Reinforcement_Learning
-
Copy Less, Ground More: Overcoming Repetitive Copying in Long-Context Reasoning via Evidence-Aware Reinforcement Learning
2026-07-21 17:59:21Lizhe Fang, Weizhou Shen, Tianyi Tang, Yisen WangAbstract
Large language models that generate step-by-step reasoning traces have achieved strong performance on complex tasks, and extending them to long-context settings has emerged as an important frontier. However, we identify a critical failure mode in this regime: \emph{repetitive copying}, where models extensively copy text from the input into their reasoning traces rather than productively solving the problem. We show that this behavior is pervasive across frontier long-context LLMs and intensifies with context length. By separating each prompt into task-relevant key evidence and irrelevant distractor context, we further show that the root cause is insufficient grounding: models copy from the prompt indiscriminately, and those that fail to focus on key evidence are far more likely to answer incorrectly. Motivated by this diagnosis, we propose GEAR (Grounding Evidence-Aware Reward), a reward shaping method that augments the accuracy signal with a grounding reward for overlap with key evidence and a distractor penalty for overlap with irrelevant context. To enable GEAR on natural-language data, we develop an automated pipeline that constructs evidence-annotated training data from arbitrary documents. We validate GEAR across multiple model scales and benchmarks, showing consistent improvements of up to +4.6 average points over standard RL with accuracy-based rewards, with larger gains at longer contexts, while also reducing repetitive copying and thinking length. Our findings suggest that, even as long-context evaluation shifts from simple retrieval toward complex reasoning, accurate grounding in relevant evidence remains an indispensable capability with substantial room for improvement.
Abstract (translated)
URL
https://arxiv.org/abs/2607.19345
PDF
-
OmniReasoner: Thinking with Long Audio-Video via Native Tool Use
2026-07-21 17:57:50Yu Chen, Caorui Li, Ziyu Xiong, Yidong Wang, Mingqi Gao, Shuman Liu, Biao Liu, Chunfeng Yang, Anxiang Zeng, Haibo Zhang, Chaofan ChenAbstract
Long audio-video reasoning is difficult for omnimodal LLMs because the decisive evidence is often sparse, cross-modal, and too expensive to preserve with uniformly high-fidelity inputs. We introduce OmniReasoner, a tool-use post-training framework for Thinking with Long Audio-Video: omni-modal LLMs learn, via supervised fine-tuning and reinforcement learning, to decide whether and where to call a zoom-in tool before answering. OmniReasoner first builds a low-cost global preview of the full stream and then, when needed, calls the zoom-in tool with a requested temporal interval for higher-fidelity visual and audio inspection before answering. Because the model observes different sampling granularities before and after this call -- a sparse global preview and a denser local clip -- we introduce TimeAnchor, which keeps the tool's temporal argument valid and round-trip-consistent across these granularities, rather than tied to frame indices from a particular sampling rate. To make this tool-use behavior trainable without expensive manual interval annotation, we build a Temporal Augmented Data Engine that synthesizes tool-use post-training trajectories by video editing and composition. Experiments across omnimodal and video benchmarks show that OmniReasoner improves both answer accuracy and temporal grounding while concentrating high-fidelity computation on informative regions. Code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2607.19339
PDF
-
ISO: An RLVR-Native Optimization Stack
2026-07-21 17:51:36Hanqing Zhu, Wenyan Cong, Zhizhou Sha, Sagnik Mukherjee, Xinyuan Song, David Gonz\'alez-Mart\'inez, Xiaoxia Wu, Yuandong Tian, Shiwei Liu, David Z. Pan, Zhangyang "Atlas" WangAbstract
Reinforcement learning with verifiable rewards (RLVR) is rapidly advancing the reasoning capabilities of language models, yet the optimization layer that converts reward feedback into weight-space updates remains poorly understood. Building on our prior analysis (Zhu et al., 2025), we study this missing layer through the singular structure of model weights and identify spectral inheritance: RLVR can reuse the base model's weight spectra while acquiring new behavior through changes in the associated input and output singular frames. We operationalize spectral inheritance as Isospectral Optimization (ISO), an RLVR-native, fixed-spectrum optimization framework with complementary offline and online instantiations. Offline, ISO-Merger combines the frame changes of shared-base specialists into a single fixed-spectrum model, requiring no post-merge data, rollouts, gradient updates, or on-policy distillation (OPD). It recovers complementary specialist capabilities and achieves the strongest aggregate performance among the compared data-free merging methods. Online, ISO-Optimizer applies a chosen base optimizer, including AdamW and Muon, to the frame variables while keeping the base spectra fixed. Across reasoning and coding tasks ranging from 1.5B to 8B parameters, ISO-Optimizer improves accuracy in the reported runs and reaches matched scores with substantially fewer training steps. On Qwen3-8B-Base, AdamW reaches an aggregate accuracy of 0.495 after 270 training steps. ISO-AdamW reaches the same accuracy after only 100 training steps and improves further to 0.509 after 210 training steps. Together, ISO offers a concrete answer to RLVR's missing optimization layer: rather than inheriting pre-training optimization wholesale, design post-training around the structure of reward-driven adaptation: inherit the spectrum, optimize the frames.
Abstract (translated)
URL
https://arxiv.org/abs/2607.19331
PDF
-
Off-Context GRPO: Learning to Reason on Hard Problems using Privileged Information
2026-07-21 17:28:40Priyank Agrawal, Ankur Samanta, Shervin Ghasemlou, Jalaj Bhandari, Kavosh Asadi, Daniel Jiang, Aditya ModiAbstract
Reinforcement learning with verifiable rewards (RLVR) improves reasoning in large language models. Yet, typical RLVR approaches fail on difficult problems: when a model cannot generate any correct solutions, it receives \textit{zero} learning signal. Providing privileged guidance during training, such as solution prefixes, can help overcome this learning cliff by steering the model towards {correct solutions with non-zero reward}. {We call these rollouts \textit{off-context}: they are generated from a training prompt that contains privileged guidance, while the target objective is defined by the original prompt without that guidance.} {We introduce} Off-Context GRPO (OC-GRPO), a minimally modified variant of GRPO that uses guided rollouts but applies an importance-corrected objective to steer the update back toward the original unguided objective, avoiding the mismatch that destabilizes uncorrected guided training. Empirically, our algorithm achieves a 3.9\% absolute improvement (13.8\% relative gain) over vanilla GRPO on average across standard mathematical reasoning benchmarks with negligible additional cost.
Abstract (translated)
URL
https://arxiv.org/abs/2607.19313
PDF
-
The Price of Reasoning: Cost-Quality Tradeoffs in Reinforcement Learning for Neural Machine Translation
2026-07-21 15:57:36Michael Jungo, Aixiu AnAbstract
Reinforcement learning with verifiable rewards (RLVR) has been established as a viable paradigm for the post-training of Large Language Models (LLMs), including downstream tasks, such as Neural Machine Translation (NMT). With the latest research indicating that RLVR could be the preferred training method for translating legal documents due to the induced reasoning capabilities, it raises the question whether it is really attributed to the reasoning or more generally to the training paradigm. We investigate the importance of including the model's reasoning trace in the generated responses during both training and inference by systematically omitting it from one of the phases. Our experiments show that including the reasoning, specifically during inference, has a positive effect on the overall translation quality. Furthermore, we recognise that the reasoning leads to an increase in output tokens, hence we study the cost-quality tradeoff between the increased computational demands and the improved translation quality.
Abstract (translated)
URL
https://arxiv.org/abs/2607.19226
PDF
-
Beyond Score Prediction: LLM-Based Essay Scoring and Feedback Generation via Reinforcement Learning with Rubric Rewards
2026-07-21 15:49:02Xuefeng Jin, Jiashuo Zhang, Teng Cao, Bin YangAbstract
Large language models (LLMs) have been widely applied to automated essay scoring (AES) and automated feedback generation (AFG). However, existing studies rely primarily on prompt engineering or supervised fine-tuning, while systematic research on reinforcement learning (RL) post-training and automated evaluation of feedback quality remains limited. We propose RLAES, a unified LLM framework that jointly optimizes essay scoring and feedback generation through RL. To make feedback quality measurable, interpretable, and usable for training, we introduce Rubric-based Feedback Evaluation (RFE), an essay-grounded feedback evaluation framework comprising 166 fine-grained binary rubric items and an LLM-as-judge. Building on RFE, we propose Adaptive Gated Feedback Optimization (AGFO), which activates rubric-based feedback rewards on demand during RL, reducing evaluation overhead while improving feedback quality. We also propose Adjacent Contrastive Reasoning (ACR) to improve ordinal score calibration by explicitly contrasting adjacent score levels. Experimental results show that the RFE framework captures essay-feedback consistency, exhibits strong pairwise discriminative power, and closely aligns with expert preferences. On the ASAP benchmark, RLAES-AGFO achieves the best scoring performance among LLM-based methods (QWK = 0.803), while maintaining feedback quality comparable to GPT-5.5 and avoiding the feedback degradation observed under score-only RL. Code and datasets are publicly available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2607.19219
PDF
-
Reasoning Before Translation: Enhancing Legal Machine Translation with Structured Reasoning
2026-07-21 15:15:56Aixiu An, Michael Jungo, Eloi Eynard, Mark Drenhaus, Andreas Fischer, Jean Hennebert, S\'ebastien RumleyAbstract
Neural machine translation (NMT) in the legal domain is a linguistically and conceptually demanding task, primarily due to the complexity of legal language and the high level of precision it requires. The recent emergence of reasoning-capable language models opens new possibilities for tackling such challenges. They add to a set of other previously proposed techniques to enhance the translation quality, which includes supervised fine-tuning and reinforcement learning. In this work, we perform a comparison between these various approaches. More particularly, we evaluate small language models such as Qwen3.5 4B, Qwen3.5 9B, and Gemma 3 12B enhanced with various re-training paradigms and compare their performances against frontier reasoning models. We focus on the Swiss legal system, which -- with its unique multilingual statutes -- offers a particularly challenging testbed for reasoning-augmented models. Our results show that the quality of small ``base'' models can be greatly enhanced, and that reinforcement learning with verifiable rewards can be applied to NMT in the legal domain and surpasses the translation quality of supervised fine-tuning. The performance of enhanced small models is close to the one of state-of-the-art reasoning models yet remains inferior. We also note that re-training paradigms yield diminishing returns as model size increase. The code and models are publicly available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2607.19181
PDF
-
Comparative Study of Multi-Agent Actor-Critic Algorithms in Parameterized Action Reinforcement Learning
2026-07-21 14:03:29Ubayd Ali Bapoo, Clement N NyirendaAbstract
Parameterized action reinforcement learning has shown strong performance in environments requiring both discrete action selection and continuous parameterization. Prior work established the effectiveness of single-agent actor-critic algorithms - Greedy Actor-Critic (GAC), Soft Actor-Critic (SAC), and Truncated Quantile Critics (TQC) - on benchmark parameterized action tasks, but their extension to multi-agent settings remains largely unexplored. This paper presents a comparative study of shared-experience multi-agent extensions of these algorithms: Multi-Agent Greedy Actor-Critic (MAGAC), Multi-Agent Soft Actor-Critic (MASAC), and Multi-Agent Truncated Quantile Critics (MATQC). Rather than following the centralized training, decentralized execution (CTDE) paradigm, the proposed framework uses multiple independent actor-critic agents that share a replay buffer while maintaining separate policy and value networks. We evaluate the algorithms on the Platform-v0 and Goal-v0 benchmarks against their single-agent counterparts, using three-, five-, and ten-agent configurations to assess scalability. Performance is measured by average evaluation return and training time across ten independent runs, with one-way ANOVA and Tukey HSD post-hoc tests used to assess statistical significance. Results show that the multi-agent framework consistently improves Greedy Actor-Critic performance, while MASAC and MATQC show comparatively modest gains over their single-agent versions. Increasing the number of agents beyond five yields limited additional performance while substantially raising computational cost, particularly for MAGAC. These results highlight a trade-off between learning performance and computational efficiency, offering insight into the scalability of shared-experience multi-agent actor-critic methods for parameterized action reinforcement learning.
Abstract (translated)
URL
https://arxiv.org/abs/2607.19117
PDF
-
DobicVLM: Aligning Chest X-Ray Report Generation with Clinically-Grounded Programmatic Rewards via Group Relative Policy Optimization
2026-07-21 11:20:09Thanni Adewuyi, Angelica Obayi, Andem Aniekan, Samuel Okoko, Angel Ezendu, Ephraim Usani, Ademide Animasaun, Philip Chibundu, Christian Maurice, Mary Donald Essien, Oluwaseun Odunsi, Oluwasegun Oguntuase, Abiodun AdereniAbstract
Medical imaging is a cornerstone of diagnostics, yet automated chest X-ray report generation struggles with structural adherence, anatomical completeness, and semantic faithfulness. We introduce DobicVLM, a vision-language model combining supervised fine-tuning on MedGemma-4B with Group Relative Policy Optimization (GRPO) and clinically-grounded programmatic rewards. Our approach uses interpretable, rule-based reward components; structural verification, anatomical checklist, semantic similarity, and length constraints to enforce clinical standards without neural reward models. Trained on 1,000 de-identified image-report pairs from a private clinical dataset (with ethics approval and compliance to local regulations), DobicVLM is evaluated via blinded expert review on 69 held-out cases. DobicVLM outperforms Gemini 2.5 Flash across the majority of criteria, achieving the highest impression accuracy (27.2%) and medical terminology (86.5%) compared to both Gemini 2.5 Flash and MedGemma 4B baselines, with minor trade-offs in completeness and referrals. This demonstrates GRPO's value for transparent alignment in resource-limited settings. Keywords: Vision-Language Models, Radiology Report Generation, Reinforcement Learning, Medical AI, GRPO
Abstract (translated)
URL
https://arxiv.org/abs/2607.18988
PDF
-
Athena-Brain Technical Report: An Efficient Robot Brain for General Intelligence and Embodied Interactio
2026-07-21 11:19:21Jialian Li, Junhong Liu, Yuchen Cao, Weiran Guo, Jiaming Song, Xutao Wang, Yi Zhao, Jiangpin Liu, Jie ChenAbstract
Large language models (LLMs) have demonstrated remarkable capabilities in language understanding, reasoning, and world knowledge. As embodied agents become increasingly capable, there is a growing demand for compact models that can serve as an on-device brain, preserving the broad general intelligence of LLMs while enabling effective high-level interaction with embodied environments. Existing approaches, however, often prioritize either general-purpose intelligence or specialized embodied capabilities, making it challenging to satisfy both requirements within a single model. We present \textbf{Athena-Brain-8B}, an 8B LLM designed to serve as an on-device brain for embodied intelligence for embodied intelligence. Through a multi-stage post-training pipeline consisting of General Supervised Fine-Tuning, General Reinforcement Learning, Embodied Expert training, and Model Merge, Athena-Brain-8B maintains strong general capabilities while acquiring strong high-level embodied interaction capabilities and generating concise responses for efficient embodied interaction. Experimental results demonstrate the effectiveness of Athena across both general and embodied evaluations. Compared with the corresponding Qwen3-8B thinking model, Athena-Brain-8B achieves comparable performance on general language and reasoning benchmarks while generating substantially shorter responses. On in-domain embodied benchmarks, Athena-Brain-8B consistently outperforms models of similar scale and surpasses several substantially larger frontier models evaluated zero-shot, demonstrating that compact language models can effectively integrate strong general intelligence with embodied capabilities.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18985
PDF
-
Fishing Out Free Riders: Shapley-Based Reward Attribution for Parallel Reasoning via Reinforcement Learning
2026-07-21 11:14:33Wentao Zhang, Haoyu Zhang, Xinke Jiang, Yuxuan Cheng, Yuhan Pan, Miao Li, Zhipeng Qiao, Tao Feng, Zhen Tao, Dengji ZhaoAbstract
Large Language Models (LLMs) excel at multi-step reasoning, yet current parallel reasoning approaches often fail to distinguish the contributions of individual reasoning paths. Many paths may be redundant, misleading, or even detrimental, but outcome-level rewards assign uniform reward, leading to ambiguous learning signals and unstable training. We propose Parallel Shapley, a reinforcement learning framework that attributes fine-grained, path-level contributions in multi-path reasoning. Treating each path as a player in a cooperative game, we leverage Shapley values to quantify marginal contributions, using a generative reward model to evaluate path utilities and Monte Carlo sampling for efficient approximation. Experiments on mathematical reasoning benchmarks show that Parallel Shapley outperforms existing baselines while providing more stable and interpretable training. Our framework effectively "fishes out the free riders," assigning reward proportionally and improving multi-path reasoning in LLMs.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18979
PDF
-
Measuring Reward-Seeking via Contrastive Belief Updates
2026-07-21 10:57:09Axel H{\o}jmark, J\'er\'emy Scheurer, Evgenia Nitishinskaya, Felix Hofst\"atter, Jason Wolfe, Theodore Ehrenborg, Bronson Schoen, Alexander MeinkeAbstract
Language models trained with reinforcement learning may learn to optimize the grader's judgment rather than the intended objective. This "reward-seeking" is difficult to measure because a model that pursues the grader's judgment and one that pursues the intended objective behave identically whenever the grader rewards the intended behavior. We measure reward-seeking using Contrastive Synthetic Document Finetuning to change a model's beliefs about what the grader rewards, putting those beliefs in conflict with what users or developers want, and measuring the rate at which the model adopts each party's preferred behavior. Applied to intermediate checkpoints of a capabilities-focused OpenAI o3 RL run, without safety training, we find that these checkpoints often side with grader preferences over those of users or developers on coding and alignment tasks. This tendency to side with the grader trends upward throughout RL training. For example, in an environment that forces a choice between keeping a promise to a supervisor and breaking it to complete the task, a late capabilities-focused o3 checkpoint breaks the promise 87% of the time when SDF documents say the grader rewards task completion, versus 9% when they say it rewards honesty (a choice its chain-of-thought often makes explicit). An earlier checkpoint is far less sensitive (40% vs. 24%). Our method also generalizes to reward-hacking models. A model organism trained to reward-hack (gpt-oss-120b) is more than twice as sensitive to grader preferences as the unmodified model, with the mean behavioral shift in favor of the grader rising from 33% to 86%. These results indicate that RL can increase reward-seeking over the course of training, producing models that may act against their developers' intentions when they believe that doing so leads to higher reward.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18966
PDF
-
H$^2$SD: Hybrid Hindsight Self-Distillation
2026-07-21 10:47:27Qiye Cai, Yichuan Ma, Linyang Li, Peiji Li, Yongkang Chen, Qipeng Guo, Yicheng Zou, Tao Gui, Xiaocheng Feng, Bing QinAbstract
Reinforcement learning with verifiable rewards (RLVR) has substantially improved the reasoning capabilities of large language models on tasks such as mathematical reasoning and code generation. However, most RLVR methods assign a scalar outcome reward to an entire trajectory, resulting in sparse supervision and limited token-level credit assignment. On-policy distillation (OPD) provides denser supervision by distilling token-level distributions from a stronger teacher model, but requires an additional teacher and typically assumes a shared vocabulary. On-policy self-distillation (OPSD) removes this dependency by conditioning the same model on privileged information to construct a teacher policy. However, directly matching the teacher distribution may cause information leakage and unstable optimization. RLSD avoids direct matching by using the teacher signal only to modulate update magnitudes, but it cannot provide an explicit correction direction when the sampled reasoning fails. To address this tradeoff, we introduce $\mathrm{H}^{2}\mathrm{SD}$, a hybrid hindsight self distillation framework that uses the teacher differently according to trajectory correctness. For successful trajectories, the teacher receives the student response confirmed as correct together with a rephrasing instruction, and its probabilities on the original response tokens are used to modulate update magnitudes without changing the direction determined by the reward. For failed trajectories, we condition the teacher on a reference hint containing key reasoning steps and a verified answer, and minimize the reverse KL divergence from the student to the teacher. Experiments on multiple challenging reasoning benchmarks show that H$^2$SD consistently outperforms representative RLVR, OPSD, and RLSD baselines while maintaining stable optimization and favorable generation efficiency.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18955
PDF
-
Circuit Claims Depend on What Is Extracted and How It Is Compared
2026-07-21 10:04:29Yang Sheng, Jie FuAbstract
Circuit extraction identifies a small set of model components whose presence preserves a target behavior under ablation, and the resulting circuit is often read as the mechanism behind that behavior. We argue that this reading is under-determined: preserving behavior does not single out one circuit, because the claim it supports depends on which circuit is reported and how two circuits are compared. We make this concrete in a synthetic Lean tactic-prediction benchmark -- predicting the next step of a proof -- where fixed proof rules with randomized surface form let differences between extracted circuits be attributed to these choices rather than to the task. Across dense and weight-sparse checkpoints (most weights constrained to zero) of the same transformer, evaluated on atomic (single-rule) and compositional (multi-rule) proofs, we vary which extracted object is reported (a compact prediction-preserving circuit, a broader graph that also keeps surrounding read, write, and routing structure, or the smallest subgraph meeting a post-ablation loss threshold), and whether each attention head's query and key are represented jointly or separately. Exact component-to-component edge overlap is low and sensitive to these choices, at times dropping to a random baseline, while two coarser summaries stay stable: the set of selected attention heads, and the circuit-size ranking of conditions that differ in which supervised checkpoint initializes reinforcement learning (RL). The largest accuracy gains from RL on compositional proofs come with the most structure beyond the atomic circuits. A circuit-level claim is therefore well defined only once one states which circuit is reported, the pruning threshold used to extract it, and the level at which circuits are compared. We distill these requirements into a reporting practice for circuit-extraction studies.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18921
PDF
-
From Trajectories to Instructions: Language-Conditioned Meta-Reinforcement Learning
2026-07-21 08:10:45Garvit Singla, Uma Maheswari Natarajan, Raghuram Bharadwaj DiddigiAbstract
Model-Agnostic Meta-Learning (MAML) is a widely used framework for reinforcement learning (RL) that enables efficient transfer by learning global policy parameters that can be rapidly adapted to new tasks. MAML training proceeds in two loops: an inner loop where the global parameters are adapted to task-specific parameters, and an outer loop where these task-specific parameters are evaluated and losses are back-propagated to improve the global parameters. Traditionally, the inner loop adaptation is performed by collecting trajectories from the task environment and applying gradient updates on the empirical expected return, which can be a costly operation. We note that it is the outer loop that drives the actual learning of global parameters, and therefore the inner loop adaptation mechanism need not be restricted to be gradient-based. This observation leads us to ask: Can we replace the inner loop trajectory collection and gradient update with a simpler, task-specific signal? In many practical settings, tasks are naturally accompanied by language instructions. Leveraging these instructions as a direct task-specific signal, we propose LA-MAML (Language Adapted MAML), which modifies the inner loop by adapting the global policy parameters in a single step through a learned embedding of the task instruction, replacing the inner loop trajectory collection and gradient-based updates. Experiments on the BabyAI benchmark demonstrate that LA-MAML achieves competitive or improved performance compared to baselines at a significantly lower per-iteration wall-clock training time. These results demonstrate that language instructions are an effective and efficient substitute for trajectory-based inner loop adaptation in meta RL.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18830
PDF
-
Stale but Stable: Staleness-Adaptive Trust Regions for Stabilizing Asynchronous Reinforcement Learning
2026-07-21 05:27:50Junyao Yang, Yucheng Shi, Zongxia Li, Zhongzhi Li, Ruhan Wang, Xiangxin Zhou, Kishan Panaganti, Haitao Mi, Leowei LiangAbstract
Asynchronous reinforcement learning improves throughput by decoupling rollout generation from optimization, but staleness is an inevitable byproduct compounded by policy lag, engine delays, and mixture-of-experts routing. From a trust-region perspective, this mismatch is critical: training-inference divergence governs approximation error in finite-horizon bounds, whereas PPO clipping only gates sampled outward updates, acting as a sampled surrogate rather than a full-policy constraint. As a result, high-staleness updates remain weakly controlled in the asynchronous regime where stale rollouts matter most. We introduce the Staleness-Adaptive Trust Region (SAT), which uses the detached sampled log-ratio as a practical staleness proxy, identifies high-mismatch tails within each batch via staleness-based kernel scaling, and contracts only the sign-selected endpoint of the nominal PPO interval. This preserves baseline behavior on ordinary tokens while enforcing more conservative updates on newly intercepted outward bands. We prove local interval containment and pointwise pessimism relative to PPO, showing how the adaptive rule reshapes update geometry under heterogeneous staleness. We evaluate SAT in a decoupled asynchronous RL setup built on Qwen3-30B-A3B-Base, using SGLang as the inference engine and Megatron for training. In this setting, SAT-GSPO w/ R3 achieves the best observed AIME24 avg@8, reaching 35.83 at lag 1 and 34.79 at lag 8, while SAT-GSPO reaches 34.17 at lag 1. Adaptive clipping and routing replay act as complementary stabilizers targeting mismatch tails and routing inconsistency, respectively. Overall, aligning clip intervals with staleness heterogeneity effectively stabilizes asynchronous RL.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18722
PDF
-
Strategy-Following Multi-Agent Deep Reinforcement Learning Considering Control Strategies Provided to Other Agents
2026-07-21 05:19:49Yamato Takahagi, Gentoku Nakasone, Yoshinari Motokawa, Toshiharu SugawaraAbstract
This study proposes a learning method for multi-agent systems that allows agents to be controlled through human manager instructions after learning and enables uninstructed agents to implicitly complement the overall work based on the actions of other agents. Multi-agent applications using deep learning have shown potential; thus, to achieve extensive social applications, humans should be able to control learned agents using simple methods to respond to environmental and social changes. Even without such changes, learned coordination often does not match the expectations of human managers, making it preferable to control coordination structures to match human intentions. Some studies have aimed to control agent behavior using simple instructions. However, they assumed that instructions are provided to all agents, which is time-consuming and not evident when designing a better cooperation regime. Ideally, specific agents should receive key action instructions, while others should automatically complete the remaining tasks. The proposed method, which extends previous work on controllability in multi-agent deep reinforcement learning, enables uninstructed agents to adaptively complement overlooked tasks and areas. The experimental results show that agents using the proposed method can shift to another cooperative structure and achieve better performance than those using conventional methods.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18719
PDF
-
Intelligent Multi-UAV Navigation in ITNTNs: A Hierarchical LLM Approach
2026-07-21 00:34:42Zijiang Yan, Hao Zhou, Wael Jaafar, Jianhua Pei, Ping Wang, Halim Yanikomeroglu, Hina TabassumAbstract
The deployment of high-speed Uncrewed Aerial Vehicles (UAVs) in 3D aerial highways necessitates robust coordination of physical flight kinematics and multi-tier network handovers. While Deep Reinforcement Learning (DRL) offers rapid tactical control, it lacks the zero-shot strategic reasoning required to quickly adapt to dynamic Integrated Terrestrial and Non-Terrestrial Networks (ITNTNs). Conversely, Large Language Models (LLMs) excel at semantic reasoning but suffer from high inference latency, rendering them unsuitable for real-time aerodynamic control. To bridge this gap, we propose a novel Hierarchical LLM-driven control framework. A massive cloud-based LLM deployed on a High-Altitude Platform Station (HAPS) manages slow-timescale global load balancing, while lightweight edge-LLMs on individual UAVs translate local observations into tactical sub-goals. These sub-goals guide a fast-timescale physical DRL controller to execute collision-free, handover-aware trajectories. Simulation results demonstrate that our agentic architecture significantly reduces collision rates and improves aggregate system throughput compared to existing baselines.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18604
PDF
-
Planning as Emergent Behavior in Reinforcement Learning with Relational Hidden States
Abstract
Reinforcement learning is conventionally divided into model-based and model-free methods. In this taxonomy, model-based methods perform lookahead planning over a learned world model, whereas model-free methods learn a reactive state-action mapping. Recent work, however, has shown that planning can emerge from model-free reinforcement learning alone. The conditions under which this behavior emerges from a pure reward-maximization objective have so far remained unclear. In this paper, we present evidence that, in the observed cases, the hidden-state structure of the neural architecture is the deciding factor. We find that a network of relational hidden states, each anchored to an environment state and exchanging messages along learned relations, acquires a planning mechanism. These hidden states recover the environment's transition structure in their learned relations, and improve the policy at decision time by planning over the learned graph. In a matched control agent that must additionally discover which cells represent which states, no such binding arises, and no planning follows from it. We argue that this explains the observed phenomenon of emergent planning in model-free reinforcement learning and raises the question of how common such emergent planning might be more generally. Finally, we hypothesize that the discovered mechanism could describe how planning emerges from pure reward maximization in the human brain through a neural architectural prior.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18589
PDF
-
Automated Data Engineering and Feature Selection for the Case Study of Warpage Detection in Fused Deposition Modeling
2026-07-20 21:15:12Saleh Valizadeh Sotubadi, Nazanin Mahjourian, Vinh NguyenAbstract
This study contributes toward development of an Automated Data Processing (ADP) framework designed to evaluate and reinforce optimal machine learning model-feature combinations for predictive tasks in fused deposition modeling (FDM) process datasets. The methodology is centered around a reinforcement learning-inspired policy updating mechanism, where multiple machine learning models are trained on both full feature sets and feature subsets selected through Shapley-based Explainable AI (SHAP XAI) across 217 datasets. At each episode, the framework assesses the predictive accuracy and F1-scores of each model-feature pair, computes a scalar reward, and updates $Q$ values to guide future model selection. SHAP XAI feature importance was employed to generate reduced yet informative feature subsets to enable the framework to explore performance with dimensionality. The policy was shown to evolve over multiple episodes, with reward distributions used to visualize performance stability. Overall, results indicate that leveraging the ADP framework through XAI algorithms successfully converges toward optimal model-feature configurations with improved accuracy and stability. Specifically, the proposed framework improves the test-set AUC from 0.9248 to 0.9731 and increases the mean reward value by more than fifty percent compared with the baseline full-feature configuration.
Abstract (translated)
URL
https://arxiv.org/abs/2607.18515
PDF
