Reinforcement_Learning
-
Visual-ERM: Reward Modeling for Visual Equivalence
2026-03-13 17:58:14Ziyu Liu, Shengyuan Ding, Xinyu Fang, Xuanlang Dai, Penghui Yang, Jianze Liang, Jiaqi Wang, Kai Chen, Dahua Lin, Yuhang ZangAbstract
Vision-to-code tasks require models to reconstruct structured visual inputs, such as charts, tables, and SVGs, into executable or structured representations with high visual fidelity. While recent Large Vision Language Models (LVLMs) achieve strong results via supervised fine-tuning, reinforcement learning remains challenging due to misaligned reward signals. Existing rewards either rely on textual rules or coarse visual embedding similarity, both of which fail to capture fine-grained visual discrepancies and are vulnerable to reward hacking. We propose Visual Equivalence Reward Model (Visual-ERM), a multimodal generative reward model that provides fine-grained, interpretable, and task-agnostic feedback to evaluate vision-to-code quality directly in the rendered visual space. Integrated into RL, Visual-ERM improves Qwen3-VL-8B-Instruct by +8.4 on chart-to-code and yields consistent gains on table and SVG parsing (+2.7, +4.1 on average), and further strengthens test-time scaling via reflection and revision. We also introduce VisualCritic-RewardBench (VC-RewardBench), a benchmark for judging fine-grained image-to-image discrepancies on structured visual data, where Visual-ERM at 8B decisively outperforms Qwen3-VL-235B-Instruct and approaches leading closed-source models. Our results suggest that fine-grained visual reward supervision is both necessary and sufficient for vision-to-code RL, regardless of task specificity.
Abstract (translated)
URL
https://arxiv.org/abs/2603.13224
PDF
-
Topo-R1: Detecting Topological Anomalies via Vision-Language Models
2026-03-13 15:05:04Meilong Xu, Qingqiao Hu, Xiaoling Hu, Shahira Abousamra, Xin Yu, Weimin Lyu, Kehan Qi, Dimitris Samaras, Chao ChenAbstract
Topological correctness is crucial for tubular structures such as blood vessels, nerve fibers, and road networks. Existing topology-preserving methods rely on domain-specific ground truth, which is costly and rarely transfers across domains. When deployed to a new domain without annotations, a key question arises: how can we detect topological anomalies without ground-truth supervision? We reframe this as topological anomaly detection, a structured visual reasoning task requiring a model to locate and classify topological errors in predicted segmentation masks. Vision-Language Models (VLMs) are natural candidates; however, we find that state-of-the-art VLMs perform nearly at random, lacking the fine-grained, topology-aware perception needed to identify sparse connectivity errors in dense structures. To bridge this gap, we develop an automated data-curation pipeline that synthesizes diverse topological anomalies with verifiable annotations across progressively difficult levels, thereby constructing the first large-scale, multi-domain benchmark for this task. We then introduce Topo-R1, a framework that endows VLMs with topology-aware perception via two-stage training: supervised fine-tuning followed by reinforcement learning with Group Relative Policy Optimization (GRPO). Central to our approach is a topology-aware composite reward that integrates type-aware Hungarian matching for structured error classification, spatial localization scoring, and a centerline Dice (clDice) reward that directly penalizes connectivity disruptions, thereby jointly incentivizing semantic precision and structural fidelity. Extensive experiments demonstrate that Topo-R1 establishes a new paradigm for annotation-free topological quality assessment, consistently outperforming general-purpose VLMs and supervised baselines across all evaluation protocols.
Abstract (translated)
URL
https://arxiv.org/abs/2603.13054
PDF
-
Mending the Holes: Mitigating Reward Hacking in Reinforcement Learning for Multilingual Translation
2026-03-13 14:52:51Yifeng Liu, Siqi Ouyang, Yatish Hosmane Revanasiddappa, Lei LiAbstract
Large Language Models (LLMs) have demonstrated remarkable capability in machine translation on high-resource language pairs, yet their performance on low-resource translation still lags behind. Existing post-training methods rely heavily on high-quality parallel data, which are often scarce or unavailable for low-resource languages. In this paper, we introduce WALAR, a reinforcement training method using only monolingual text to elevate LLMs' translation capabilities on massive low-resource languages while retaining their performance on high-resource languages. Our key insight is based on the observation of failure modes (or "holes") in existing source-based multilingual quality estimation (QE) models. Reinforcement learning (RL) using these QE models tends to amplify such holes, resulting in poorer multilingual LLMs. We develop techniques including word alignment and language alignment to mitigate such holes in WALAR's reward for RL training. We continually trained an LLM supporting translation of 101 languages using WALAR. The experiments show that our new model outperforms LLaMAX, one of the strongest open-source multilingual LLMs by a large margin on 1400 language directions on Flores-101 dataset.
Abstract (translated)
URL
https://arxiv.org/abs/2603.13045
PDF
-
ARL-Tangram: Unleash the Resource Efficiency in Agentic Reinforcement Learning
2026-03-13 14:25:20Bangjun Xiao, Yihao Zhao, Xiangwei Deng, Shihua Yu, Yuxing Xiang, Huaqiu Liu, Qiying Wang, Liang Zhao, Hailin Zhang, Xuanzhe Liu, Xin Jin, Fuli LuoAbstract
Agentic reinforcement learning (RL) has emerged as a transformative workload in cloud clusters, enabling large language models (LLMs) to solve complex problems through interactions with real world. However, unlike traditional RL, agentic RL demands substantial external cloud resources, e.g., CPUs for code execution and GPUs for reward models, that exist outside the primary training cluster. Existing agentic RL framework typically rely on static over-provisioning, i.e., resources are often tied to long-lived trajectories or isolated by tasks, which leads to severe resource inefficiency. We propose the action-level orchestration, and incorporate it into ARL-Tangram, a unified resource management system that enables fine-grained external resource sharing and elasticity. ARL-Tangram utilizes a unified action-level formulation and an elastic scheduling algorithm to minimize action completion time (ACT) while satisfying heterogeneous resource constraints. Further, heterogeneous resource managers are tailored to efficiently support the action-level execution on resources with heterogeneous characteristics and topologies. Evaluation on real-world agentic RL tasks demonstrates that ARL-Tangram improves average ACT by up to 4.3$\times$, speeds up the step duration of RL training by up to 1.5$\times$, and saves the external resources by up to 71.2$\%$. This system has been deployed to support the training of the MiMo series models.
Abstract (translated)
URL
https://arxiv.org/abs/2603.13019
PDF
-
Long-form RewardBench: Evaluating Reward Models for Long-form Generation
2026-03-13 13:05:17Hui Huang, Yancheng He, Wei Liu, Muyun Yang, Jiaheng Liu, Kehai Chen, Bing Xu, Conghui Zhu, Hailong Cao, Tiejun ZhaoAbstract
The widespread adoption of reinforcement learning-based alignment highlights the growing importance of reward models. Various benchmarks have been built to evaluate reward models in various domains and scenarios. However, a significant gap remains in assessing reward models for long-form generation, despite its critical role in real-world applications. To bridge this, we introduce Long-form RewardBench, the first reward modeling testbed specifically designed for long-form generation. Our benchmark encompasses five key subtasks: QA, RAG, Chat, Writing, and Reasoning. We collected instruction and preference data through a meticulously designed multi-stage data collection process, and conducted extensive experiments on 20+ mainstream reward models, including both classifiers and generative models. Our findings reveal that current models still lack long-form reward modeling capabilities. Furthermore, we designed a novel Long-form Needle-in-a-Haystack Test, which revealed a correlation between reward modeling performance and the error's position within a response, as well as the overall response length, with distinct characteristics observed between classification and generative models. Finally, we demonstrate that classifiers exhibit better generalizability compared to generative models trained on the same data. As the first benchmark for long-form reward modeling, this work aims to offer a robust platform for visualizing progress in this crucial area.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12963
PDF
-
Efficient Real-World Autonomous Racing via Attenuated Residual Policy Optimization
2026-03-13 13:02:56Raphael Trumpp, Denis Hoornaert, Mirco Theile, Marco CaccamoAbstract
Residual policy learning (RPL), in which a learned policy refines a static base policy using deep reinforcement learning (DRL), has shown strong performance across various robotic applications. Its effectiveness is particularly evident in autonomous racing, a domain that serves as a challenging benchmark for real-world DRL. However, deploying RPL-based controllers introduces system complexity and increases inference latency. We address this by introducing an extension of RPL named attenuated residual policy optimization ($\alpha$-RPO). Unlike standard RPL, $\alpha$-RPO yields a standalone neural policy by progressively attenuating the base policy, which initially serves to bootstrap learning. Furthermore, this mechanism enables a form of privileged learning, where the base policy is permitted to use sensor modalities not required for final deployment. We design $\alpha$-RPO to integrate seamlessly with PPO, ensuring that the attenuated influence of the base controller is dynamically compensated during policy optimization. We evaluate $\alpha$-RPO by building a framework for 1:10-scaled autonomous racing around it. In both simulation and zero-shot real-world transfer to Roboracer cars, $\alpha$-RPO not only reduces system complexity but also improves driving performance compared to baselines - demonstrating its practicality for robotic deployment. Our code is available at: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12960
PDF
-
Thinking in Streaming Video
2026-03-13 12:33:36Zikang Liu, Longteng Guo, Handong Li, Ru Zhen, Xingjian He, Ruyi Ji, Xiaoming Ren, Yanhao Zhang, Haonan Lu, Jing LiuAbstract
Real-time understanding of continuous video streams is essential for interactive assistants and multimodal agents operating in dynamic environments. However, most existing video reasoning approaches follow a batch paradigm that defers reasoning until the full video context is observed, resulting in high latency and growing computational cost that are incompatible with streaming scenarios. In this paper, we introduce ThinkStream, a framework for streaming video reasoning based on a Watch--Think--Speak paradigm that enables models to incrementally update their understanding as new video observations arrive. At each step, the model performs a short reasoning update and decides whether sufficient evidence has accumulated to produce a response. To support long-horizon streaming, we propose Reasoning-Compressed Streaming Memory (RCSM), which treats intermediate reasoning traces as compact semantic memory that replaces outdated visual tokens while preserving essential context. We further train the model using a Streaming Reinforcement Learning with Verifiable Rewards scheme that aligns incremental reasoning and response timing with the requirements of streaming interaction. Experiments on multiple streaming video benchmarks show that ThinkStream significantly outperforms existing online video models while maintaining low latency and memory usage. Code, models and data will be released at this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2603.12938
PDF
-
Finite Difference Flow Optimization for RL Post-Training of Text-to-Image Models
2026-03-13 10:54:09David McAllister, Miika Aittala, Tero Karras, Janne Hellsten, Angjoo Kanazawa, Timo Aila, Samuli LaineAbstract
Reinforcement learning (RL) has become a standard technique for post-training diffusion-based image synthesis models, as it enables learning from reward signals to explicitly improve desirable aspects such as image quality and prompt alignment. In this paper, we propose an online RL variant that reduces the variance in the model updates by sampling paired trajectories and pulling the flow velocity in the direction of the more favorable image. Unlike existing methods that treat each sampling step as a separate policy action, we consider the entire sampling process as a single action. We experiment with both high-quality vision language models and off-the-shelf quality metrics for rewards, and evaluate the outputs using a broad set of metrics. Our method converges faster and yields higher output quality and prompt alignment than previous approaches.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12893
PDF
-
Beyond Imitation: Reinforcement Learning Fine-Tuning for Adaptive Diffusion Navigation Policies
2026-03-13 10:14:32Junhe Sheng, Ruofei Bai, Kuan Xu, Ruimeng Liu, Jie Chen, Shenghai Yuan, Wei-Yun Yau, Lihua XieAbstract
Diffusion-based robot navigation policies trained on large-scale imitation learning datasets, can generate multi-modal trajectories directly from the robot's visual observations, bypassing the traditional localization-mapping-planning pipeline and achieving strong zero-shot generalization. However, their performance remains constrained by the coverage of offline datasets, and when deployed in unseen settings, distribution shift often leads to accumulated trajectory errors and safety-critical failures. Adapting diffusion policies with reinforcement learning is challenging because their iterative denoising structure hinders effective gradient backpropagation, while also making the training of an additional value network computationally expensive and less stable. To address these issues, we propose a reinforcement learning fine-tuning framework tailored for diffusion-based navigation. The method leverages the inherent multi-trajectory sampling mechanism of diffusion models and adopts Group Relative Policy Optimization (GRPO), which estimates relative advantages across sampled trajectories without requiring a separate value network. To preserve pretrained representations while enabling adaptation, we freeze the visual encoder and selectively update the higher decoder layers and action head, enhancing safety-aware behaviors through online environmental feedback. On the PointGoal task in Isaac Sim, our approach improves the Success Rate from 52.0% to 58.7% and SPL from 0.49 to 0.54 on unseen scenes, while reducing collision frequency. Additional experiments show that the fine-tuned policy transfers zero-shot to a real quadruped platform and maintains stable performance in geometrically out-of-distribution environments, suggesting improved adaptability and safe generalization to new domains.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12868
PDF
-
Rethinking Multiple-Choice Questions for RLVR: Unlocking Potential via Distractor Design
2026-03-13 09:28:36Xu Guo, Qiming Ge, Jian Tong, Kedi Chen, Jin Zhang, Xiaogui Yang, Xuan Gao, Haijun Lv, Zhihui Lu, Yicheng Zou, Qipeng GuoAbstract
Reinforcement Learning with Verifiable Rewards (RLVR) significantly enhances the reasoning capabilities of Large Language Models. When applied to RLVR, Multiple-Choice Questions (MCQs) offer a scalable source of verifiable data but risk inducing reward hacking, where models shortcut reasoning via random guessing or simple elimination. Current approaches often mitigate this by converting MCQs to open-ended formats, thereby discarding the contrastive signal provided by expert-designed distractors. In this work, we systematically investigate the impact of option design on RLVR. Our analysis highlights two primary insights: (1) Mismatches in option counts between training and testing degrade performance. (2) Strong distractors effectively mitigate random guessing, enabling effective RLVR training even with 2-way questions. Motivated by these findings, we propose Iterative Distractor Curation (IDC), a framework that actively constructs high-quality distractors to block elimination shortcuts and promote deep reasoning. Experiments on various benchmarks demonstrate that our method effectively enhances distractor quality and yields significant gains in RLVR training compared to the original data.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12826
PDF
-
Reinforcement Learning for Elliptical Cylinder Motion Control Tasks
2026-03-13 09:10:20Pawel Marczewski, Paulina Superczynska, Jakub Bernat, Szymon SzczesnyAbstract
The control of devices with limited input always bring attention to solve by research due to its difficulty and non-trival solution. For instance, the inverted pendulum is benchmarking problem in control theory and machine learning. In this work, we are focused on the elliptical cylinder and its motion under limited torque. The inspiration of the problem is from untethered magnetic devices, which due to distance have to operate with limited input torque. In this work, the main goal is to define the control problem of elliptic cylinder with limited input torque and solve it by Reinforcement Learning. As a classical baseline, we evaluate a two-stage controller composed of an energy-shaping swing-up law and a local Linear Quadratic Regulator (LQR) stabilizer around the target equilibrium. The swing-up controller increases the system's mechanical energy to drive the state toward a neighborhood of the desired equilibrium, a linearization of the nonlinear model yields an LQR that regulates the angle and angular-rate states to the target orientation with bounded input. This swing-up + LQR policy is a strong, interpretable reference for underactuated system and serves a point of comparison to the learned policy under identical limits and parameters. The solution shows that the learning is possible however, the different cases like stabilization in upward position or rotating of half turn are very difficult for increasing mass or ellipses with a strongly unequal perimeter ratio.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12807
PDF
-
FLUX: Accelerating Cross-Embodiment Generative Navigation Policies via Rectified Flow and Static-to-Dynamic Learning
2026-03-13 09:10:01Zeying Gong, Yangyi Zhong, Yiyi Ding, Tianshuai Hu, Guoyang Zhao, Lingdong Kong, Rong Li, Jiadi You, Junwei LiangAbstract
Autonomous navigation requires a broad spectrum of skills, from static goal-reaching to dynamic social traversal, yet evaluation remains fragmented across disparate protocols. We introduce DynBench, a dynamic navigation benchmark featuring physically valid crowd simulation. Combined with existing static protocols, it supports comprehensive evaluation across six fundamental navigation tasks. Within this framework, we propose FLUX, the first flow-based unified navigation policy. By linearizing probability flow, FLUX replaces iterative denoising with straight-line trajectories, improving per-step inference efficiency by 47% over prior flow-based methods and 29% over diffusion-based ones. Following a static-to-dynamic curriculum, FLUX initially establishes geometric priors and is subsequently refined through reinforcement learning in dynamic social environments. This regime not only strengthens socially-aware navigation but also enhances static task robustness by capturing recovery behaviors through stochastic action distributions. FLUX achieves state-of-the-art performance across all tasks and demonstrates zero-shot sim-to-real transfer on wheeled, quadrupedal, and humanoid platforms without any fine-tuning.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12806
PDF
-
Think and Answer ME: Benchmarking and Exploring Multi-Entity Reasoning Grounding in Remote Sensing
2026-03-13 08:47:19Shuchang Lyu, Haiquan Wen, Guangliang Cheng, Meng Li, Zheng Zhou, You Zhou, Dingding Yao, Zhenwei ShiAbstract
Recent advances in reasoning language models and reinforcement learning with verifiable rewards have significantly enhanced multi-step reasoning capabilities. This progress motivates the extension of reasoning paradigms to remote sensing visual grounding task. However, existing remote sensing grounding methods remain largely confined to perception-level matching and single-entity formulations, limiting the role of explicit reasoning and inter-entity modeling. To address this challenge, we introduce a new benchmark dataset for Multi-Entity Reasoning Grounding in Remote Sensing (ME-RSRG). Based on ME-RSRG, we reformulate remote sensing grounding as a multi-entity reasoning task and propose an Entity-Aware Reasoning (EAR) framework built upon visual-linguistic foundation models. EAR generates structured reasoning traces and subject-object grounding outputs. It adopts supervised fine-tuning for cold-start initialization and is further optimized via entity-aware reward-driven Group Relative Policy Optimization (GRPO). Extensive experiments on ME-RSRG demonstrate the challenges of multi-entity reasoning and verify the effectiveness of our proposed EAR framework. Our dataset, code, and models will be available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12788
PDF
-
EvolveCoder: Evolving Test Cases via Adversarial Verification for Code Reinforcement Learning
2026-03-13 06:26:50Chi Ruan, Dongfu Jiang, Huaye Zeng, Ping Nie, Wenhu ChenAbstract
Reinforcement learning with verifiable rewards (RLVR) is a promising approach for improving code generation in large language models, but its effectiveness is limited by weak and static verification signals in existing coding RL datasets. In this paper, we propose a solution-conditioned and adversarial verification framework that iteratively refines test cases based on the execution behaviors of candidate solutions, with the goal of increasing difficulty, improving discriminative power, and reducing redundancy. Based on this framework, we introduce EvolveCoder-22k, a large-scale coding reinforcement learning dataset constructed through multiple rounds of adversarial test case evolution. Empirical analysis shows that iterative refinement substantially strengthens verification, with pass@1 decreasing from 43.80 to 31.22. Reinforcement learning on EvolveCoder-22k yields stable optimization and consistent performance gains, improving Qwen3-4B by an average of 4.2 points across four downstream benchmarks and outperforming strong 4B-scale baselines. Our results highlight the importance of adversarial, solution-conditioned verification for effective and scalable reinforcement learning in code generation.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12698
PDF
-
RetroReasoner: A Reasoning LLM for Strategic Retrosynthesis Prediction
2026-03-13 05:20:56Hanbum Ko, Chanhui Lee, Ye Rin Kim, Rodrigo Hormazabal, Sehui Han, Sungbin Lim, Sungwoong KimAbstract
Retrosynthesis prediction is a core task in organic synthesis that aims to predict reactants for a given product molecule. Traditionally, chemists select a plausible bond disconnection and derive corresponding reactants, which is time-consuming and requires substantial expertise. While recent advancements in molecular large language models (LLMs) have made progress, many methods either predict reactants without strategic reasoning or conduct only a generic product analysis, rather than reason explicitly about bond-disconnection strategies that logically lead to the choice of specific reactants. To overcome these limitations, we propose RetroReasoner, a retrosynthetic reasoning model that leverages chemists' strategic thinking. RetroReasoner is trained using both supervised fine-tuning (SFT) and reinforcement learning (RL). For SFT, we introduce SyntheticRetro, a framework that generates structured disconnection rationales alongside reactant predictions. In the case of RL, we apply a round-trip accuracy as reward, where predicted reactants are passed through a forward synthesis model, and predictions are rewarded when the forward-predicted product matches the original input product. Experimental results show that RetroReasoner not only outperforms prior baselines but also generates a broader range of feasible reactant proposals, particularly in handling more challenging reaction instances.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12666
PDF
-
FastDSAC: Unlocking the Potential of Maximum Entropy RL in High-Dimensional Humanoid Control
2026-03-13 03:27:25Jun Xue, Junze Wang, Xinming Zhang, Shanze Wang, Yanjun Chen, Wei ZhangAbstract
Scaling Maximum Entropy Reinforcement Learning (RL) to high-dimensional humanoid control remains a formidable challenge, as the ``curse of dimensionality'' induces severe exploration inefficiency and training instability in expansive action spaces. Consequently, recent high-throughput paradigms have largely converged on deterministic policy gradients combined with massive parallel simulation. We challenge this compromise with FastDSAC, a framework that effectively unlocks the potential of maximum entropy stochastic policies for complex continuous control. We introduce Dimension-wise Entropy Modulation (DEM) to dynamically redistribute the exploration budget and enforce diversity, alongside a continuous distributional critic tailored to ensure value fidelity and mitigate high-dimensional value overestimation. Extensive evaluations on HumanoidBench and other continuous control tasks demonstrate that rigorously designed stochastic policies can consistently match or outperform deterministic baselines, achieving notable gains of 180\% and 400\% on the challenging \textit{Basketball} and \textit{Balance Hard} tasks.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12612
PDF
-
Swap-guided Preference Learning for Personalized Reinforcement Learning from Human Feedback
2026-03-13 02:51:50Gihoon Kim, Euntai KimAbstract
Reinforcement Learning from Human Feedback (RLHF) is a widely used approach to align large-scale AI systems with human values. However, RLHF typically assumes a single, universal reward, which overlooks diverse preferences and limits personalization. Variational Preference Learning (VPL) seeks to address this by introducing user-specific latent variables. Despite its promise, we found that VPL suffers from posterior collapse. While this phenomenon is well known in VAEs, it has not previously been identified in preference learning frameworks. Under sparse preference data and with overly expressive decoders, VPL may cause latent variables to be ignored, reverting to a single-reward model. To overcome this limitation, we propose Swap-guided Preference Learning (SPL). The key idea is to construct fictitious swap annotators and use the mirroring property of their preferences to guide the encoder. SPL introduces three components: (1) swap-guided base regularization, (2) Preferential Inverse Autoregressive Flow (P-IAF), and (3) adaptive latent conditioning. Experiments show that SPL mitigates collapse, enriches user-specific latents, and improves preference prediction. Our code and data are available at this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2603.12595
PDF
-
Reinforcement Learning for Diffusion LLMs with Entropy-Guided Step Selection and Stepwise Advantages
2026-03-13 01:38:44Vishnu Teja Kunde, Fatemeh Doudi, Mahdi Farahbakhsh, Dileep Kalathil, Krishna Narayanan, Jean-Francois ChamberlandAbstract
Reinforcement learning (RL) has been effective for post-training autoregressive (AR) language models, but extending these methods to diffusion language models (DLMs) is challenging due to intractable sequence-level likelihoods. Existing approaches therefore rely on surrogate likelihoods or heuristic approximations, which can introduce bias and obscure the sequential structure of denoising. We formulate diffusion-based sequence generation as a finite-horizon Markov decision process over the denoising trajectory and derive an exact, unbiased policy gradient that decomposes over denoising steps and is expressed in terms of intermediate advantages, without requiring explicit evaluation of the sequence likelihood. To obtain a practical and compute-efficient estimator, we (i) select denoising steps for policy updates via an entropy-guided approximation bound, and (ii) estimate intermediate advantages using a one-step denoising reward naturally provided by the diffusion model, avoiding costly multi-step rollouts. Experiments on coding and logical reasoning benchmarks demonstrate state-of-the-art results, with strong competitive performance on mathematical reasoning, outperforming existing RL post-training approaches for DLMs. Code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12554
PDF
-
CALF: Communication-Aware Learning Framework for Distributed Reinforcement Learning
Abstract
Distributed reinforcement learning policies face network delays, jitter, and packet loss when deployed across edge devices and cloud servers. Standard RL training assumes zero-latency interaction, causing severe performance degradation under realistic network conditions. We introduce CALF (Communication-Aware Learning Framework), which trains policies under realistic network models during simulation. Systematic experiments demonstrate that network-aware training substantially reduces deployment performance gaps compared to network-agnostic baselines. Distributed policy deployments across heterogeneous hardware validate that explicitly modelling communication constraints during training enables robust real-world execution. These findings establish network conditions as a major axis of sim-to-real transfer for Wi-Fi-like distributed deployments, complementing physics and visual domain randomisation.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12543
PDF
-
Beyond Motion Imitation: Is Human Motion Data Alone Sufficient to Explain Gait Control and Biomechanics?
2026-03-12 19:42:19Xinyi Liu, Jangwhan Ahn, Edgar Lobaton, Jennie Si, He HuangAbstract
With the growing interest in motion imitation learning (IL) for human biomechanics and wearable robotics, this study investigates how additional foot-ground interaction measures, used as reward terms, affect human gait kinematics and kinetics estimation within a reinforcement learning-based IL framework. Results indicate that accurate reproduction of forward kinematics alone does not ensure biomechanically plausible joint kinetics. Adding foot-ground contacts and contact forces to the IL reward terms enables the prediction of joint moments in forward walking simulation, which are significantly closer to those computed by inverse dynamics. This finding highlights a fundamental limitation of motion-only IL approaches, which may prioritize kinematics matching over physical consistency. Incorporating kinetic constraints, particularly ground reaction force and center of pressure information, significantly enhances the realism of internal and external kinetics. These findings suggest that, when imitation learning is applied to human-related research domains such as biomechanics and wearable robot co-design, kinetics-based reward shaping is necessary to achieve physically consistent gait representations.
Abstract (translated)
URL
https://arxiv.org/abs/2603.12408
PDF
