RL for VLMs

RL: Reinforcement Learning · VLM: Vision-Language Model

Follow this topic Ask about this topic

Momentum

36 papers in the last four weeks, up 227% on the four weeks before. 0.4% of all new papers.

Jul 13Week of Sep 28

Latest papers 287

Oct 7, 2026cs.CV

VideoEvolve: Co-Evolving Memory and Retrieval for Long Video Understanding

Long video understanding increasingly relies on external memory to organize massive visual streams into compact representations. However, most memory-based methods dynamically adapt how information is retrieved for different questions, while largely fixing what is remembered. This mismatch makes missing details costly to recover, whereas stored information is valuable only when it can be reliably retrieved. To address this issue, we propose VideoEvolve, a novel self-evolving framework that jointly evolves memory and retrieval for long video understanding. Specifically, starting from a coarse low-frame-rate overview, VideoEvolve couples a Memory Evolver for selective memory augmentation with a Retrieval Evolver for adaptive retrieval over the evolving memory. We then co-evolve the two Evolvers through alternating agentic reinforcement learning (Agentic RL), updating one while freezing the other. To steer this alternating evolution, Bottleneck-Aware Evolution Feedback (BEF) identifies whether the current bottleneck lies in memory or retrieval and directs optimization toward the more limiting side. Furthermore, VideoEvolve introduces Capability-Aware Evolution Feedback (CEF) to alleviate downstream feedback from over-specializing memory to a fixed set of training questions, shifting training toward underdeveloped yet learnable video capabilities. By integrating Agentic RL with BEF and CEF, VideoEvolve transforms downstream reasoning experience into transferable capability updates, providing a concrete path from static long-video systems toward experience-driven, self-improving multimodal intelligence. Extensive experiments on multiple long video understanding benchmarks demonstrate the effectiveness of VideoEvolve.

Yongchao Xu, Bowen Ye, Jiefeng Gan +5

Alibaba Group, Hangzhou, China · University of Science and Technology of China, Hefei, China · Shanghai Jiao Tong University, Shanghai, China +2

Long-Video Understanding Memory-Augmented Video Understanding Video Retrieval RL for VLMs

Oct 6, 2026cs.CV

Selective Transfer of RL Updates for Visual Reasoning

Model merging provides a training-free way to transfer reasoning capabilities from language models to vision-language models (VLMs), but endpoint-based transfer can conflate pre-existing model differences with changes acquired during reasoning post-training. We instead formulate capability transfer around the training-stage update, isolating the parameter changes induced by reinforcement learning (RL). Yet transferring this update in full remains suboptimal: we find that its components differ substantially in cross-model transferability, with dominant directions transferring more effectively than the complete update. Based on this finding, we introduce Selective-RL, which isolates the RL-stage update, retains its dominant matrix-wise directions with magnitude preservation, and transfers them to the language modules of a VLM. Across three model families and five visual-reasoning benchmarks, Selective-RL improves full-update interpolation in 12 of 15 comparisons, including an 8.55 percentage-point MathVision gain on the Qwen recipient. Matched controls show that update magnitude or arbitrary low rank alone does not reproduce these gains. These results highlight a distinction between what is acquired during post-training and what remains transferable across models, providing a training-stage perspective on cross-model capability transfer. Code is available at https://anonymous.4open.science/r/selective-rl.

Suxin Ji, Hungtao Wan, Mingjun Liu +1

University of Pennsylvania · Independent Researcher · University of Electronic Science and Technology of China +1

Reinforcement Learning VLM Reasoning RL for VLMs

Oct 5, 2026cs.CV

MeSD: Multi-Evidence Self-Distillation for VideoLLM

While reinforcement learning with verifiable rewards provides reliable outcome supervision for VideoLLMs, sequence-level rewards offer limited token-level guidance. On-policy self-distillation addresses this limitation by conditioning a self-teacher on privileged information to provide dense token-level supervision. However, aggregating heterogeneous evidence within a single teacher context obscures cross-evidence agreement and conflict. A further challenge lies in determining whether teacher guidance should refine reward-based updates or provide corrective supervision for failed trajectories. To address these issues, we propose MeSD, a multi-evidence self-distillation framework for VideoLLMs. MeSD constructs three evidence-conditioned teachers with shared parameters, using the ground-truth answer as a common semantic context while separately incorporating temporal and spatial evidence. Given the same student-generated prefixes, MeSD evaluates evidence-specific preferences relative to the Answer Teacher and fuses teacher-common preferences with gated teacher-specific residuals. Furthermore, MeSD introduces Verification-Guided Optimization to classify trajectories as Success, Failure, or Indeterminate. For Success and Indeterminate trajectories, MeSD refines token-level advantage magnitudes while preserving reward-derived signs. For verified failure trajectories that contain the required evidence, MeSD applies failure-conditioned distillation, using reverse-KL correction toward the fused distribution. Experiments on multiple video benchmarks demonstrate consistent gains over reinforcement learning and self-distillation baselines.

Weijie Zhu, Han Fang, Hanyu Fu +13

UCAS · Xingchen AGI Lab, China Telecom Artificial Intelligence Technology (Beijing) Co., Ltd · PKU +2

On-Policy Self-Distillation Video-Language Models RL for VLMs

Oct 1, 2026cs.CL

Controllable Multi-label Video Safety Detection via Adaptive Tversky Policy Optimization

The rapid growth of video-based social media has increased users' exposure to harmful content, creating a need for reliable automated video safety detection. Although recent Vision-Language Models (VLMs) show strong video understanding capabilities, existing harmful video detection systems face two key limitations: they typically reduce safety detection to binary classification, overlooking the inherently multi-label nature of unsafe videos, and they rely on static training objectives that do not support controllable precision-recall trade-offs, though the desired operating point may vary across moderation pipelines and unsafe categories. To address these gaps, we propose Adaptive Tversky Policy Optimization (ATPO), a reinforcement learning framework for Multi-label Video Safety Detection (Multi-VSD). ATPO introduces the Adaptive Tversky Reward (ATR), which dynamically adjusts false-positive and false-negative penalties during training to enable controllable precision-recall trade-offs. Experiments on SafeWatch-Bench and XD-Violence show that ATPO substantially improves multi-label performance, increasing the Jaccard Index from 40.66 to 75.44 on SafeWatch-Bench-Real. Moreover, ATR enables reliable steering of the precision-recall operating point, supporting deployment scenarios with heterogeneous policy requirements. Code and checkpoints are provided at https://bruceyg.github.io/ATPO-project-page/ .

Guangyu Yang, Jingbiao Mei, Mingsheng Sun +5

University of Cambridge · Xiaohongshu Inc. · AntGroup +2

Video Content Moderation RL for VLMs

Oct 1, 2026cs.LG

Same Reward, Different Skills: When Multimodal RL Learns to Look

Reinforcement learning with verifiable rewards (RLVR) improves vision-language benchmark scores even without visual information during training. With images at test, blind-trained models recover roughly half of the real-image gain at 3B and nearly four fifths at 7B. Prolonged real-image training can erode grounding while benchmark gains persist. Both findings expose the same gap: an image in the prompt is not an image in the learning signal. Our design rule, visual resolvability, asks that visual evidence be necessary for a correct answer and that the task remain learnable. We test it on counterfactual coordinate scenes in which the question stays fixed and the target is never named, so a correct answer requires finding the target in the image. With standard GRPO and correctness-and-format rewards, a 7B model raises its accuracy at finding the target (discovery) from 0.425 to 0.875 on held-out scenes denser than any it trained on, and it improves on question types it never trained on. Two controls locate the source of the gain. Replacing test images with gray canvases drops discovery to zero; training on gray canvases instead, at matched step 30 and in each of four seeds, yields essentially none of the gain even when the model is then tested with real images. The learned skill carries over to grounding tasks built independently of the training corpus. A caption that answers the training question, added to the same images, reward and budget, cuts the gain by nearly two thirds. Changing what reward requires changes what RL learns.

Haocun Ye, Xinlong Jiang, Qile Chen +6

University of the Chinese Academy of Sciences · Institute of Computing Technology, Chinese Academy of Sciences · Independent Researcher

Reinforcement Learning with Verifiable Rewards Visual Grounding Visual RL RL for VLMs

Oct 1, 2026cs.RO

eRLT: Efficient VLA Reinforcement Learning via Action-Relevant Token Routing

Vision-Language-Action (VLA) models provide strong behavioral priors for robotic manipulation, yet efficiently adapting them to downstream tasks remains challenging. Recent work addresses this challenge by adapting frozen VLAs through online reinforcement learning (RL), whose sample efficiency depends on the quality of the state representation used by the actor and critic. Existing methods construct such representations either with VLA-independent visual encoders or through fixed compression of internal VLA representations. Neither design explicitly extracts the task-specific action-relevant VLA features most useful for downstream action refinement and action-value estimation, therefore limiting sample efficiency. To address this limitation, we introduce eRLT, which constructs an effective state representation by routing task-specific action-relevant information across both tokens and layers of the frozen VLA. Specifically, learned routing tokens dynamically aggregate visual-language features at multiple depths, while a lightweight layer router combines these summaries into a fixed-dimensional RL token. The routing module is initialized using expert demonstrations to capture features predictive of expert actions and then refined using critic feedback from online interactions for action-value estimation. Across seven LIBERO and RoboTwin tasks, eRLT improves mean normalized learning-curve AUC by up to 23.7% over representative baselines. Real-robot experiments on USB connector insertion and motherboard ribbon-cable insertion further show AUC improvements of 108.9% and 46.7%, respectively, over the strongest baseline.

Dehao Huang, Jianbang Liu, Jianpan Gao +7

Southern University of Science and Technology, Shenzhen, China. · Beijing Zhongguancun Academy, Beijing, China. · Samsung Robotics eXperience. +2

Efficient VLA Models Robotic Manipulation Representation Learning for RL RL for VLMs

Sep 30, 2026cs.CV

EviRover: Reinforcing Agentic Perception Beyond a Glance

Visual perception is conventionally formulated as a one-shot prediction from a single glance at the image, under the assumption that the image content and the model's parametric knowledge suffice to resolve the query. This assumption often fails in real-world scenarios that hinge on fine-grained visual details or require knowledge-intensive and up-to-date information. We term such cases \textit{perception under insufficient evidence} and formulate perception as an agentic process that can obtain information beyond a single glance. To address the absence of data for this setting, we design two dedicated data generation pipelines, yielding EviRover-SFT-5K and EviRover-RL-12K for training. We further construct EviLens, a human-verified benchmark comprising 688 instances across five perception categories. Building on these data, we present EviRover, to our knowledge the first perception agent explicitly trained to resolve perceptual queries through interaction, using supervised fine-tuning followed by agentic reinforcement learning. Experiments show that the 4B EviRover outperforms its backbone by 30 points on average on EviLens, reaching performance comparable to advanced proprietary models. The gains transfer beyond EviLens to WebEyes, conventional perception benchmarks, and general multimodal benchmarks, including a 15-point improvement on BrowseComp-VL. All code, models, and data are released.

Kaixuan Fan, Kaituo Feng, Tianshuo Peng +4

MMLab, The Chinese University of Hong Kong · Fudan University

Tool-Using Vision-Language Agents Active Visual Perception Fine-Grained Visual Perception RL for VLMs

Sep 30, 2026cs.CV

Aligning Thoughts with Answers: Probability Rewards to Tame Thinking Drift

This paper studies \textbf{thinking--answer consistency} in vision-language models. We focus on Visual Intention Grounding, where a model infers a target object based on a human intention query and predicts a bounding box. We reveal that previous IoU-based reinforcement learning (RL) frameworks suffer from ``thinking drift'', where the model produces a correct bounding box, despite having an incorrect reasoning process pointing to a different target object. Thus, we propose \textbf{Rita} (\textit{ReInforcing Thinking--Answer consistency}) as a novel RL paradigm to tame the drift. Specifically, Rita introduces two reasoning-label-free RL rewards, constructed from the conditional probability of reference answers: a \textbf{thinking reward} and a \textbf{consistency reward}. It also adopts a difficulty-aware \textbf{data filtering} strategy that selects informative easy-to-medium samples for RL using rollout error rate and reward variance. Extensive experiments on EgoIntention and the new RefEgo-Int benchmarks show that Rita performs consistently superior to the supervised finetuning approaches and vanilla RL-finetuned frameworks.

Pengzhan Sun, Shiu-hong Kao, Shijie Li +4

National University of Singapore · A*STAR Institute of Advanced Intelligence and Computing, Singapore · South China University of Technology +2

VLM Reasoning RL for VLMs

Sep 30, 2026cs.AI

Reinforcing Multimodal Reasoning via Token-Level Perception-Grounded Advantage Estimation

Reinforcement Learning with Verifiable Rewards (RLVR) has improved the reasoning capabilities of Multimodal Large Language Models (MLLMs), yet existing frameworks rely on coarse, sequence-level reward signals that lack the fine-grained supervision over the visually-grounded steps within a multimodal reasoning chain. We investigate this gap through the lens of two token-level metrics: visual dependency (i.e. how much a token's prediction relies on the input image features) and predictive entropy. Our empirical analysis reveals two key findings: (1) correct reasoning chains exhibit a markedly sharper entropy reduction as visual grounding intensifies, compared to incorrect ones; (2) pivotal tokens, those whose misprediction triggers reasoning collapse, are statistical outliers in the joint distribution of visual dependency and predictive entropy derived from correct chains. Motivated by these findings, we propose token-level perception-grounded advantage estimation (TPAE), which estimates token-level advantages by measuring each token's statistical consistency with the vision-entropy patterns of correct rollouts. TPAE leverages this granular score to modulate the sequence-level advantage, producing a fine-grained supervision signal that can be integrated into various RLVR frameworks. Extensive experiments on seven benchmarks show that TPAE consistently outperforms leading strong baselines, yielding more stable and efficient optimization for multimodal reasoning. The code is publicly available at https://github.com/Zhihan72/TPAE.

Zhihan Zhang, Lizi Liao

Singapore Management University Singapore, Singapore

Token-Level Credit Assignment Multimodal Large Language Models Multimodal Reasoning Reinforcement Learning with Verifiable Rewards RL for VLMs

Sep 30, 2026cs.AI

Beyond Prediction: Steering VLM Agents with Retrospective World Modeling

Equipping VLM agents with world modeling capabilities has shown strong potential for complex reasoning and long-horizon planning, while reducing the dependence of policy learning on costly real-world interactions. Existing methods mainly rely on prospective simulation to predict the consequences of candidate actions. However, this forward-only paradigm focuses on what will happen next and provides limited constraints for verifying whether an action is causally consistent with the observed state transition, which can lead to plausible-looking but physically incoherent behaviors. In this paper, we challenge the view of world modeling as only prospective prediction and introduce Retrospective World Modeling, a new agent learning paradigm that enables agents to reason backward by estimating the retrospective attribution distribution

P(\hat{a}{t}|s_t, s{t+1})

for the action that most likely caused a given transition. Based on this capability, we formulate the Self-Consistency Reward (SCR), an intrinsic signal that measures the probabilistic consistency between the policy action and the retrospective explanation. Integrating SCR into reinforcement learning provides dense transition-level feedback and steers agents toward behaviors that are both task-effective and physically grounded. Extensive experiments across diverse agentic tasks show that our method substantially improves policy robustness and generalization over prospective-only world modeling baselines.

Yongjiang Liu, Jie Zhang, Haoyue Zhang +3

The Hong Kong University of Science and Technology · The Hong Kong Polytechnic University · China University of Geoscience

World Model Learning World Models World Model-Based Planning Inverse Dynamics Robustness of VLA Models RL for VLMs

Sep 30, 2026cs.CV

Frame Differential On-Policy Self-Distillation for Video Reasoning

Reinforcement learning (RL) has substantially improved the reasoning ability of multimodal language models through verifiable rewards and increasingly fine-grainedvisual or temporal credit assignment. In video reasoning, however, current RL methods typically train with a fixed sparse frame budget: increasing the number of frames makes autoregressive rollouts expensive, while too few frames may miss temporally localized events and fine-grained visual details. We present \textbf{Frame Differential On-Policy Self-Distillation (FD-OPSD)}, which transfers the useful evidence of dense frame observations to a sparse frame policy during RL training. FD-OPSD compares the policy's token level preferences for the same sampled response under sparse and dense views, and distills the resulting frame differential signal without an external teacher or dense autoregressive rollout. The method preserves sparse-frame rollouts and leaves inference unchanged. Across Qwen2.5-VL-7B and Qwen3-VL-4B on six video reasoning benchmarks, FD-OPSD yields higher overall average performance than the strongest corresponding GRPO, T-GRPO, or Video-KTR baselines across the 16, 32, and 64 frame evaluation settings. These results show that dense visual evidence can be transferred selectively during training through token level self-distillation while retaining sparse frame rollouts and unchanged inference.

Haiying He, Xin Zheng, Shaoli Hu +4

HKUST · NKU · SEU +1

On-Policy Self-Distillation Video Reasoning VLM Reasoning RL for VLMs

Sep 30, 2026cs.CV

SpatialCORE: Confidence-Aware Grounded Spatial Reasoning in Large Vision--Language Models

Large Vision-Language Models (LVLMs) have made remarkable progress across visual perception tasks, yet spatial reasoning remains a persistent weakness, especially for questions that require reasoning over visual space. Recent spatial-reasoning methods incorporate generated grounding, where models predict bounding boxes, masks, or other localization outputs for task-relevant objects as part of their reasoning trace. However, these approaches typically optimize final-answer correctness alone, allowing correct answers to be rewarded even when the model does not reason from confidently localized task-relevant objects. We introduce SpatialCORE (Spatially COnfident REasoning), a post-training framework that turns the model's own confidence in generated grounding into a learning signal for spatial reasoning. Its central idea is to reinforce grounding that is both accurate and confident, encouraging the model to reason from confidently localized task-relevant objects. SpatialCORE realizes this through a self-regulating spatial reward that weights each predicted bounding box's matching quality by its coordinate-token confidence. An answer gate further ties grounding optimization to final-answer correctness. SpatialCORE achieves state-of-the-art results among open-source and specialized spatial reasoning models across diverse benchmarks, and transfers effectively in zero-shot settings to unseen data distributions. The source code is available at https://github.com/rafiibnsultan/SpatialCORE.

Rafi Ibn Sultan, Xiangyu Zhou, Md. Sajid Alam Chowdhury +4

Department of Computer Science, Wayne State University · Department of Radiation Oncology, Henry Ford Health · Department of Electrical and Computer Engineering, The Ohio State University +1

Visual Spatial Reasoning Large Vision-Language Models Vision-Language Grounding RL for VLMs

Sep 29, 2026cs.CV

MOBA-VL: Event-Localized Multi-Turn Reinforcement Learning for Real-Time MOBA Commentary

Real-time commentary for Multiplayer Online Battle Arena (MOBA) esports requires a vision-language model (VLM) to narrate a live match second by second, both fluently and accurately. Existing streaming VLMs sound natural but often miss key events such as kills and objectives. To address this limitation, we use game telemetry, which records exactly when each event occurs, as a supervision signal. We introduce MOBA-VL, a 9B-parameter model trained on this signal with event-localized multi-turn reinforcement learning, which rewards the turns that describe each event. We also collect MOBACast, 860 professional matches (about 460 hours) across three MOBA games with word-level timestamped commentary, and MOBACast-Bench, a benchmark from held-out tournaments. On MOBACast-Bench, MOBA-VL achieves the highest Overall score on full matches (63.25 vs. 55.12 for StreamingVLM) and clips (63.45 vs. 56.22 for DeepSeek-V4.1-Flash). Event-localized credit also raises event recall from 34.5 to 42.1 over supervised fine-tuning. Code and data will be released, and demos are available on an anonymous project page at https://moba-vl.github.io.

Shengyun Zhong, Xinkang Zhao, Ziyuan Chu +1

Zhejiang University, China · Northeastern University, USA

Reinforcement Learning Video-Language Models RL for VLMs

Sep 29, 2026cs.CV

VISTA: Internalizing Collective Visual Experience via On-Policy Distillation for Active Multimodal Agents

Active multimodal agents use visual tools to acquire task-relevant evidence while reasoning. Although reinforcement learning samples multiple interaction trajectories per input, outcome-based objectives primarily use the group to estimate scalar advantages, leaving complementary visual discoveries underused. We introduce VISTA, which internalizes collective visual experience through on-policy distillation by turning observations from same-input rollouts into shared supervision. Collective visual experience distillation (CVED) organizes these observations with their interaction context and aligns them with individual decisions, while heterogeneity-aware policy improvement (HAPI) reinforces successful trajectories and provides experience-guided distillation for unsuccessful attempts. An experience-conditioned teacher evaluates the student's sampled response prefixes, allowing discoveries from one trajectory to guide learning in another without replacing the student's original history or generating new target trajectories. The trained agent retains its visual tools and acts using its own interaction history. VISTA achieves the strongest average performance among the evaluated active multimodal agents of comparable size and consistently outperforms same-backbone training baselines across fine-grained perception and general reasoning tasks, demonstrating the value of collective experience for active multimodal learning.

Zheng Jiang, Houde Qian, Yiming Chen +5

Tsinghua University

On-Policy Distillation Multimodal Agents RL for VLMs Tool Use

Sep 29, 2026cs.AI

Improving OCR Faithfulness via Gated and Attenuated On-Policy Distillation

Vision-language models may rewrite anomalous text in images into linguistically plausible expressions, compromising OCR transcription faithfulness. Sequence-level task rewards and local teacher guidance are complementary, but guidance from the same teacher may not remain equally effective as the student improves. Offline analysis shows that supervision from a fixed teacher becomes progressively less favorable as the student improves, both across training checkpoints and across response groups with different task rewards. Motivated by this observation, we introduce GAD-RL, which adaptively regulates teacher supervision during joint post-training according to the student's current task performance and local distributions. A frozen teacher conditions on reference transcriptions and student-generated prefixes. GAD-RL disables distillation for response groups containing an output with task reward at least 0.95 and continuously attenuates distillation strength as group-mean reward increases. It also weights forward KL by the student's probability of the teacher's Top-1 token, moderating local auxiliary updates when student support for that candidate is low. On Qwen3.5-2B, GAD-RL achieves 59.92% Micro Recall on CHAOS-Bench, surpassing GRPO and GRPO+OPD (fixed-weight) by 8.45 and 4.43 percentage points, respectively, while achieving an Overall score of 91.18 on OmniDocBench v1.6.

Baode Wang, Zuming Huang, Kexuan Ren +2

VLM Distillation On-Policy Distillation Optical Character Recognition RL for VLMs

Sep 29, 2026cs.CV

MG-Thinker: Bi-Axial Self-Reflection for Multi-Image Reasoning Grounding

Reinforcement learning (RL) has recently delivered substantial gains in multimodal reasoning, opening a promising route for fine-grained visual perception. Yet for multi-image reasoning grounding (MRG), reasoning over real-world multi-image contexts toward pixel-precise localization, existing RL-based approaches overlook two characteristics intrinsic to this paradigm: a coarse-to-fine hierarchical reasoning pattern, and heterogeneously distributed task--sample difficulties. In this work, we present MG-Thinker, a post-training RL framework that advances a new MRG paradigm featuring such hierarchical reasoning, supported by a curated 25K MRG dataset with task-adaptive Chain-of-Thought (CoT) annotations that elicit multi-perspective evidence before conclusion. To remedy the heterogeneous task--sample difficulties, we further propose Bi-Axial DAPO (BiA-DAPO), which decomposes rollout advantages along an intra-group signal axis and an inter-group competence axis through two complementary mechanisms, both grounded on our defined candidate pool for stable group-level statistics. Extensive experiments show that MG-Thinker achieves state-of-the-art performance on multi-image reasoning grounding while consistently improving generalization across multi-image understanding and diverse multimodal benchmarks.

Heyu Huang, Chi Chen, Zonghao Guo +3

Huazhong University of Science and Technology Wuhan, China · Tsinghua University Beijing, China

Multimodal Grounding Multimodal CoT Reasoning Visually Grounded Reasoning RL for VLMs

Sep 29, 2026cs.AI

Seek Before You Move: Evidence Seeking for Progress Grounding in Vision-Language Navigation

Vision-Language Navigation (VLN) requires agents to continuously ground task progress from long-horizon instructions and partial egocentric observations. Existing VLM-based navigation agents typically reason only over available observations and may remain confident even when task-relevant evidence is missing. For example, an agent may confidently proceed forward and get lost even though the landmark indicating the next turn lies outside its current field of view. We term this failure mode Progress Myopia: the agent fails to recognize unreliable progress grounding and continues acting on insufficient evidence. To address it, we propose SeekVLN, an evidence-seeking framework that couples semantic progress reasoning with active acquisition of task-relevant observations. SeekVLN is trained in two stages: First, Future-guided Reverse Generation (FRG) uses future expert actions to augment offline expert trajectories with supplementary views and evidence annotations. Supervised fine-tuning on these trajectories establishes a prior for evidence seeking and progress reasoning without additional expert interaction. However, imitation alone does not reveal whether seeking improves subsequent navigation. We therefore introduce Counterfactual Contrastive Policy Optimization (C2PO) for reinforcement fine-tuning. By comparing each evidence-seeking branch with a counterfactual direct-navigation branch from the same state, C2PO uses a contrastive reward to assign credit to seeking decisions based on subsequent navigation benefit. Experiments on simulated benchmarks show that SeekVLN achieves state-of-the-art performance, improving success rate by 12.7% and 7.5% over the base model on R2R-CE and RxR-CE, respectively. Both simulated and real-world evaluations exhibit human-like evidence-seeking behaviors for more reliable progress grounding.

Zhimin Wang, Meiyuan Zhu, Duo Wu +8

Tsinghua University · Pengcheng Laboratory · South China University of Technology +3

Vision-Language Navigation RL for VLMs

Sep 29, 2026cs.CV

HaPRL: Human-Anchored Process Reinforcement Learning for Visual Search Agent

Multi-turn visual search agents answer questions about high-resolution images by iteratively deciding where to look. Reinforcement learning for these agents rewards only the final answer, leaving the search process unsupervised. Consequently, faulty routes in which the reasoning process is erroneous yet the final result is correct arise frequently, which in turn leads to ineffective training, i.e., scaling along the wrong paths. In this paper, we introduce HaPRL, the first framework to reinforce the search process with human search behavior. We first build an annotation platform and collect 1K+ human-annotated data with fine-grained behavioral signals. During training, a carefully designed judge scores each rollout with task-adaptive weights, anchored on the distilled trace of how a human annotator actually searched the same image. Extensive experiments show that HaPRL consistently outperforms outcome-based RL, and early-stage process supervision yields 6.7x more improvement in subsequent outcome-based scaling. Our results also demonstrate the importance of aligning model behavior with human process annotation signals, which offer new insight into the training of foundation models.

Zhangquan Chen, Yaoxin Niu, Xiang An +5

Tsinghua University · Peng Cheng Laboratory · LMMs-Lab +2

Human-in-the-Loop Annotation Process Supervision Visual Search RL Post-Training RL for VLMs

Sep 29, 2026cs.AI

Learn from the Gap: Differential-Aware Advantage Pruning with Adaptive Rollout Sampling for GRPO

Recently, Group Relative Policy Optimization (GRPO) and its variants have been developed for policy optimization and demonstrated notable performance gains. However, these methods usually incur substantial computational overhead due to per-question multi-rollout sampling and repeated per-token probability evaluation across rollouts. Furthermore, low-information or highly homogeneous trajectories can degrade downstream learning signal efficiency, hindering model optimization and limiting final performance. To address these issues, we propose FastRL, a novel plug-and-play reinforcement learning framework that simultaneously improves training efficiency and the effectiveness of policy learning. Specifically, 1) We introduce an advantage-aware pruning strategy to selectively preserve high-advantage trajectories while maximizing inter-trajectory gradient diversity. 2) Then, we design an adaptive rollout sampling mechanism to dynamically adjust the sampling scale across different training stages based on historical pruning distributions, balancing exploration adequacy and computational efficiency. Experiments demonstrate that FastRL can be seamlessly integrated into GRPO, DAPO, and GSPO variants, achieving an average 2.07

\times

training speedup on Geometry3K and GeoQA8K-R1V, along with an approximately 1.64% improvement in average accuracy on visual reasoning benchmarks. Source codes will be available at https://github.com/Nicozwy/FastRL.

Jiahua Yang, Zhiwei Yang, Xianpeng Zhang +7

Guangdong Institute of Smart Education, Jinan University, Guangzhou, China · OPPO AI Center, Shenzhen, China · Ragentile Intelligence Inc, Edmonton, Canada

Group Relative Policy Optimization RL for VLMs

Sep 29, 2026cs.CL

Momentum-Coupled Rubric Adaptation for Detailed Image Captioning

Detailed image captioning requires accurate and comprehensive descriptions of fine-grained visual content, yet caption quality spans factual accuracy, information coverage, and clarity. Compared with conventional methods that rely mainly on high-quality supervision or holistic rewards, rubric-based reinforcement learning decomposes these requirements into explicit criteria and provides targeted, structured feedback. However, existing methods often use separate models for caption generation, rubric construction, and judging, which may lead to inconsistent interpretations across roles. Some dynamic rubric methods alternate updates between the caption policy and rubric generator while keeping the judge fixed, but staged optimization may still leave rubric construction and judging out of step with policy optimization. We propose MoCo Rubric, a two-stage framework that coordinates these roles. First, role-conditioned, shared-parameter multi-task supervised fine-tuning equips a single vision--language model to serve as the Caption Policy, Rubric Generator, and Rubric Judge. Then, the Generator constructs rubrics online from captions sampled by the current Policy, reference captions, and image evidence. The Judge provides rubric-based rewards, and only the Policy receives GRPO updates. As Policy updates change the candidates being evaluated, we use an exponential moving average of the Policy parameters to update one momentum model shared by the Generator and Judge. This gradual transfer lets both rubric roles track Policy updates without separate RL optimization while smoothing parameter changes that could disrupt their rubric capabilities under direct synchronization. Across five captioning benchmarks, MoCo Rubric achieves an average pairwise win rate of 72.83%, the best mean rank in blind ranking, and the highest average score in caption-based question answering.

Zhenwen Ji, Lei Jin, Shanyong Wang +6

Xiaohongshu Inc. · the Joint SDU-NTU Centre for Artificial Intelligence Research (C-FAIR), Shandong University

Image Captioning Rubric-Based RL RL for VLMs

Sep 29, 2026cs.CV

DSPO: Diversity-aware Subjective Policy Optimization for Robust Emotional Reasoning

Reinforcement Learning has significantly advanced the complex reasoning capabilities of MLLMs. However, prevailing RL algorithms suffer a severe failure in emotion reasoning tasks. These methods heavily rely on deterministic hard-label supervision and point-wise isolated evaluation, creating a fundamental gap with the inherently subjective and continuously distributed nature of human emotions. Furthermore, unlike explicit physical objects, emotional states are deeply implicit within visual cues. This abstract nature exacerbates visual hallucinations in MLLMs, leading to plausible yet ungrounded emotional evidence. To address these limitations, we propose Diversity-Aware Subjective Policy Optimization (DSPO), a reinforcement learning framework that jointly promotes subjective affective coverage and visual grounding. First, we construct a context-grounded emotional distribution prior in the VAD space by combining the lexical prior of the annotated emotion with image-specific contextual information. Based on this prior, we introduce a Distribution-Aligned Emotional Diversity Reward (DEDR), which measures the leave-one-out marginal contribution of each candidate emotion within a rollout. DEDR rewards candidates whose inclusion brings the predicted affective set closer to the context-grounded prior, thereby preserving plausible subjective interpretations without encouraging unconstrained dispersion. We further develop Counterfactual Visual Intervention Gating (CVIG), which masks the visual region highlighted in the reasoning process and uses the resulting candidate-wise probability changes to reduce the weights of interpretations unsupported by visual evidence. Extensive experiments demonstrate that DSPO achieves state-of-the-art performance across multiple public benchmarks, especially on the cross-domain performance, i.e., improving +10.8% on average cross-domain accuracy than EMO-R3.

Cheng Ye, Weidong Chen, Bingyan Xu +1

University of Science and Technology of China, Hefei

VLM Robustness Multimodal Reasoning VLM Reasoning Visual Grounding RL for VLMs

Sep 28, 2026cs.CV

PIVOT: Pivot-Aware On Policy Self Distillation for Multi-Turn VLM Agents

Reinforcement learning with verifiable rewards (RLVR) via Group-Relative Policy Optimization (GRPO) is widely used for multi-turn VLM agent training, yet it suffers from zero-gradient silence on uniform failures and coarse episode-level credit assignment. While On-Policy Distillation (OPD) and On-Policy Self-Distillation (OPSD) mitigate sparse rewards using hindsight information, their underlying mechanisms remain poorly understood. Through controlled counterfactual rollback probes across five multi-turn VLM agent benchmarks, we reveal that performance gains in OPSD/OPD are largely driven by physical state rollback at the pivot step, defined as the first unrecoverable action without remaining step budget. However, physical state rollbacks are computationally prohibitive and infeasible in real-world environments. To bridge this gap, we present Pivot-Aware Internalized Visual On-Policy Training (PIVOT), an RL framework that internalizes pivot localization and state restoration directly into token-level parameter updates, eliminating environment rollbacks during RL training and additional skill hints at test time. PIVOT unifies three functional roles within a single architecture: a failure Analyzer non-invasively localizes the pivot step and diagnoses failure modes from visual trajectory collages and action logs; a detached Teacher re-scores failed tokens under this privileged diagnostic context; and a Student optimizes joint GRPO and confidence-gated OPD objectives. At test time, both Teacher and Analyzer branches are stripped. Evaluated on five multi-turn VLM agent tasks across cognitive grid puzzles, 3D embodied control and navigation, and generative reasoning, PIVOT achieves 0.90 overall accuracy on Qwen2.5-VL-3B (+8% over SFT+GRPO baseline and +5% over previous SOTA) and scales to 0.92 on Qwen3-VL-2B (+12% over SFT+GRPO baseline).

Jiazhou Zhou, Hu Zhou, Yucheng Chen +3

AI Thrust, The Hong Kong University of Science and Technology (Guangzhou) · International Digital Economy Academy (IDEA) · The Hong Kong Polytechnic University +2

On-Policy Self-Distillation Reinforcement Learning with Verifiable Rewards RL Post-Training Policy Learning RL for VLMs

Sep 28, 2026cs.CV

Beyond Saying Less: Fine-Grained Alignment for Informative and Faithful Vision-Language Models

Object hallucination remains a major challenge for large vision-language models. While off-policy preference optimization proves to be an effective solution, on-policy reinforcement learning provides a more promising direction as it directly targets a model's current failure modes. However, we find that without fine-grained reward formulation and allocation, on-policy optimization often falls into an easy shortcut: reducing hallucinations merely by saying less---making fewer valid claims. To comprehensively resolve this, we propose a fine-grained alignment framework that couples dense reward signals at the data level with precise credit assignment at the algorithmic level. Specifically, we first construct the Dense Object Presence and Absence (DOPA) dataset to address sparse annotations that prevent valid object claims from being verified and rewarded. DOPA exhaustively annotates the deterministic presence and absence of every concept across an expanded vocabulary, significantly increasing the density of reliable reward signals during on-policy rollouts. Second, we propose Subsentence-level Credit Assignment for on-Policy Optimization (SCAPO) to prevent response-level shared advantages from allowing local hallucinations to compromise all other valid outputs within the same response. By assigning credit to each subsentence independently based on its object claims, SCAPO can precisely reinforce faithful generations and penalize hallucinations. Furthermore, we leverage the resulting faithful image descriptions as auxiliary context to transfer generative gains to discriminative tasks. Experiments demonstrate that our method produces highly informative, faithful descriptions in generative tasks while yielding clear performance gains on discriminative evaluation.

Xingming Long, Jie Zhang, Yuecong Min +2

State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences · University of Chinese Academy of Sciences · Zhongguancun Academy

Credit Assignment in RL Step-Level Credit Assignment in RL Object Hallucination in VLMs Vision-Language Alignment Faithfulness in VLMs VLM Hallucination Mitigation+1

Sep 28, 2026cs.LG

Adjoint Guidance Flow: Amortized Critic Guidance for VLA Policies

Flow-based Vision-Language-Action (VLA) policies are typically trained by behavior cloning and thus do not explicitly optimize long-term task return. Critic guidance steers generation toward higher-value actions, but existing methods differentiate the critic through a one-step surrogate of the sampler and back-propagate a critic ensemble at every flow step. In contrast, here we propose Adjoint Guidance Flow (AGF), which amortizes trajectory-aware critic guidance into a lightweight guidance network while preserving the pretrained VLA policy. Specifically, we formulate critic-guided flow generation as a deterministic optimal control problem, whose optimal guidance is a costate that carries the terminal critic gradient back through the remaining flow, and regress the guidance network onto this costate while keeping both the VLA and critic frozen. This design provides favorable memory and throughput scaling during training, and inference needs one guidance-network forward pass per step, without the critic ensemble, back-propagation, or adjoint computation. Across LIBERO, RoboCasa, and LIBERO-Pro, AGF consistently improves pretrained VLAs, remains competitive with critic-guidance and policy-fine-tuning baselines, and is the most robust method when a single guidance strength is deployed across tasks. Compared with QGF, AGF runs

3.6\times

faster per guidance step with

7.0\times

fewer parameters, with comparable and even better performance, showing that critic guidance can be trajectory-aware and lightweight.

Jeongsol Kim, Youngjun Jun, Kyumin Choi +6

KAIST · SKKU · GIST

Efficient VLA Models Vision-Language-Action Models RL for VLMs

Sep 28, 2026cs.LG

The Low-Rank Structure of VLA Reinforcement Learning

Reinforcement learning (RL) is increasingly used to post-train vision-language-action (VLA) models, yet how RL reshapes these policies remains poorly understood. We find that RL across widely used flow-based VLA models, including

π_{0.5}

and GR00T~N1.5/N1.6, on LIBERO, ManiSkill, MetaWorld, and CALVIN induces substantially lower-rank parameter updates that are highly concentrated in the action expert's Timestep Modules, a small and previously overlooked component. Through systematic module-replacement experiments, we further show that these modules capture a disproportionate share of the performance gains from RL. We then characterize what is encoded in these Timestep Modules. First, we show that RL specializes them to the discrete denoising timesteps used during rollouts, and that this discrete-timestep training underlies the low-rank updates. Second, we find that among their outputs, the shift vector changes most distinctly under RL, and through probing, we show that shift update directions strongly predict task success (ROC-AUC up to

99.6\%

). Third, we find that the geometry of shift updates reflects task relationships, as their pairwise similarity correlates with cross-task transfer patterns. Building on these findings, we show that steering along shift update directions further improves RL-trained policies without additional RL training. Overall, we provide a systematic understanding of how RL reshapes VLA policies by studying how learned signals are encoded in parameter space, offering insights into more efficient and interpretable VLA post-training.

Minjae Oh, Yoonah Park, Jongwon Lim +1

Graduate School of Data Science, Seoul National University

Vision-Language-Action Models RL Post-Training RL for VLMs

Sep 28, 2026cs.CV

Reinforcement Learning from Intermediate Renders for Image-to-Code Generation

Reinforcement learning is increasingly used to post-train vision-language models for image-to-code generation, such as generating SVG code from a reference image, by optimizing rewards computed from the final rendered output. However, relying on a single terminal reward provides sparse feedback that is poorly aligned with the contribution of individual tokens. A generated program may contain operations that accurately reproduce some parts of the target image alongside others that introduce errors, yet all tokens are trained from the same final outcome. We observe that many intermediate code prefixes are not only executable, but already produce meaningful partial renders that reflect progress toward the target. This property provides a natural source of denser supervision during generation. Based on this observation, we introduce IR4RL, an RL framework with a token-level render-progress reward that turns changes between intermediate renders into localized feedback for the generated sequence. We evaluate our approach on Image-to-SVG and Image-to-TikZ generation. Across both tasks, our method improves over supervised fine-tuning and standard GRPO, yielding new state-of-the-art open-source models. This shows that intermediate rendering provides a simple and effective source of process supervision for RL post-training of image-to-code models.

Omri Kaduri, Kate Feingold, Phillip Isola +1

Weizmann Institute of Science · MIT

RL for Code Generation Code Generation Process Supervision RL for VLMs

Sep 28, 2026cs.CV

Rethinking Latent Visual Reasoning: Grounding Latent Reasoning in Visual Evidence

Latent visual reasoning (LVR) enables multimodal large language models (MLLMs) to perform intermediate computation in continuous latent tokens rather than expressing every reasoning step in words. However, unlike textual CoT, latent reasoning is not directly observable, making it difficult to supervise what latent tokens learn. In this work, we first conduct a thorough analysis of latent-token behavior and identify a latent evidence-credit gap: latent tokens respond only weakly to image perturbations that alter the correct answer. We hypothesize that this issue stems from the lack of explicit supervision during GRPO training. These findings suggest that a final-answer reward provides too little guidance on what visual evidence to preserve or how credit should be assigned across latent tokens. To bridge this gap, we propose ReaLVR, which brings visual-evidence supervision to the model's own free-running latent trajectories. ReaLVR contrasts correct and model-generated wrong answers to determine where stronger supervision is needed, and relevant and mismatched visual evidence to specify what to preserve. Across three model families, ReaLVR consistently outperforms evaluated LVR baselines, achieving the highest five-task average of 63.7% on Qwen2.5-VL-7B. Crucially, we are the first to scale visual reasoning in latent space, showing that our framework continues to deliver robust improvements at frontier model scales up to 235B. Further analyses show more question-sensitive latent-token positions, stronger alignment with relevant visual regions, and greater fixed-context dependence on the most attended latent tokens.

Xi Xiao, Tianchen Zhao, Youngeun Kim +10

University of Alabama at Birmingham · Amazon AGI · Work done during an internship at Amazon AGI.

Large Vision-Language Models Latent Visual Reasoning Visually Grounded Reasoning RL for VLMs

Sep 27, 2026cs.CV

TTRSD: Test-Time Reinforcement Learning with Self-Distillation for Vision-Language Models

Test-time reinforcement learning enables vision-language models (VLMs) to adapt using unlabeled inputs. However, repeated sampling under fixed visual conditions can reinforce shared perceptual errors, while sequence-level rewards fail to isolate visual perception the foundational bottleneck that anchors multimodal reasoning risking the degradation of pre-trained reasoning capabilities. We propose TTRSD, a test-time reinforcement learning framework combining multi-view answer-level self-distillation with visual contrastive token selection. A shared policy aggregates teacher predictions across original, cropped, and downsampled views into an answer distribution. Student trajectories generated from the original image receive rewards based on the support for their final answers in this distribution. To allocate this feedback precisely toward perceptual bottlenecks, we compare the log-probabilities of the same sampled tokens under original and visually ablated inputs while holding their textual prefixes fixed, selecting visually sensitive positions for policy-gradient updates. TTRSD separates update direction, determined by group-relative advantages, from update position, determined by visual sensitivity, without requiring ground-truth labels, external verifiers, or a separate teacher. With only 20 unlabeled adaptation samples, TTRSD improves performance across seven benchmarks and three VLMs, raising InternVL3-2B's MMMU accuracy from 35.79% to 49.32%(+13.53%), demonstrating cross-dataset generalization while preserving inherent reasoning integrity.

Shuning Wang, Zhiheng Wu, Xun Zhou +8

Zhejiang University · Hong Kong University of Science and Technology · Harbin Institute of Technology +1

Test-Time Adaptation of VLMs VLM Distillation RL for VLMs

Sep 23, 2026cs.AI

DEEPO: Dual-Entropy Enhanced Policy Optimization for Hallucination in MLLMs

Reinforcement learning (RL) is widely used to sharpen reasoning in multimodal large language models (MLLMs), yet its effect on hallucination is uneven. We trace this to two weak points in the \emph{correction chain} from reward to parameter update. At the rollout level, hard queries---those with high semantic entropy---frequently produce unanimously wrong sample groups, collapsing the group-relative advantage to zero exactly where hallucination risk is highest. At the optimization level, confident-but-wrong tokens are gradient-invisible: a categorical policy's expected score-gradient norm vanishes as its distribution sharpens, so the predictions that most need correction receive the weakest updates. We propose Dual-Entropy Enhanced Policy Optimization (DEEPO), a dual-stage enhancement combining signal variance regularization with gradient preconditioning: semantic-entropy-triggered expert prefixes inject grounded continuations on high-uncertainty queries, providing direct supervision and restoring advantage variance, while advantage-sign-aware Renyi preconditioning counteracts logit-level saturation so correction reaches confident errors in the operational confidence regime. Both branches improve over GRPO individually; their interaction is statistically significant on VideoMMMU---the most complex long-horizon task in our evaluation suite (+4.0$, 95% CI [1.1, 6.9])---and additive elsewhere. DEEPO reduces hallucination while preserving accuracy and training stability.

Yingxuan Zhuang, Miao Pan, Wangjie Gan +6

Zhejiang University · Shanghai Artificial Intelligence Laboratory

Multimodal Large Language Models VLM Hallucination Mitigation RL for VLMs

Sep 22, 2026cs.CV

Video-HopChain: Multi-Hop Questions and Confidence-Gated Exploration for Video Reasoning Models

HopChain has shown on still images that multi-hop data synthesis improves vision-language reasoning, because long chain-of-thought reasoning exposes errors that compound across steps, while most data used for reinforcement learning with verifiable rewards (RLVR) rarely demands a chain of visual evidence, so these weaknesses are likely to stay unexposed. We observe the same problem in video, where this framework has not yet been explored. We therefore build Video-HopChain, a dataset of 22,550 multi-hop video questions over 13,378 videos, together with a held-out benchmark of 1,000 questions. Each question chains three to six yes/no questions about moments in one video, and each yields one of two integers depending on its answer. The final answer is the sum of these integers, so an exact match on that sum gives the verifiable reward that RLVR needs. We first train Qwen3-VL-8B with GRPO on a standard video dataset, and a second stage on Video-HopChain then raises the mean over eight video understanding and reasoning benchmarks from 55.4 to 57.9 and improves every one of them. Training on such a dataset, however, exposes a known limitation of GRPO: its learning signal comes from the reward variance within a group, so hard questions whose rollouts are all incorrect and easy questions whose rollouts are all correct both leave the group with no gradient. To recover these groups at the same compute budget, we introduce Confidence-Gated Exploration (CGE). With 8 rollouts per question, CGE samples the first 4 as usual. If these 4 are either all correct or all incorrect, it samples the last 4 with the policy's most confident token masked inside the reasoning span, and removes the masked positions from the loss while all 8 rollouts enter the advantage. With CGE, the mean rises further to 59.3. We release the dataset, the checkpoint, and the data generation and training code.

Trung Nguyen Quang, Yuhao Dong, Shuo Sun +4

S-Lab, Nanyang Technological University (NTU) · Johns Hopkins University · NTU

Group Relative Policy Optimization Video Reasoning Reinforcement Learning with Verifiable Rewards RL for VLMs