RL for LLM Agents

RL: Reinforcement Learning · LLM: Large Language Model

Follow this topic Ask about this topic

Momentum

70 papers in the last four weeks, up 141% on the four weeks before. 0.7% of all new papers.

Jul 13Week of Sep 28

Latest papers 402

Sep 30, 2026cs.LG

From Imitation to Reward Discovery: On-Policy Warmup for Agentic RL

Reinforcement learning with a verifiable reward (RLVR) offers a scalable approach to training language-model agents, yet sparse outcome rewards can leave early training with little signal for policy improvement. We identify an On-Policy Acceleration Phenomenon: in our main comparisons, RLVR initialized with on-policy distillation reaches high performance earlier in training and achieves both higher average performance during subsequent RLVR and higher final performance than the alternative baselines. Motivated by this observation, we study On-Policy Warmup (OPW), a teacher-guided stage in which the student trains with teacher supervision on its own interaction trajectories before transitioning to RLVR. Unlike imitation on fixed teacher-generated trajectories, OPW targets states induced by the student's own decisions, including imperfect actions and recovery situations. We provide a theoretical explanation by connecting on-policy reverse-KL distillation to trajectory-level distribution matching. Under a competent teacher and sufficiently small population distillation loss, this connection yields a lower bound on initial verifier success and a corresponding bound on reward-discovery complexity. For group-relative RLVR, we further characterize when increased success probability produces more reward-informative groups. Together, our findings support on-policy distillation as an effective warmup for agentic RLVR and identify initial reward discovery as a mechanism that can contribute to the observed acceleration.

Yitong Qiao, Tiantian He, Lei Liu +4

Zhejiang University · Ant Healthcare, Ant Group

Agentic RL Reinforcement Learning with Verifiable Rewards On-Policy Distillation Sparse-Reward RL Policy Learning RL for LLM Agents

Sep 30, 2026cs.LG

Mitigating the Length-Scaling Tax with Online Distillation

Length scaling during reinforcement-learning (RL) post-training is often viewed as a sign of improved reasoning ability, especially on difficult problems, but may also make responses to already-solved problems unnecessarily verbose. We quantify this side effect as the length-scaling tax (LST): excess response length on already-solved queries without a commensurate accuracy gain. To mitigate LST, we propose Length Self-Distillation (LSD), which routes solved prompts to on-policy distillation and retains the original RL objective for unsolved prompts. LSD uses an exponential moving average of the online policy as its teacher, requiring no external model. We find that LSD achieves comparable or better performance than RL across multiple variants, while substantially curbing response-length growth on easy queries. LSD reduces LST from 19.0% to -3.7% on single-turn reasoning and from 31.4% to 13.7% on multi-turn agentic tasks, demonstrating that LSD effectively preserves concise response patterns on easy queries while supporting efficient exploration on difficult queries during RL post-training.

Xu Wan, Wenyue Xu, Shengjie Zhao +1

ByteDance Seed · Tongji University · Peking University

RL for Language Model Reasoning On-Policy Distillation RL Post-Training RL for LLM Agents

Sep 30, 2026cs.LG

Explicit Trajectory Diversity for RL-Based Post-Training of LLM Agents

LLM agents often admit multiple high-quality solutions to the same task, differing in reasoning structure, tool-use pattern, or interaction trajectory. Yet existing notions of diversity in LLM post-training are mostly implicit, arising from general stochasticity and regularization mechanisms rather than explicitly targeting task-relevant behavioral variation. While such implicit diversity can be useful, it does not directly specify which forms of behavioral variation should be encouraged for a given task. In this work, we study explicit trajectory diversity in RL-based post-training for LLMs. Our key idea is to define diversity through user-specified, task-specific trajectory descriptors, which map each sampled trajectory to an interpretable behavioral representation, and then measure diversity as a set-level functional over the resulting descriptor matrix. Building on this formulation, we introduce Trajectory-guided Joint Policy Optimization(TJPO), a single-policy framework that optimizes explicit diversity over sampled trajectory groups, avoiding the need for population-based policy training, and instantiate it within group-based policy optimization through trajectory-level learning signals. This design makes the diversity objective both interpretable and controllable. Experiments on Sokoban and ALFWorld show that TJPO improves task-specific trajectory diversity while maintaining competitive task performance. Descriptor and trajectory analyses show that the learned variation follows the specified behavioral dimensions and includes distinct successful strategies. Extra experiment results suggest that explicitly shaping trajectory diversity can help LLM agents satisfy user requirements and remain effective when task conditions change.

Huaiyu Fu, Heng Cao, Hao Wang +2

Microsoft

LLM Agent Training RL Post-Training RL for LLM Agents

Sep 29, 2026cs.CL

EVOKE: Eliciting World Knowledge in Agents for Transferable Decision-Making

Large language models (LLMs) are increasingly deployed as agents for multi-step decision-making, yet transfer poorly to unseen environments. World-model methods address this by training agents to predict future observations, at the cost of additional training and errors that compound when predictions are used for planning. However, for LLM agents operating in digital environments, much of this world knowledge is already internalized during pretraining, which shifts the problem from acquiring it to eliciting it. We argue that typical post-training provides little pressure for such elicitation, since supervision under a single goal at each visited state inadvertently drives policies to rely on superficial contextual habits. We introduce EVOKE, a post-training method that supplies this pressure through goal diversity at fixed states. Motivated by theory showing that an agent competent across diverse goals must encode a world model recoverable from its action preferences, EVOKE holds the environment state and interaction history fixed and ranks the same candidate actions under alternative goals, forcing action preferences to change, so that a policy relying on contextual habits or single-goal correlations cannot order them correctly. This implicitly elicits the policy's pretrained world knowledge to inform decisions. We evaluate EVOKE across diverse tasks in three backbones, demonstrating improved task performance, unseen environment generalization, and data efficiency. We further conduct controlled analyses to better understand what drives these gains. These findings offer a new perspective on eliciting internalized world knowledge for transferable action through direct decision supervision.

Yuhan Guo, Jinming Liu, Liang Xu +8

Shanghai Jiaotong University · Eastern Institute of Technology, Ningbo · Zhongguancun Academy, Beijing, China +1

LLM Agent Training Generalization in RL Goal-Conditioned RL RL for LLM Agents

Sep 29, 2026cs.AI

AdviSD: Learning to Advise Frontier LLMs via Targeted Multi-Turn Self-Distillation

A small trainable advisor can steer a frozen language-model executor using natural-language advice. In addition to learning from task rewards, the advisor can use feedback from completed interactions to improve its advice. However, a plausible correction need not change execution, yet learning from such corrections can still affect the advisor's future decisions in other contexts. In a shared-parameter model, we prove that such corrections can limit learning if their targets favor useful advice less strongly than those of other corrections. Keeping them less often than the rest improves the model's eventual performance compared to learning from every correction. Motivated by this, our method, Advisor Self-Distillation (AdviSD), pairs outcome-based reinforcement learning with self-distillation from a feedback-conditioned copy of the advisor selectively. Reflection proposes corrections, and the advisor scores the same recorded executor response with and without its issued advice, using the magnitude of the difference to select decisions for supervision. This approach does not require executor likelihoods or additional executor rollouts. Experiments with Qwen3-8B advisors for Gemini and Claude show that AdviSD outperforms advisor-GRPO by 4.2-6.4 percentage points on BFCL-v3 and by 3.9-5.1 score points on EnvScaler. The trained advisors generalize to out-of-domain tasks and transfer across different executor versions and model families. AdviSD also beats matched-count random selection, supporting the value of its selection rule.

Rishabh Agrawal, Hejie Cui, Shasha Li +2

Google · University of Southern California

Language Model Steering LLM Agent Self-Improvement Self-Distillation RL for LLM Agents

Sep 29, 2026cs.AI

ReMem: Rethinking Perception and Memory in Long-Context Recommendation Agents

Recent Recommendation Agents (RecAgents) offer a promising alternative by shifting recommendation to an active, user-side paradigm, where generative agents autonomously perceive external platforms, reason over user preferences, and execute decisions. However, existing RecAgents still suffer from two critical limitations: brittle item perception based on noisy and heterogeneous item pages, and inefficient long-context reasoning over extended user histories and multi-step interaction traces. To address these challenges, we propose a novel recommendation agent framework, termed as ReMem, that combines OCR-based multimodal perception with time-evolving dynamic memory. Instead of parsing raw HTML, ReMem observes item pages through screenshots and extracts structured multimodal information via an OCR tool, enabling a more humanoid and platform-agnostic perception mechanism. To support long-horizon preference modeling, ReMem further introduces a chunk-wise sequential memory update strategy, where the agent selectively maintains a fixed-size memory of informative historical interactions while processing arbitrarily long contexts with linear inference complexity and bounded context length. This design allows the agent to preserve evolving user preferences without relying on external memory modules or disrupting the standard autoregressive generation process. To enhance the dynamic memory instruction, we further develop a multi-memory GRPO variant, which propagates the final-answer advantage to all intermediate conversations that contribute to the final response. Extensive experiments on three datasets demonstrate that ReMem consistently outperforms state-of-the-art baselines, achieving an average improvement of 5.16% across three recommendation agent tasks, namely searching, ranking, and judging.

Haohao Qu, Yongcheng Jing, Chun Hin Chan +3

The Hong Kong Polytechnic University Hong Kong · Nanyang Technological University Singapore

Long-Context Modeling Recommender Systems Large Language Model-Based Recommendation RL for LLM Agents Memory-Augmented Language Model Agents

Sep 29, 2026cs.AI

Learning from Viable Failure Prefixes: Milestone Viability Potential Policy Optimization for Long-Horizon LLM Agents

Long-horizon LLM agents require reinforcement learning methods that can assign credit to intermediate decisions under sparse and delayed rewards. Existing group-based methods such as GRPO and GiGPO alleviate this issue by comparing rollout returns or repeated anchor states, but they still fail when the compared returns have no variation. We identify this failure mode as zero-credit failure: during early training, many failed rollouts contain useful prefixes, yet existing methods assign them no task-discriminative advantage. To address this issue, we propose Milestone Viability Potential Policy Optimization (MVPO), a potential-routed policy optimization algorithm that learns from viable failure prefixes. MVPO estimates prefix potential over Union-Find viability regions, repairs zero-credit groups with potential-difference advantages, and attenuates the potential branch according to relative performance progress. Experiments with Qwen2.5-1.5B-Instruct show that MVPO outperforms eight strong baselines, including GRPO and GiGPO. Under the same training length, MVPO improves over the GiGPO baseline by +4.4 success points on ALFWorld and +5.3 on WebShop, while adding only 0.16%-0.20% advantage-construction overhead.

Qi Zhou, Yuanfan Li

Xi’an Jiaotong University

Credit Assignment in RL Long-Horizon LLM Agents Policy Learning RL for LLM Agents

Sep 29, 2026cs.LG

VACE: Validation-Gated Alternating Co-Evolution of Agent Models and Harnesses

Language model agents can be improved by updating their model weights or refining the harness that guides task execution. These components are coupled: weight updates change how the model uses the harness, while harness updates change the trajectories used for training. We propose VACE, Validation-Gated Alternating CoEvolution, which alternates agentic reinforcement learning with trajectory-driven harness refinement. After each RL stage, VACE reuses the collected trajectories to propose a harness revision and evaluates the incumbent and candidate with the updated model held fixed. The candidate guides subsequent training only if it improves validation performance. With Qwen3.5-9B, VACE achieves 45.26% test accuracy on OfficeQA and a mean partial-credit score of 75.19% on AutomationBench, exceeding weight-only RL by 6.43 and 9.09 percentage points and ungated alternation by 4.59 and 6.95 points, respectively. Across 44 harness proposals, 17 reduce validation performance at the updated checkpoint and are rejected before subsequent RL training, highlighting the importance of validation gating.

Jiexing Qi, Yu He, Jun Liu +9

ICT AI Competence Center, Huawei Technologies Co., Ltd., Shanghai, China · Shanghai Jiao Tong University, Shanghai, China

Agentic RL Agent Harness Optimization RL for LLM Agents

Sep 29, 2026cs.CL

Traverse: Learning When to Remember, Reset, and Redirect for Long-Horizon Web Search

Long-horizon information-seeking agents often accumulate noisy or misleading context, causing early mistakes to persist and making recovery increasingly difficult. We introduce an autonomous search harness in which the agent manages its own search process through three states: Rubric, Answer, and Verify. The agent first defines criteria for a valid answer, searches under these criteria, and then independently verifies the result before deciding whether to terminate or continue searching. It is further equipped with a Seal Memory tool that enables active context management. Training this behavior with reinforcement learning, however, can induce Seal Collapse, resulting in unstable training and preventing the agent from reliably learning when and how to use its memory tools. We solve this with a simple strategy that trains only the final segment after context management. Our 35B model achieves 72.83 on BrowseComp, outperforming comparable open-source systems, and consistently improves over the base model across BrowseComp-ZH, xbench, DeepSearchQA, WideSearch, financial investigation, and product search. Ablations show that autonomous compression outperforms automatic compaction and validate our RL design.

Jingyuan Ma, Lynx Aster, He Zhang +5

State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University · ByteDance

Long-Horizon Agent Evaluation Web Search Agents RL for LLM Agents Memory-Augmented Language Model Agents

Sep 29, 2026cs.AI

EASE: Behavior-Adaptive Skill Curation for Self-Evolving Agents

Agent skills provide a lightweight mechanism for self-evolving agents to accumulate reusable procedural knowledge without updating model parameters. However, existing learned skill curators typically optimize curation without explicitly modeling downstream executor behavior. We show that this can cause systematic cross-executor degradation: curators trained with different executors perform best when paired with their own training executor, indicating that effective skill curation is executor-dependent. We formulate behavior-adaptive skill curation and introduce EASE, a framework that learns a single curator that adapts its decisions to different executor behaviors. EASE maintains an online behavioral profile of recent execution patterns and conditions the curator on this profile, the current trajectory, and retrieved skills to add, modify, or remove skills from an evolving repository. We train the shared curator jointly across multiple frozen executors with reinforcement learning, using retrieval-aware and behavior-aware temporal attribution to focus optimization on curation actions with observable downstream influence. Across ALFWorld, ScienceWorld, and WebShop, with executors ranging from Qwen3-8B/32B and GPT-OSS-120B to unseen Kimi K2.6, DeepSeek V4 Flash, and Gemini 3.5 Flash, EASE outperforms strong skill- and memory-based baselines without per-executor finetuning. EASE also maintains 34.5--41.0% fewer skills, improves skill retrieval by 36.3--38.7% and measured edit utility by 51.8--60.0%, and reduces deployment-time inference tokens by 9.1--14.5%. These results establish behavior-adaptive skill curation as an effective principle for building self-evolving agents.

Zhen Xiong, Qiaoyu Tan

New York University · New York University Shanghai

LLM Agent Skill Learning LLM Agent Skill Retrieval RL for LLM Agents Self-Evolving Agents

Sep 29, 2026cs.AI

BRIDGE: Bilevel Retrieval-Credit-Aware Agentic Reinforcement Learning

Agentic reinforcement learning (ARL) with verifiable rewards improves the ability of large language models (LLMs) to tackle knowledge-intensive tasks by learning to interleave search and reasoning. However, most existing ARL methods optimize only LLM-generated tokens and treat retrieved evidence as environment observations. This creates an information-credit gap: failures caused by missing or misleading evidence are attributed to the LLM policy rather than to the retriever, which motivates training the LLM and the retriever jointly. In this paper, we show that retrieval and LLM policy learning are order-sensitive: adapting the retriever before optimizing the policy yields a larger reward gain than the reverse order. To preserve this hierarchy while allowing both components to co-adapt, we formulate retrieval-augmented agentic RL as a bilevel optimization problem. To solve it efficiently, we introduce BRIDGE, a memory-efficient first-order bilevel method motivated by a loss-landscape analysis of the RL and retrieval objectives. Across seven open-domain QA benchmarks, BRIDGE achieves the highest average accuracy with both 3B and 7B backbones, improving the multi-hop average over the strongest baseline by 9.6 and 3.4 EM points, respectively. It also achieves the best averaged answer accuracy and reasoning quality across medical QA benchmarks.

Quan Xiao, Mingda Liu, Gaowen Liu +2

Department of Electrical and Computer Engineering, Cornell University · Institute of Integrated Research, Institute of Science Tokyo · Cisco Research

Reinforcement Learning Agentic RAG Bilevel Optimization RL for LLM Agents

Sep 28, 2026cs.LG

ROSS: Relearning from Self-Generated Rollouts through Selective Supervision

Large language model post-training generates self-generated rollouts through reinforcement learning and on-policy distillation, yet this experience is often treated as stale once the policy advances. Historical rollouts can remain compatible with a later policy while preserving behaviors that the policy no longer expresses reliably. However, they may also contain mistakes, abandoned attempts, and redundant actions that should not be imitated, motivating finer-grained selective supervision. We introduce ROSS (Relearning from Self-Generated Rollouts through Selective Supervision), which preserves the full historical trajectory as context while applying loss only to selected model-generated continuations. Across domain-specific reinforcement learning, multi-teacher on-policy distillation, and agentic reinforcement learning, ROSS consistently improves upstream checkpoints and outperforms baselines across mathematics, code generation, instruction following, and software engineering. On Qwen3.6-35B-A3B, ROSS improves the six-benchmark MOPD average from 58.40% to 62.20% and SWE-bench Verified from 64.20% to 68.40%. These results show that self-rollout training leaves behind reusable behavioral experience that can yield further gains through offline supervised fine-tuning (SFT), without additional policy rollouts.

Zhiwei Zhang, Huayu Deng, Fei Zhao +4

AllSpark Team

Supervised Fine-Tuning Self-Training for Language Models On-Policy Distillation RL Post-Training RL for LLM Agents

Sep 28, 2026cs.CL

Harness Learning Enables Generalizable Test-Time Adaptation

A language-model agent is jointly defined by its model and its harness, the executable program that organizes model calls, tool use, and information flow. Because different tasks call for different ways of organizing these operations, the harness needs to be adapted using feedback from the task at hand. We introduce harness learning, which trains a proposer model to revise a solver's harness using execution feedback. We formulate this process as meta-learning over executable programs, with harness revisions playing the role of weight updates in gradient-based adaptation. We train the proposer with reinforcement learning, using the task performance of revised harnesses as the reward. At test time, the proposer uses feedback from successive executions on a new task to refine the harness, without performing any parameter-space update. Experiments on reasoning and multi-hop question answering show that harness learning improves revision quality and that the ability to adapt at test time transfers to unseen tasks. Policies trained on individual revisions can continue improving harnesses over multiple rounds, while the benefits of training on revision sequences vary across settings. These findings suggest a path towards continually learning agents that turn accumulated experience into generalizable improvements.

Alvin Zhang, Xuecheng Liu, Zixuan Wang +6

Test-Time Adaptation Meta-Learning Agent Harness Optimization RL for LLM Agents

Sep 28, 2026cs.AI

Reinforcing Agentic Creativity in Scientific Ideation with Night Science

Large language models (LLMs) excel at structured, verifiable tasks, but their low-entropy bias can produce homogeneous and predictable outputs, limiting their utility for open-ended scientific ideation. Effective discovery, however, spans a broader creative spectrum: from structured day science to loosely structured, serendipitous night science that reaches ideas beyond those typically considered. We introduce AI Night-Scientist, an agentic framework that uses reinforcement learning to teach models when and how to depart from predictable reasoning. Grounded in cognitive science, we model creativity along three axes: action (what to do and how creatively), process (when to explore versus exploit), and outcome (the novelty and usefulness of the resulting idea). We use these axes to train models with GRPO, exposing them to varying degrees and forms of creativity throughout training. This produces substantially more diverse scientific proposals, expanding the range of research directions by 27.8% and contribution types by 14.9% over the base model. It also improves predicted citation impact by up to 32.0 percentage points and originality by 66.2 points. These gains cannot be reproduced by simply increasing decoding temperature; instead, we find that semantic guidance specifying what kind of creativity to pursue is critical. Overall, our results suggest that creativity is a learnable, multi-level ability that can be shaped to help researchers reach ideas beyond those typically explored by LLMs.

Priyanka Kargupta, Silviu Cucerzan, Shweti Mahajan +4

University of Illinois Urbana-Champaign · Microsoft · Microsoft Research

AI Agents for Scientific Discovery Scientific Hypothesis Generation Computational Creativity RL for LLM Agents

Sep 28, 2026cs.AI

Continuous Context Management

Long-horizon large language model (LLM) agents commonly retain their complete interaction history until compaction is triggered at a predefined threshold. We study Continuous Context Management (CCM), which performs compaction at every turn to prevent interaction history from accumulating in the active prompt. At each turn, a CCM agent emits an updated memory together with an environment action; its next prompt contains the original task, retained memory, and newest observation rather than the complete transcript. We first evaluate CCM without fine-tuning on TerminalBench-2 using Claude Sonnet 4.6, Claude Opus 4.6, GLM-5, and Kimi K3. CCM substantially reduces cumulative input usage and active-prompt size, although it lowers task success for most models while preserving performance for Kimi K3. We use GRPO with privileged full-history distillation to improve CCM in open-weight models. A frozen copy of the student's initial model scores each sampled student action under the complete history reconstructed from that student's rollout, providing dense action-token supervision without a separate teacher rollout or reference solution. On WebShop, this objective substantially improves CCM over GRPO at both evaluated model scales and surpasses full-history GRPO for Qwen3-4B-Instruct, though not for Qwen3-8B. On Endless Terminals, the augmented method provides a modest improvement over GRPO, with both CCM policies outperforming the untrained full-history baseline. These results demonstrate that CCM is a viable inference paradigm for agents operating with substantially reduced retained context and that its performance can be improved through reinforcement learning with privileged full-history distillation.

William Hoy, Jingxuan Fan, Nurcin Celik +1

University of Miami · Harvard University

Context Distillation Long-Horizon LLM Agents LLM Context Management RL for LLM Agents

Sep 28, 2026cs.AI

ASCT: Attentive Search over Counterfactual Trees for Credit Assignment in Agentic Reinforcement Learning

Terminal utility evaluates a complete agentic workflow, but learning requires credit for the decisions within it. We introduce Attentive Search over Counterfactual Trees (ASCT), a framework that turns training-time multi-step search into local action credit. At actor-visited states, an auxiliary tree evaluates alternative legal actions from the same recoverable prefix. Its action-value table is centered by the frozen actor's probabilities and supplies credit for PPO on actor-sampled trajectories. This protocol connects counterfactual evaluation to policy learning while deploying the actor alone. Uniform, UCT, and cost-aware AgentUCT instantiate the framework. On HotpotQA agentic retrieval-augmented generation, all three improve mean held-out utility over trajectory-return PPO and workflow-adapted VinePPO. Across three seeds, ASCT-AgentUCT reaches 0.6187 utility versus 0.5939 for VinePPO, with gains in answer F1 and execution cost, and uses 50.3% fewer recorded auxiliary Qwen tokens. Transfer and component-description studies examine the learned policies beyond the training setting.

Yang Li, Jinhan Yang, hai liu +8

University of Hong Kong · The Chinese University of Hong Kong · Jiangxi Science and Technology Normal University +1

Agentic RL Credit Assignment in RL Step-Level Credit Assignment in RL Monte Carlo Tree Search Policy Learning Counterfactual Credit Assignment+2

Sep 28, 2026cs.LG

GraphHCA: Closed-Form Hindsight Credit Assignment for Long-Horizon LLM Agents

Group-based reinforcement learning (RL) has advanced large language models (LLMs) and is increasingly extending to agentic tasks, where sparse terminal rewards make step-level credit assignment essential. Existing methods assign credit from what follows an action in sampled rollouts, but do not explicitly capture its retrospective relation to the realized outcome. Hindsight credit assignment (HCA) instead attributes credit through the ratio of hindsight to behavior-policy probabilities, but estimating the hindsight distribution requires an auxiliary model or an extra pass. To address this estimation bottleneck, we propose GraphHCA, a model-free realization of HCA that eliminates explicit hindsight-distribution estimation. For terminal-goal tasks with deterministic transitions, Bayes' rule reduces the hindsight ratio to a ratio of behavior-policy success probabilities at consecutive states. Taking logs yields a state-wise success potential, whose increment across a transition provides step-level credit. GraphHCA estimates this potential from pooled rollouts through a discounted recursion on the induced transition graph, which admits a unique fixed point on any directed graph. The resulting step-level signal is combined with the trajectory-level advantage, requiring neither a learned hindsight model nor an extra forward pass and recovering GRPO when the step-level weight is zero. Among all compared baselines, GraphHCA achieves state-of-the-art results on ALFWorld and WebShop at both LLM scales, and on Sokoban with a vision-language agent. For example, on ALFWorld it improves overall success rate by up to 24.6 points over GRPO and by up to 4.7 points over the strongest step-level baseline.

Haodong Zhu, Yangyang Ren, Changbai Li +4

Beihang University · Zhongguancun Academy · Communication University of China +1

Credit Assignment in RL Step-Level Credit Assignment in RL Model-Free RL Long-Horizon LLM Agents Policy Learning RL for LLM Agents

Sep 28, 2026cs.LG

Cross-Rollout Bellman Closure for Long-Horizon Agentic Reinforcement Learning

Group-based reinforcement learning such as GRPO trains LLM agents by comparing rollouts sampled for each task, without a learned critic. In long-horizon settings, these rollouts revisit shared anchor states, offering cross-rollout evidence for step-level credit. Ideally, step-level credit should incorporate evidence beyond the realized suffixes observed at an anchor while aggregating alternative continuations according to their empirical frequencies. Visit-local averaging pools realized suffix returns at shared anchors and respects observed frequencies, but does not recursively propagate evidence across rollouts, whereas shortest-path estimators have global reach but allow a rarely observed route to dominate an anchor's value. We introduce Cross-Rollout Bellman Closure (CRBC), which merges each rollout group into a finite empirical process with absorbing success and failure boundaries and evaluates its behavior-policy Bellman fixed point with one linear solve. This fixed point uses the same empirical action and transition frequencies to propagate evidence through shared anchors and aggregate alternative continuations. Backing up the resulting state values through observed transitions yields action values, whose gain over the corresponding state value provides step-level credit. A corresponding finite-depth family recovers visit-local return averaging at zero depth and converges to the exact closure as depth increases. The normalized closure credit is combined with the trajectory-level group advantage for policy optimization, without additional environment rollouts. Across ALFWorld, WebShop, and Sokoban benchmarks with multiple model scales, CRBC consistently improves final performance and learning efficiency. For example, CRBC outperforms the strongest evaluated baseline by 5.59 percentage points on ALFWorld with Qwen2.5-1.5B-Instruct.

Yangyang Ren, Haodong Zhu, Linlin Yang +3

Beihang University · Zhongguancun Academy · Communication University of China +2

Agentic RL Step-Level Credit Assignment in RL Temporal Credit Assignment Critic-Free RL Policy Learning RL for LLM Agents

Sep 28, 2026cs.LG

TIDE: Teacher-Student Transition via Informative Distillation and Exploration for Agentic RL

Effective multi-turn agents require interaction strategies that coordinate information gathering, actions, and feedback over long horizons. GRPO is a reinforcement learning algorithm used to train these agents, but sparse trajectory-level rewards limit early exploration in small models. Recent methods augment RL with on-policy distillation (OPD) from a stronger teacher. However, a fixed mixture assumes that teacher guidance and reward optimization should retain a constant relative role throughout training and across interaction turns. This assumption can fail at two scales. Globally, as training progresses, maintaining strong distillation pressure can constrain the model from moving beyond the teacher's capabilities. Locally, teacher--student disagreement identifies where the student departs from the teacher, but cannot tell whether that departure is exploration supported by better outcomes or low-quality policy drift. Our methodological insight is that teacher guidance and reward optimization should be dynamically rebalanced over training and jointly allocated across turns. We instantiate this insight in \tide. Globally, \tide uses the measured disagreement trend as a practical schedule signal, advancing an OPD-to-RL handoff when discrepancy reduction becomes slow but remains positive and progressively increasing the relative weight of RL. Locally, \tide jointly modulates teacher-guided and reward-driven updates: relative action value and disagreement prioritize the OPD signal, whereas relative action value supplies the RL advantage and normalized disagreement reweights it across turns. Coupled with the global handoff, \tide allocates stronger teacher guidance early and gives reward-driven updates greater relative weight later in training. Experiments across multiple benchmarks, student scales, and controlled ablations support the effectiveness of TIDE's adaptive OPD--RL coordination.

Yibin Huang, Xinming Xu, Conghui Zhu

Faculty of Computing, Harbin Institute of Technology · Tsinghua University

Agentic RL On-Policy Distillation Sparse-Reward RL RL for LLM Agents

Sep 28, 2026cs.AI

Proactive Dialogue Policy Optimization via Cognitive-State Transition

Proactive dialogue requires agents to continually adapt their policies to user feedback while progressing toward task objectives over multiple turns. To move beyond imitation learning on static datasets, recent approaches use user simulators to collect interactive data for policy optimization. However, many simulators do not explicitly model the evolution of user cognition, limiting the consistency and state dependence of feedback across turns. Moreover, representing each action only by a high-level strategy label overlooks the large utterance space and cannot distinguish alternative realizations of the same strategy. To this end, we jointly design a

\textbf{Cog}

nitive User

\textbf{Sim}

ulator

\textbf{(Cog-Sim)}

and

\textbf{C}

ognitive-

\textbf{S}

tate

\textbf{T}

ransition--Driven

\textbf{P}

olicy

\textbf{O}

ptimization

\textbf{(CSTPO)}

. Cog-Sim maintains the user's cognitive and affective states and generates responses through constrained state transitions across turns, so feedback depends on both the realized utterance and the user's current state. CSTPO organizes each action as a hierarchical strategy--utterance representation: a high-level strategy label constrains utterance sampling, and utterances are optimized within each label. Sparse complete-branch sampling reuses shared dialogue prefixes and estimates separate strategy-level and utterance-level advantages, enabling fine-grained optimization at both levels. Across three tasks, Cog-Sim exhibits monotonic dose--response relationships and is preferred over prompt-based simulators for naturalness. CSTPO improves Qwen3-14B's performance to a level comparable to that of GPT-5.5-based planning methods.

Minghui Ma, Mengqi Chen, Bin Guo +1

Northwestern Polytechnical University

Policy Optimization Conversational Agents User Simulation Task-Oriented Dialogue Systems RL for LLM Agents

Sep 28, 2026cs.AI

UniOPSD: Unifying Outcome and Hindsight Feedback for Agentic Reinforcement Learning

Reinforcement learning has become an effective approach to training language model agents, but sparse and delayed outcome rewards provide limited guidance for credit assignment across long interaction sequences. Recent work on on-policy self-distillation (OPSD) offers complementary supervision by evaluating a policy's sampled responses under privileged training-time context. However, our diagnostics show that positive average agreement between outcome and hindsight feedback coexists with substantial local disagreement, raising the question of how to allocate influence between them at each decision. We introduce UniOPSD (Unified On-Policy Self-Distillation), which unifies these feedback sources through adaptive local credit arbitration. UniOPSD constructs comparable credit estimates from environmental returns and successful-peer hindsight at shared interaction anchors. Historical agreement determines the global mixing level, while current signal availability and relative precision adjust each source's influence at individual decisions. The episode-level outcome contribution is retained, and bounded token modulation refines the fused step credit for policy optimization. With Qwen2.5-3B-Instruct and Qwen2.5-7B-Instruct, UniOPSD achieves ALFWorld success rates of

82.8\%

and

83.6\%

, WebShop success rates of

75.0\%

and

82.0\%

, and Search-QA aggregate accuracies of

45.3\%

and

49.8\%

, respectively. On 3B WebShop, UniOPSD improves over SDAR by

7.0

percentage points. Our code is available at https://github.com/Zenghuang-Fu/Uniopsd

Zenghuang Fu, Zhaoyang Li, Qiuyuan Ai +8

University of Chinese Academy of Sciences · Institute of Automation, Chinese Academy of Sciences · Peking University +3

Agentic RL Credit Assignment in RL Step-Level Credit Assignment in RL On-Policy Distillation Sparse-Reward RL Policy Learning+1

Sep 28, 2026cs.AI

SIPO: Selective-Inference Policy Optimization for Tree-Structured Agentic RL

Tree-structured reinforcement learning trains search agents by comparing alternative continuations and propagating terminal rewards to intermediate decisions. Adaptive expansion, however, creates a statistical asymmetry: an incumbent is selected using its own generation statistic, whereas fresh siblings are sampled after selection. When that statistic is associated with return, branch values can reflect selection history as well as continuation quality, even for a shared parent. We propose Selective-Inference Policy Optimization (\SIPO{}), which incorporates this distinction into tree-based credit estimation. Its scale-free branch criterion keeps generation scores and sibling penalties on a consistent relative scale; exchangeable branching supplies multiple fresh continuations from each selected parent; and order-statistic correction adjusts retained incumbent values using selection rank and the estimated score--outcome association. These mechanisms preserve the leaf budget and the host policy optimisation objective. Across seven QA benchmarks using Qwen3-4B, Qwen3-8B, and Qwen2.5-7B, \SIPO{} achieves the highest reported multi-hop and single-hop averages among the compared methods. On Qwen3-8B, it improves these averages over AT\textsuperscript{2}PO by

1.31

and

1.07

percentage points, respectively, and ranks first on six of seven benchmarks. Component ablations evaluate the individual and combined changes, while early-training paired diagnostics show a selected--fresh value gap alongside a near-zero fresh--fresh reference. Together, these results support accounting for selection history when constructing and evaluating search-agent rollouts. Our code is available at https://github.com/Zenghuang-Fu/SIPO

Zenghuang Fu, Ningqi Chen, Mingda Jia +10

University of Chinese Academy of Sciences · Institute of Automation, Chinese Academy of Sciences · The University of Hong Kong +4

Agentic RL Reinforcement Learning Policy Learning RL for LLM Agents

Sep 28, 2026cs.LG

Coding Agent Memory Post-training: Unlocking the Memory Potential of Pre-trained File Operations for Long-Horizon Tasks via Reinforcement Learning

Language-model agents increasingly tackle long-horizon tasks whose interaction histories exceed the model's active context. Recent work has begun to use reinforcement learning to make memory control part of the policy, often relying on predefined memory tools within domain-specific training environments of relatively short horizons. This setup ties learned memory behavior to environment-specific interfaces that lie outside the base model's pre-training and must be learned from scratch, so even after post-training, agents struggle to use memory in long-horizon tasks. To address these limitations, we introduce Coding Agent Memory Gym (CAMG), a suite of long-horizon agentic-RL environments spanning Shop, Coding, DeepResearch, and AutoResearch. Alongside each environment's native task interface, CAMG provides executable shell access and an episode-persistent workspace, enabling agents to create, revise, search, and reuse files as memory throughout an episode. We also introduce CAMG-RL, which trains a single policy jointly across all four environments with fully asynchronous PPO, learning this file-based memory behavior directly from downstream task reward, and we train CAMG-RL-4B and CAMG-RL-9B from Qwen3.5 models of matching size. On SWE-bench Verified and MLE-bench Lite, CAMG-RL-4B and CAMG-RL-9B are competitive with Qwen3.5-35B-A3B and Qwen3.5-122B-A10B, respectively.

Lirui Luo, Kelong Mao, Heming Xia +10

LLM Agent Memory Persistent Memory for Language Models RL Post-Training Long-Horizon Agent Tasks RL for LLM Agents Memory-Augmented Language Model Agents

Sep 28, 2026cs.CV

Marathoner: Ultra-Long-Horizon Autonomous Intelligence

Humans naturally possess the ability to work persistently toward long-term goals. Given a challenging task, humans can continuously work for months or even years to accomplish a specific objective. In this paper, we propose Marathoner, an autonomous agentic model possessing the ability of ultra-long-horizon execution. Specifically, we propose a comprehensive post-training pipeline to instill this critical capability into base model. For Ultra-Long-Horizon Task Synthesis, we leverage major release PRs containing 1000+ lines of new code from diverse GitHub repositories as the primary source for synthesizing challenging task-level data. Additionally, we introduce Multi-Task Chaining, which chains multiple generated tasks into a single more challenging task, enabling the synthesis of tasks with frontier-level difficulty. For rejection sampling finetuning, we combine strong teacher model with diverse harnesses to generate trajectories on our synthesized tasks and conduct supervised finetuning on base model with rejection sampled trajectories. For reinforcement learning, cold-started model performs real-world execution through harnesses in independent sandboxes during rollout process, effectively facilitating the acquisition of genuine ultra-long-horizon execution capability. We further propose a novel reward strategy, Later Stage Bonus Reward, which explicitly encourages model to perform meaningful maneuvers during later stages of execution. Through extensive evaluation on 5 benchmarks containing ultra-long-horizon tasks, Marathoner achieves consistent and substantial performance improvements over base model and even surpasses performance of strong proprietary model. Further analysis shows that Marathoner can consistently work for 10+ hours and conduct 1000+ tool calls on highly challenging tasks.

Zhang Ruiyang, Ou Jinpeng, Xie Yifan +4

FIC, University of Macau · Ant Group · Peking University

Long-Horizon Agent Tasks Long-Horizon LLM Agents RL for LLM Agents

Sep 28, 2026cs.AI

GlyphBench: A Playground for Language-Model Reinforcement Learning

We introduce GlyphBench, an environment suite for reinforcement learning (RL) post-training of language-model agents, with over 360 tasks spanning diverse games. GlyphBench renders spatial observations as two-dimensional Unicode grids and connects training, evaluation, and trajectory replay through a unified interface designed to support efficient and reproducible research. We use GlyphBench to study how observation interfaces, reasoning effort, and agent harnesses affect performance, and how RL configurations shape learning dynamics. Our results show that glyph observations outperform native text and pixels in our Craftax experiments, with further gains on several BALROG environments. RL on 100 GlyphBench tasks improves Qwen3.5-4B on held-out Reasoning Gym problems, reaching 63.48% accuracy and outperforming the base model, a math-trained baseline, and a code-trained baseline. These experiments provide empirical evidence that reasoning gains from gameplay can yield stronger transfer than math or code. Together, these results highlight GlyphBench's value as a testbed for systematic research on how language-model agents learn, interact, and generalize.

Roger Creus Castanyer, Marc-Alexandre Côté, Matthew James Sargent +3

Mila - Quebec Artificial Intelligence Institute · Université de Montréal · Vmax

RL Benchmarks RL for Language Models RL for Language Model Reasoning RL for LLM Agents

Sep 28, 2026cs.LG

Learning Perturbation Robust Policies for LLM Agents with Stable Optimization

Reinforcement learning (RL) has become an effective post-training paradigm for long-horizon large language model (LLM) agents. However, we find that the resulting policies can be sensitive to various policy perturbations, such as hidden-state noise, pruning, and quantization. In this work, we study how to improve perturbation robustness during policy optimization. We first introduce the notion of a perturbation robust policy and analyze conditions under which perturbed policy updates preserve stable monotonic improvement. Based on this analysis, we introduce Stable Perturbation-Robust Policy Optimization (SPrPO), which applies adaptive and sensitivity-aware perturbations during RL training. We evaluate SPrPO on ALFWorld and WebShop and conduct systematic experiments across multiple perturbation types and scales, showing improved perturbation robustness while maintaining stable policy optimization.

Pengxin Wang, Yuanzhe LI, Yuxin Ren +2

Department of Electrical and Computer Engineering University of Arizona

RL Post-Training Robust RL Policy Learning RL for LLM Agents

Sep 27, 2026cs.LG

QwenGyre: An Elastic Reinforcement Learning Framework for Training xLong-Horizon Agents

Large language model (LLM) agents increasingly undertake extreme-long (xlong) horizon tasks, where a single execution can span hours, hundreds of model--environment interactions, and nearly 1M tokens per rollout. Applying online reinforcement learning (RL) to such executions poses two fundamental challenges: (1) severe execution variance and prolonged rollout delays cause massive GPU idling; and (2) complex non-linear branching generates massive trajectory redundancy, crippling training efficiency. To address these, we presents QwenGyre, an end-to-end framework for xlong-horizon online RL. QwenGyre elastically reallocates GPUs between rollout and training without interrupting live executions, while its trajectory processor reconstructs branching histories, scores partial progress, and deduplicates redundant paths to bound training costs. Scaled to our flagship model, Qwen~3.8 2.4T, with 700K tokens per rollout, QwenGyre yields a 6.0% absolute gain on NL2RepoBench (52.5%

\to

58.5%) in 48 steps. Across our evaluations on diverse domains of training datasets, QwenGyre delivers up to

1.85\times

and

1.78\times

speedups over Colocate and Async, respectively.

Weiqi Wang, Yuxin Zhou, Mouxiang Chen +9

Alibaba Token Hub, Alibaba Group · University of Science and Technology of China · Tsinghua University

Reinforcement Learning Long-Horizon Agent Tasks Long-Horizon LLM Agents RL for LLM Agents

Sep 27, 2026cs.AI

CompoWorld: Compositional Environment Scaling for General Agents

Automatically generated environments provide a scalable source of interaction data for training general agents. However, existing approaches mainly generate tasks within a single environment, while real-world workflows require agents to connect information and actions across multiple services. We introduce Compositional Environment Scaling (\textbf{CompoWorld}), which expands the task space by composing a finite library of reusable services. Coding agents turn tool specifications into verified services with typed states and shared interfaces, while a world model handles tools that cannot be reliably implemented. A random-walk procedure connects services through dependency graphs, enabling the generation and verification of tasks that require information to flow across services. Verified trajectories support supervised fine-tuning (SFT), while our Completion-Focused Rubric Reward guides reinforcement learning (RL) toward full task completion by emphasizing criteria with lower pass rates within each rollout group. We construct 448 services exposing 10,130 tools and use 3K SFT trajectories and 1K RL tasks to train Qwen3.6-35B-A3B. Experimental results show that CompoWorld improves on its backbone by 9.17 points on average across eight benchmarks. On AutomationBench, it surpasses frontier models such as Claude Opus 4.6 and leads all compared agent-specialized 35B-A3B models.

Xiao-Wen Yang, Weiyi Xu, Wen Da +9

AllSpark Team

Tool-Augmented Language Model Agents Tool-Using Agents Procedural Content Generation LLM Agent Training RL for LLM Agents

Sep 27, 2026cs.LG

Climbing the Hill: Prompt Injection Red-Teaming Against Frontier Models with Curriculum Reinforcement Learning

Prompt injection is a leading security risk for LLMs and LLM-based applications such as agents. State-of-the-art red-teaming methods for prompt injection leverage reinforcement learning (RL) to train an attacker LLM to generate effective injected prompts. However, when targeting frontier LLMs such as GPT-6-Luna, a major challenge is the cold-start problem: every attack attempt by the attacker LLM fails and thus receives zero reward, providing no signal for learning. In this work, we propose a curriculum learning-based method to address the cold-start problem. In particular, we propose to train the attacker LLM against a sequence of increasingly robust target LLMs, with each stage warm-starting from the attacker LLM obtained in the previous one. However, simply training against a weak target (e.g., GPT-4o-mini) may not sufficiently prepare the attacker LLM to obtain useful learning signals against a frontier LLM (e.g., GPT-5.6-Terra). Instead, we find that the design of the curriculum is critical: after each stage, the attacker LLM needs to partially succeed against the next target LLM such that it can learn from successful attempts to attack the new target. Our extensive evaluation shows that our method can effectively red-team frontier LLMs, achieving an attack success rate (ASR@10) of 93.8% and 45.0% against GPT-5.6-Luna and GPT-5.6-Terra on AgentDyn, whereas state-of-the-art RL methods such as RL-Hammer and PISmith achieve 0% ASR under the same setting. Moreover, we find that the attacker LLM transfers across targets, e.g., an attacker LLM trained to defeat one strong LLM (GPT-5.6-Terra) also succeeds against six other frontier LLMs (e.g., GPT-6-Luna) it was never trained on. Our code is available at https://github.com/albert-y1n/PIForge.

Chenlong Yin, Xiaolong Jin, Wei Zou +2

The Pennsylvania State University · Purdue University

Adversarial Attacks on LLMs LLM Red Teaming Curriculum RL Prompt Injection Attacks on LLMs Prompt Injection Attacks on AI Agents RL for LLM Agents

Sep 27, 2026cs.LG

Beyond Timestamps: Decision-Aligned On-Policy Distillation for Long-Horizon Agents

Reinforcement learning with verifiable rewards (RLVR) often relies on sparse outcome rewards, providing coarse supervision for long-horizon agents. On-policy self-distillation (OPSD) complements this signal with dense privileged feedback. However, we identify \emph{Decision--Timestamp Mismatch}: privileged guidance may be misaligned with the student's functional decision because the corresponding decision can occur at a different timestep, while the student's decision itself may span multiple timesteps rather than being tied to a single timestamp. Thus, timestamp-local supervision can misalign both the context and the temporal scope of credit. To address this mismatch, we introduce \textsc{AlignOPSD}, following the principle of aligning supervision before assigning credit. Decision-Aligned Supervision Rectification re-scores the same student-sampled response in functionally matched contexts across sibling rollouts to calibrate local teacher evidence. Semi-Markov Hierarchical Credit Assignment then derives variable-duration decision spans from correspondence changes and uses rectified evidence to allocate outcome-grounded credit across spans and their constituent turns. We evaluate \textsc{AlignOPSD} with Qwen2.5-3B and Qwen2.5-7B on ALFWorld, WebShop, and Search-QA against representative baselines. \textsc{AlignOPSD} outperforms both GRPO and StepOPSD across all eight backbone--aggregate-metric comparisons, improving on GRPO by 5.5--8.7 % and ranking first in six. Additional analyzes examine the two alignment stages and hyperparameter sensitivity between tasks. Our code is avaliable at https://github.com/mingju-c/Align-OPSD

Mingju Chen, Can Lv, Jinrong Liu +3

Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing, Beihang University · School of Artificial Intelligence, Beihang University · Tsinghua University

Credit Assignment in RL Reinforcement Learning with Verifiable Rewards On-Policy Distillation RL for LLM Agents