Reinforcement_Learning
-
When Pairwise Preferences Cycle: A Reliability-Anchored Preference Walk for Video GRPO
2026-10-14 12:30:39Xinrui Lin, Fan Li, Shumin Wang, Dehong Kong, Yanyong ZhangAbstract
Pairwise preference feedback has become a widely adopted supervision signal for reinforcement learning in video generation models, as pairwise comparisons provide a more reliable alternative than pointwise scalar ratings for evaluating the quality of generated videos. Specifically, we define a preference graph for each GRPO group under a pairwise evaluator, where nodes represent sampled rollouts and edges encode pairwise preferences. Auditing 6,976 on-policy comparison graphs, we find that cyclic preferences are prevalent and persist throughout training. Naive pairwise reward aggregation based on average win rates discards the underlying comparison structures and treats each rollout independently, which may introduce unreliable advantage estimates for certain rollouts and consequently affect reinforcement learning optimization. Therefore, we propose the \emph{Reliability-Anchored Preference Walk} (\ours{}), a graph-aware advantage aggregation method that strengthens the global preference signal from win rates with reliable relational evidence from cyclic preference graphs through a reliability-guided preference walk. Under matched evaluator, comparison protocol, group size, prompt schedule, and query budget, \ours{} requires no additional evaluator queries beyond the observed comparisons within each rollout group and improves over Pairwise-GRPO across controlled comparisons, transfer settings, and blind human evaluations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.20089
PDF
-
SAM3RL: Memory Control via Reinforcement Learning for Visual Object Tracking
2026-10-14 10:17:47Tom\'a\v{s} \v{C}\'i\v{z}ek, Matej Straka, Illia Volkov, Mat\v{e}j Kripner, Ervin Maci\'c, Jiri Matas, Klara Janouskova, Martin SchmidAbstract
Segment Anything Model 3 (SAM 3) is a foundation model for image and video segmentation that sets the state of the art in visual object tracking. Its tracker, inherited from SAM 2, stores information from past frames in a memory bank, managed by a simple First-In-First-Out rule, allowing temporal consistency in complex videos. Several recent methods improve SAM 2 with hand-crafted memory update rules, each targeting a specific failure mode, such as distractors, occlusions, or object motion. However, their gains transfer only partially to the stronger SAM 3 backbone. We instead present a more general, data-driven approach that frames memory control as a sequential decision-making task. SAM3RL is an extension of the SAM 3 tracker with a lightweight memory controller, trained by reinforcement learning, that decides which frames enter the memory bank in order to maximize tracking performance. SAM3RL sets a new state of the art on 10 of 12 tracking and video object segmentation benchmarks, outperforming all hand-crafted rules without any runtime overhead. The same approach applied to SAM 2 gives SAM2RL, which outperforms hand-crafted rules on 9 of the 12 benchmarks, showing the generality of our method across model generations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19931
PDF
-
MolHop: Scaffold-Hopping Molecular Design with Agentic Reinforcement Learning
2026-10-14 10:11:10Yanyi Su, Chenyan Wang, Yulei Gong, Yeyun Chen, Xiaohui Yang, Jun ChengAbstract
Molecular design is often formulated as an optimization problem over existing molecular spaces. However, molecular optimization and the discovery of structurally distinct replacement molecules represent two fundamentally different search problems. LLM-based molecular design agents typically perform the former, conducting local refinement within the neighborhood of existing structures to improve target properties through structural modifications. In contrast, the latter requires exploration of a broader chemical space to discover molecules with entirely different scaffolds while preserving the functional characteristics of reference molecules. We propose MolHop, an agentic framework for cross-scaffold molecular discovery that enables models to navigate and edit within a pre-organized chemical space. MolHop organizes approximately 1.6 million molecular scaffolds from commercial molecules into a coarse-to-fine hierarchical action space. The agent selects new exploration regions through scaffold navigation actions and adjusts candidate properties through molecular editing. The framework maintains historical trajectory memory to support long-horizon search processes. On drug molecule and commercial compound tasks, connecting Grok-4.6 and DeepSeek-Flash-V4.1 to the MolHop framework substantially improves their cross-scaffold joint success rates. We develop MolHop-9B through agentic reinforcement learning, and MolHop-9B outperforms generative baselines including GPT-5.6-sol, Claude-5-Opus, and Gemini-3.1-Pro, while generating candidate molecules with larger scaffold differences from the reference molecules. These results demonstrate that effective molecular discovery beyond the traditional optimization paradigm requires structured chemical and action spaces, enabling agents to explore new structural regions rather than merely refine existing molecules.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19920
PDF
-
Not All Criteria Matter: Importance-Aware Criteria Weighting for Rubric-Based LLM Evaluation
2026-10-14 10:05:43Haocheng Yang, Licheng Pan, Yiming Dong, Xiaoxi Li, Yuan Lu, Zhenhao Chen, Jiajun Fan, Shuai Shao, Chunyuan Zheng, Hao WangAbstract
Rubric-based evaluation assesses the responses of large language models (LLMs) by aggregating the verification results of explicit criteria, yet not all criteria matter equally. However, existing methods commonly aggregate the verification results with uniform weighting, which assigns minor and overlapping criteria the same weight as essential ones and thus produces misleading evaluation scores. Although some methods prompt an LLM to directly assign weights, direct weighting requires a holistic judgment over all criteria that LLMs struggle to make reliably. To address these limitations, we propose Importance-Aware Criteria Weighting (ICW), which derives the weights of criteria from pairwise importance comparisons made by an LLM. Specifically, ICW transforms the holistic judgment over all criteria into a series of pairwise importance comparisons, checks and revises inconsistent comparisons, and derives the weights from the principal eigenvector of the comparison matrix. Extensive experiments demonstrate that ICW consistently outperforms uniform weighting and competitive weighting methods, and translates into better policy models in downstream reinforcement learning.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19915
PDF
-
BluePixel: Toward Native 4K Agentic Image Editing
2026-10-14 09:37:09Shanyan Guan, Shiyu Li, Xin Li, Zhizhou Chen, Yunzhe Xu, Yuehao Liu, Ruixue Shen, Shiju Zhang, Min Jin, Qiang Liu, Junbo Liu, Penggang Shi, Sheng Sun, Xiao Wang, Yong Zhang, Zhongang Qi, Wei Li, Chao Ma, Ying Tai, Yanhao GeAbstract
Instruction-driven image editing has reached native 4K, yet how such editors are trained remains undisclosed, and pretraining one from scratch is beyond most teams. The task itself compounds the difficulty: a single end-to-end editor cannot make precise parametric adjustments, match specialized experts, or interpret the vague instructions users actually write. We argue that 4K editing is a system problem rather than a single-model problem, and present BluePixel, which couples a native-4K editing model with an agent harness; the complete training recipe is disclosed in this paper. The model makes 4K training practical: it turns a 1K-pretrained editor into a native-4K one through progressive VAE adaptation that keeps 4K training as affordable as 1K, absorbs a long tail of vertical editing skills through continual learning, and refines the aesthetic-fidelity trade-off with reinforcement learning. Distillation via source-anchored target projection (SATP) then compresses 50 denoising steps to 4, completing a 4K edit in 8.7 s on a single Kunlunxin P800 XPU while preserving task success rate and background consistency. The harness, BluePixel Agent, makes editing controllable: it decomposes user instructions into specialist and general operations, verifies every result with the same reward model that guided training, and retries within bounded attempts. On VINS-4KEval++ (3,305 native-4K cases), BluePixel ranks second among seven models, behind only GPT Image 2 in non-edited-region consistency.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19882
PDF
-
Learning Perception to Action Rules through Reinforcement Learning for Theory-of-Mind Reasoning
2026-10-14 08:40:06Soyoung Oh, Nursulu Sagimbayeva, Nicholas Jennings, Paulina Hussein, Mehwish Alam, Vera DembergAbstract
In Theory-of-Mind (ToM) reasoning, large language models may correctly infer the agent's mental state but fail to use that inference when predicting or judging the agent's subsequent actions. In this work, we investigate whether explicitly connecting perception, belief, and action with rules improves ToM reasoning. We generate concise \textit{if--then} rules that express relations between perception to action and use Group Relative Policy Optimization (GRPO) to train Qwen2.5 and Llama3 models across four ToM benchmarks. GRPO with rule generation outperforms prompting, supervised fine-tuning, and think reasoning. Moreover, rules generated by larger models ($\geq$3B) provide transferable knowledge that improves downstream performance of GPT-4o-mini. These findings suggest that explicit rules help models bridge the reasoning gap in ToM and support knowledge transfer across models.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19796
PDF
-
More Methods, Better Agents? A Reality Check on Self-Distillation in Agentic Reinforcement Learning
2026-10-14 08:36:31Linhao Zhong, Bozhen Fang, Xuehao Wang, Duochao Shi, Zekai Luo, Guangming Yao, Yi Yuan, Hao Chen, Jingdong Chen, Chunhua ShenAbstract
Self-distillation in agentic reinforcement learning has produced a growing number of methods. However, existing evaluations often reuse published baseline scores under differing experimental conditions, provide few direct comparisons among contemporaneous methods, and give limited attention to training stability and computational cost. These limitations obscure the source of reported gains and the practical strengths and weaknesses of competing methods. Costly agent training further motivates a common basis for comparison. To address these gaps, we compare ten representative methods against GRPO on ALFWorld, ScienceWorld, Search-QA, and WebShop across three model sizes. Our study comprises 132 training runs totaling 12,124.4 recorded H800 GPU-hours, with repeated evaluations over eight rollout seeds for the pretrained backbones and all eleven trained methods. Beyond final task performance, we examine training progress, stability, and measured computational cost. Our evaluation yields three key findings. First, gains over GRPO remain limited in scope, with no method consistently outperforming it across settings. Second, strong final performance can mask unstable training. Third, additional training cost does not consistently translate into better performance. We establish this benchmark to support fair, comprehensive comparison and guide future research toward reliable and cost-effective improvements in agent training. Project page: this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2610.19789
PDF
-
Soloist: A Generative Model for Blind Voice Separation in the Wild
2026-10-14 07:42:57Chen Yang, ChuFan Yu, Botian Jiang, Jingqi Chen, Wenxuan Wang, Yitian Gong, Zhaoye Fei, Xipeng QiuAbstract
Real-world voice separation must recover individual voices from overlapping speech, environmental noise, and music, yet isolated source recordings are rarely available for supervision. We introduce Soloist, a generative model for blind voice separation that autoregressively generates individual voice streams as audio-codec sequences conditioned on the mixture, without requiring a target-speaker query or a predefined voice count. We propose a two-stage training paradigm to learn from synthetic mixtures and adapt to real recordings. The first stage combines supervised separation on synthetic mixtures with isolated source targets and an auxiliary task that predicts speaker timelines from real recordings, learning who speaks when from natural acoustics. The second stage applies reinforcement learning to real recordings, with rewards supplied by a model judge that compares candidate separations based on voice purity and content preservation without isolated source targets. Across diverse scenarios, Soloist recovers clean speech from multi-speaker overlap and noisy recordings, and extracts singing vocals from music accompaniment while maintaining relatively high pitch accuracy and intelligibility. On real-world recordings with unknown voice counts, Soloist also separates voices well. Finally, we integrate Soloist into a full-duplex conversation processing pipeline that assembles separated outputs into continuous speaker tracks, demonstrating its practical utility in real conversations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19716
PDF
-
Function-Space Temporal Self-Distillation for Policy Optimization
2026-10-14 07:10:49Aryan Das, Arnab Bhattacharya, Vinay Kumar VermaAbstract
Reinforcement Learning with Verifiable Rewards (RLVR) improves reasoning in language models, but scalar outcomes provide coarse supervision over response trajectories. Self-distillation offers denser token-level guidance by comparing an unassisted policy with a feedback-informed self-teacher, yet the teacher evolves with the student, creating unstable targets and motivating additional regularization. We introduce Function-Space Policy Optimization (FSPO), a single-policy framework that combines current privileged adaptation with function-space temporal consolidation. FSPO stores sparse historical next-token distributions as fixed temporal targets, providing temporal separation without maintaining a persistent teacher model. It further introduces privileged disagreement correction to explicitly suppress locally overvalued actions, reliability-adaptive grounding that strengthens reference supervision when self-generated evidence repeatedly fails, and quality and coverage-aware temporal memory that retains verified behaviors and improves privileged demonstrations. Across reasoning and tool-use benchmarks, FSPO achieves the highest average accuracy under a 5-hour training budget, improving over the strongest baseline by approximately 4% points on average while reducing peak memory from about 70 GB to 56 GB on Olmo3-7B-Instruct.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19687
PDF
-
DEFlow: Deep Equilibrium Flow Maximum-Entropy Reinforcement Learning
2026-10-14 06:48:06Ruslan Ageev, Huaze Tang, Runqing Miao, Jiarong Li, Wenbo DingAbstract
Expressive flow policies can represent multiple high-value actions, but sampling them with few integration steps while retaining tractable action densities remains challenging. We introduce DEFlow, a deep equilibrium flow policy that replaces explicit Euler sampling with implicit Euler integration and solves all intermediate actions jointly as an equilibrium system. We combine implicit differentiation with a discrete change-of-variables formula for the final-action density, enabling off-policy training within soft actor-critic without backpropagating through solver iterations. DEFlow achieves an average relative improvement of approximately 10% in mean successful coverage over the strongest baseline across four Multi-Goal docking variants. It also achieves the highest mean time-averaged return on four of five manipulation tasks, while remaining broadly competitive across five MuJoCo tasks. These results support implicit flow policies for tasks requiring both diversity and precision, at the cost of additional nonlinear and linear solves. Primary
Abstract (translated)
URL
https://arxiv.org/abs/2610.16597
PDF
-
Model What Matters: Decision-Guided Selective World Modeling for Language Agents
2026-10-14 06:11:08Maolin Ran, Minda Chen, Di Huang, Wei Xiong, Fei Huang, Weiwen LiuAbstract
Reinforcement learning provides a direct mechanism for optimizing task objectives in large language model-driven agents, but task rewards are typically sparse trajectory-level signals that only evaluate task success, without directly supervising action-conditioned environment transitions, leaving the model with limited explicit understanding of action consequences and environment dynamics. Recent work introduces auxiliary supervision through policy and world modeling co-training, yet typically treats complete environmental observations uniformly as world modeling targets, overlooking the varying decision relevance of observation content. We propose DGS-WM (Decision-Guided Selective World Modeling), a decision-guided policy-world modeling co-training approach for language agents. DGS-WM employs a Decision-Guided Observation Selection component that filters observation units with decision value as world modeling targets based on their support for subsequent policy responses, and a Confidence-Adaptive Weighting component that adaptively adjusts supervision strength according to the model's prediction confidence on selected targets. Together, these components form a decision-aware auxiliary world modeling loss, jointly optimized with the RL loss on the same policy model. Across multiple benchmarks covering single-hop/multi-hop question answering and interactive web tasks, DGS-WM achieves consistent improvements under different model scales and RL algorithms, demonstrating more stable and effective performance compared to full-observation world modeling methods.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19642
PDF
-
E2DRL: End-to-End Differentiable Reinforcement Learning for 3D Obstacle Avoidance in Quadruped Robots
2026-10-14 05:54:25Cheng Fan, Bingyang Zhou, Gang Yang, Zixuan Zhuang, Aoqian Zhang, Pengzhi Yang, Xin Liu, Chunzheng Wang, Yan Rui Tan, Fan ShiAbstract
Differentiable reinforcement learning shows strong promise for robot policy learning, but its application to legged locomotion has remained limited to blind walking or simulated 2D tasks. Extending it to end-to-end visual obstacle avoidance is challenging because hard collision constraints and early termination upon contact corrupt first-order gradients, trapping optimizers in local minima. We present E2DRL, which addresses this with a soft differentiable obstacle-interaction formulation: penetration penalties across multiple body keypoints with velocity-conditioned stop-gradient shaping to provide the first-order training signals for 3D obstacle avoidance. To realize a deployable perception-control system, a privileged teacher trained via Soft Analytic Policy Optimization (SAPO) in a differentiable simulator is distilled into a recurrent depth-based student via DAgger, then blended with a base locomotion policy using a terrain-aware gating module for deployment. In simulation, E2DRL achieves higher sample efficiency than PPO and demonstrates superior velocity-tracking performance and 3D obstacle-avoidance capability compared with both PPO and Duality-based DCBF NMPC. We present the first sim-to-real validation of end-to-end differentiable RL for 3D obstacle avoidance on a Unitree Go2 in static and dynamic environments.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19627
PDF
-
Instruction-Conditioned Motion Control for Robotic Arms: Language-to-Code Decomposition and Reinforcement Learning
2026-10-14 05:38:21Fan Huang, Dezhong Tong, Changyu Chen, Haewoon Kwak, Jisun AnAbstract
Instruction-Conditioned Motion Control (ICMC) separates parameter specification, instruction generation, and code synthesis for language-driven robot control, making execution outcomes traceable to instructions and programs. We compare one-shot generation with ICMC under o1 and o4-mini on cube pick-and-place and letter-reshaping tasks for a UR5e arm in MuJoCo. We then train an open-weight instruction policy by GRPO through a frozen code generator on cube placement onto two- and five-cube letter layouts, averaging the rewards of five programs per instruction. The logged positive-reward rate rose from 31.6% to 35.3% (253/800 to 282/800) between the first and last five training steps. A log audit identifies positive binary labels associated with geometric-failure messages: at least 19 in that run, 18 in the dense-reward run, and 5 in the run with one program per instruction. In one case, all five rewards were 1.0 while the log reports a 0.9515 m miss against a 0.05 m tolerance; the verdict came from the program's own output stream. We distinguish logged reward rates from independently verified task success and report descriptive results for each configuration's single training trajectory. We introduce an independent evaluation protocol that checks task completion against fixed specifications, separating robot control from success assessment to support reproducible evaluation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19617
PDF
-
PolyMorph: Learning Locomotion Policy for Cross-Embodiment Adaptation
2026-10-14 05:36:59Hyunsik Oh, Seunghun Jeon, Donghoon Youm, Jungwoo Hur, Hyeongjun Kim, Jeong Hyun Lee, Moonkyu Jung, Jemin HwangboAbstract
Reinforcement learning has enabled robust locomotion for legged robots, but learned policies are typically tied to a specific robot morphology and require additional engineering when the number of legs, actuators, or mechanical properties change. We present PolyMorph, a morphology-conditioned framework that learns a single locomotion policy for heterogeneous legged robots. PolyMorph converts each robot's URDF into joint-link modules and combines a recurrent encoder, graph-based Transformer, and per-joint decoder to generate actions across variable leg counts and joint configurations. To improve training stability, PolyMorph uses morphology-normalized rewards, adaptive robot sampling, and URDF-derived PD-gain initialization. We evaluate PolyMorph in simulation across bipedal, tripedal, and quadrupedal robots, including unseen morphology variants and platforms, and deploy the policy on real quadruped platforms operated in multiple configurations. The results show that a single PolyMorph policy generalizes across diverse robot morphologies without per-robot policy retraining.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19615
PDF
-
RELAY: Selective Self-Distillation with Adaptive Hand-off for Rubric-Based Reinforcement Learning
2026-10-14 05:22:37Kunwang Zheng, Xinyu Sun, Yu Cao, Shuocheng Li, Lei Cheng, Ying Liu, Qilong Zhang, Jinlong Chen, Hua Yang, Qianglong ChenAbstract
Rubric-based reinforcement learning grades a response against many written criteria but collapses them into one reward, which discards the detail the rubric carries and gives no GRPO reward gradient when a group's responses tie. A frozen teacher that reads the rubric as privileged context provides token-level supervision, but we observe an early plateau in the gains over GRPO. RELAY uses advantages and rewards to select positive-advantage and low-scoring zero-advantage rollouts, and monitors the distillation loss to decide when to retire the teacher. Stage 1 combines selective distillation from a frozen teacher with GRPO; a plateau detector then removes the teacher, and Stage 2 continues with GRPO alone. Calibrated on one model-domain pair and reused unchanged in three others, the plateau detector fires between steps 122 and 171. RELAY scores highest in-domain among the compared methods in all four settings, ahead of GRPO on RaR-Medicine by 4.62 points with Qwen3-1.7B and 6.02 with Qwen3-4B. On Qwen3-1.7B/Medicine, removing the teacher from a shared checkpoint outperforms keeping, weakening or refreshing it. Distilling selected rollouts covering about half the response tokens also outperforms distilling all rollouts at a matched average coefficient.
Abstract (translated)
URL
https://arxiv.org/abs/2610.17623
PDF
-
Retrieve What Helps, Only When It Helps: Agentic Reinforcement Learning for Selective Retrieval-Augmented Machine Translation
2026-10-14 03:45:40Shijun Li, Yang Liu, Najmeh Sadoughi, Dongmei Wang, Xinjie Li, Zhu Liu, Rohith Mysore Vijaya Kumar, Yuzong LiuAbstract
Retrieving external knowledge can improve large language model (LLM) translation by providing supplementary information such as idiom interpretations and terminology. However, existing retrieval-augmented machine translation systems typically apply predefined retrieval strategies uniformly across all inputs. We argue that translation operates in a partial-evidence regime: LLM's parametric knowledge suffices for many inputs, while external evidence benefits a subset and usually supports rather than directly provides the desired translation. This creates two challenges: deciding when to retrieve and what evidence to acquire. We introduce Selective Retrieval-Augmented Translation (SRAT), an agentic reinforcement learning framework that learns when to translate directly or retrieve, what to retrieve, and how to integrate these decisions by optimizing final translation quality. To align evidence retrieval with downstream translation quality rather than simple semantic similarity, we further develop joint Plackett--Luce optimization and translation-quality contrastive training for retriever-training. Experiments on idiom, knowledge-intensive, and document-level translation with offline or online retrieval show that SRAT consistently outperforms never- and always-retrieve baselines, and quality-aligned retriever training provides further gains when similarity-based retrieval is insufficient. The results demonstrate the benefit of treating retrieval as a learnable translation action rather than fixed preprocessing.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19528
PDF
-
Diagnosing and Correcting Reward Hacking in Verifiable Instruction Following
2026-10-14 03:42:59Sangyeon Yoon, Wooseok Seo, Hamish Ivison, Hyesoo Hong, Donghoon Kim, Wonjun Kang, Albert No, Jaehyung KimAbstract
Verifiable instruction-following benchmarks emphasize output-constraint satisfaction but can overlook whether responses meaningfully answer the underlying request, creating an evaluation blind spot. We show that frontier models can receive full credit for responses with little substantive answer content. Using the same verifier-based criterion as a training reward can induce reward hacking, as publicly released checkpoints exhibit higher constraint satisfaction but lower response quality after reinforcement learning with verifiable rewards. To address this reward hacking, we introduce CALIF, a lightweight method that augments verifier rewards with a calibrated likelihood-based bonus to preserve response quality. CALIF evaluates response likelihood under the request alone using the existing frozen reference policy, requiring no additional judge or learned reward model. Across models ranging from 1.7B to 8B in reasoning and non-reasoning modes, CALIF improves constraint satisfaction on IFBench while maintaining response quality near or above pre-RL levels. It also outperforms corresponding base policies in 40 of 45 model-metric comparisons across three rubric-based instruction-following benchmarks, demonstrating gains beyond training verifiers.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19525
PDF
-
VLRLA: Deployment-Time Adaptation of Vision-Language-Action Models through In-Context Reinforcement Learning
2026-10-14 03:35:08Diyuan Shi, Yiqi Tang, Hongyin Zhang, You Wu, Donglin WangAbstract
Vision-language-action (VLA) models connect visual perception and language understanding with robot control, providing a foundation for instruction-conditioned manipulation. However, conventional pretrained VLA policies lack a mechanism for gradual adaptation through accumulated interaction experience during deployment. We therefore propose VLRLA, a general framework that equips existing VLA models with \textit{in-context reinforcement learning} capabilities for deployment-time adaptation. Our approach augments a VLA model with a reinforcement learning module that learns from historical observations, actions, rewards, and termination signals. During deployment, the policy gradually refines its action generation and progressively improves task success while keeping model parameters fixed. Our approach is also designed for general applicability and compatibility with diverse learning history construction strategies and model training pipelines. Finally, through extensive experiments, we systematically investigate the learning properties of VLRLA in simulation and demonstrate its ability to adapt to unseen scenarios during real-world deployment.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19517
PDF
-
ForkGRPO: Efficient Exploration and Fine-Grained Credit Assignment for Diffusion Language Models
2026-10-14 01:56:46Zhaoyan Jia, Chengwei Zhou, Gourav Datta, Massoud PedramAbstract
Diffusion large language models (dLLMs) have advanced rapidly, yet strengthening their reasoning through reinforcement learning (RL) remains an open challenge. Any-order generation can suffer from entropy degradation, which narrows the reasoning paths available for exploration. Recent work mitigates this issue by applying standard GRPO with left-to-right generation. However, independently generating every rollout from the input prompt can leave alternative continuations from intermediate states underexplored, while trajectory-level advantages assign uniform credit to all tokens without distinguishing the contributions of individual reasoning segments. We propose ForkGRPO, an RL framework that couples efficient exploration with fine-grained segment-level credit assignment. It samples multiple continuations from high-entropy intermediate states along a reference trajectory, prioritizing textual boundaries and reusing shared prefixes. The continuations provide complementary learning signals: within-state group-relative advantages guide which continuations to reinforce, while Monte Carlo state-value differences indicate which reasoning segments improve expected outcomes. Among the evaluated architecture-preserving RL methods, ForkGRPO achieves the best performance on three of four benchmarks, while improving over the directly comparable JustGRPO by up to 1.2 percentage points and using up to 27.8% fewer generated tokens. ForkGRPO-Fast further achieves a $2.49\times$ training wall-clock speedup with negligible accuracy loss.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19408
PDF
-
Learning What to Keep: Transfer-Aware Personal Skill Learning from User Feedback
2026-10-14 01:27:19Li Wang, Chen Zhang, Yirong Yang, Xingyu Chen, Jinkun Zhao, Tianhao Peng, Wenjun WuAbstract
User feedback enables LLMs to refine reusable personalized skills, but often mixes persistent preferences with request-specific corrections. Indiscriminately retaining both risks negative transfer, motivating selective retention at the level of individual requirements. We frame this challenge as selective transfer over atomic rules and propose SkillGrain, a framework for learning reusable personalized skills from user feedback. SkillGrain decomposes feedback into independently actionable rules, predicts their transferability, and retains only those suitable for future requests. To learn these decisions, we combine rubric-guided skill rewards with fine-grained credit assignment that separately evaluates rule content and retention decisions. We further jointly optimize skill induction and skill-conditioned generation through reinforcement learning, ensuring that learned skills support downstream personalization. We construct a benchmark spanning five personalized writing tasks to evaluate cross-request generalization. Experiments across three base models in both in-domain and out-of-domain settings show that SkillGrain achieves the best overall average performance among competing methods.
Abstract (translated)
URL
https://arxiv.org/abs/2610.19380
PDF
