereby providing a reliable positive reference and prompt-specific hard negatives without treating the golden trajectory as an on-policy rollout. Its old-policy token probabilities are evaluated through teacher forcing and used solely to constrain the update, rather than as an importance-sampling correction for the teacher distribution. A first-order analysis shows that the resulting update increases a length-normalized likelihood margin between reliable reasoning and observed failures. Experiments across two model scales and 10 core mathematical reasoning benchmarks demonstrate that GCPO improves training efficiency and final reasoning performance over competitive supervised and RLVR baselines.

Hao Wu, Wei Liu, Shizhe Xiang +3

RL for Language Model ReasoningLLM ReasoningReinforcement Learning with Verifiable RewardsContrastive RLGroup Relative Policy Optimization
AskPDF
Oct 5, 2025

Selective Expert Guidance for Effective and Diverse Exploration in Reinforcement Learning of LLMs

Reinforcement Learning with Verifiable Rewards (RLVR) has become a widely adopted technique for enhancing the reasoning ability of Large Language Models (LLMs). However, the effectiveness of RLVR strongly depends on the capability of base models. This issue arises because it requires the model to have sufficient capability to perform high-quality exploration, which involves both effectiveness and diversity. Unfortunately, existing methods address this issue by imitating expert trajectories, which improve effectiveness but neglect diversity. To address this, we argue that the expert only needs to provide guidance only at critical decision points rather than the entire reasoning path. Based on this insight, we propose MENTOR: Mixed-policy Expert Navigation for Token-level Optimization of Reasoning, a framework that provides expert guidance only at critical decision points to perform effective and diverse exploration in RLVR. Extensive experiments show that MENTOR enables models capture the essence of expert strategies rather than surface imitation, thereby performing high-quality exploration and achieving superior overall performance. Our code is available online.

Zishang Jiang, Jinyi Han, Tingyun Li +7

School of Data Science, Fudan University · Shanghai Institute of Artificial Intelligence for Education, East China Normal University · College of Computer Science and Artificial Intelligence, Fudan University +1

RL for Language Model ReasoningLLM-Guided RLRL ExplorationReinforcement Learning with Verifiable Rewards
AskPDF
Oct 1, 2025

Reinforcement Learning with Verifiable yet Noisy Rewards under Imperfect Verifiers

Reinforcement Learning with Verifiable Rewards (RLVR) replaces costly human labeling with automated verifiers. To reduce verifier hacking, many RLVR systems binarize rewards to {0,1}\{0,1\}, but imperfect verifiers inevitably introduce false negatives (rejecting correct answers) and false positives (accepting incorrect ones). We formalize verifier unreliability as a stochastic reward channel with asymmetric noise rates ρ0ρ_0 and ρ1ρ_1 -- the FP rate and the FN rate, respectively. From this abstraction we derive two lightweight corrections: (i) a backward correction that yields an unbiased surrogate reward and thus an unbiased policy-gradient estimator in expectation, and (ii) a forward correction that reweights score-function terms so the expected update aligns with the clean gradient direction and requires only the FN rate. We implement both as lightweight hooks in a group relative policy optimization pipeline, both corrections improve RLVR for math reasoning under synthetic and real verifier noise, with the forward variant being more stable under heavier noise. Finally, an appeals mechanism with a lightweight LLM verifier estimates the FN rate online and further improves performance.

Xin-Qiang Cai, Wei Wang, Feng Liu +3

RIKEN AIP, Tokyo, Japan · The University of Tokyo, Tokyo, Japan · The University of Melbourne, Melbourne, Australia

Reinforcement LearningRL for Language Model ReasoningReinforcement Learning with Verifiable RewardsRL Post-Training
AskPDF
Sep 30, 2025

Thinking-Free Policy Initialization Makes Distilled Reasoning Models More Effective and Efficient Reasoners

Reinforcement Learning with Verifiable Reward (RLVR) effectively solves complex tasks but demands extremely long context lengths during training, leading to substantial computational costs. While multi-stage training can partially mitigate this, starting with overly short contexts often causes irreversible performance degradation, ultimately failing to reduce overall training compute significantly. In this paper, we introduce Thinking-Free Policy Initialization (TFPI), a simple yet effective adaptation to RLVR that bridges long Chain-of-Thought (CoT) distillation and standard RLVR. TFPI employs a simple ThinkingFree operation, explicitly discarding the thinking content via a direct </think> append, to reduce token usage during inference. Training with ThinkingFree-adapted inputs improves performance and lowers token consumption, even in the original slow-thinking mode. Extensive experiments across various benchmarks have shown that TFPI accelerates RL convergence, achieves a higher performance ceiling, and yields more token-efficient reasoning models without specialized rewards or complex training designs. With TFPI only, we train a 4B model to reach 89.0% accuracy on AIME24 and 65.5% on LiveCodeBench using less than 4K H20 hours.

Xin Xu, Clive Bai, Kai Yang +4

LLM Department, Tencent · The Hong Kong University of Science and Technology · The University of Hong Kong

Efficient Language Model ReasoningRL for Language Model ReasoningReinforcement Learning with Verifiable RewardsCoT DistillationLanguage Model Distillation
AskPDF
Sep 30, 2025

Clip-Low Increases Entropy and Clip-High Decreases Entropy in Reinforcement Learning of Large Language Models

Reinforcement learning with verifiable rewards (RLVR) has recently emerged as the leading approach for enhancing the reasoning capabilities of large language models (LLMs). However, RLVR is prone to entropy collapse, where the LLM quickly converges to a near-deterministic form, hindering exploration and progress during prolonged RL training. In this work, we reveal that the clipping mechanism in PPO and GRPO induces biases on entropy. Through theoretical and empirical analyses, we show that clip-low increases entropy, while clip-high decreases it. Further, under standard clipping parameters, the effect of clip-high dominates, resulting in an overall entropy reduction even when purely random rewards are provided to the RL algorithm. Our findings highlight an overlooked confounding factor in RLVR: independent of the reward signal, the clipping mechanism influences entropy, which in turn affects the reasoning behavior. Furthermore, our analysis demonstrates that clipping can be deliberately used to control entropy. Specifically, with a more aggressive clip-low value, one can increase entropy, promote exploration, and ultimately prevent entropy collapse in RLVR training.

Jaesung R. Park, Junsu Kim, Gyeongman Kim +4

Department of Mathematics, UCLA · Department of Mathematical Sciences, Seoul National University · KRAFTON +2

RL for Language ModelsReinforcement LearningReinforcement Learning with Verifiable Rewards
AskPDF