On-Policy Self-Evolution

Follow this topic Ask about this topic

Momentum

7 papers in the last four weeks, up 75% on the four weeks before. 0.1% of all new papers.

Jul 6Week of Sep 21

Latest papers 39

Oct 1, 2026cs.CL

My FAULT: Self-Diagnosis as Credit Assignment in Self-Evolving Agentic Reinforcement Learning

Agentic reinforcement learning (RL) has emerged as a powerful approach for training large language model agents on multi-step tasks, yet reliance on terminal outcome rewards creates two credit-assignment problems, particularly in long-horizon tasks. First, same-outcome rollout groups provide no learning signal from terminal rewards. Second, terminal rewards provide only trajectory-wide feedback, making it difficult to identify which decisions caused a failure. Recent work supplements terminal rewards with finer-grained information from trajectory analysis, such as natural-language reflections on intermediate decisions and errors. However, natural-language diagnoses are difficult to use directly for credit assignment: their error claims may be unreliable, and they do not quantify how much each error should affect learning. We propose Self-Diagnosis-guided Terminal Credit Redistribution (FAULT), which turns diagnosed errors into explicit step-level credit anchored by terminal outcomes. FAULT checks diagnostic evidence and learns relative error costs from task outcomes. During training, the policy and self-diagnoser co-evolve, while error costs are updated online from recent outcomes. On ALFWorld, FAULT recovers learning signals from same-outcome groups, reaching 95% signal coverage versus 41% for GRPO and 72% for GiGPO, while better localizing credit to specific error steps. Across two model scales, FAULT delivers strong. improvements on the long-horizon ALFWorld and WebShop tasks while remaining competitive on short-horizon Search-based QA.

Yihua Zhu, Qianying Liu, Weixu Qiao +10

Alibaba Group · Kyoto University · NII LLMC +6

Credit Assignment Agentic Reinforcement Learning Step-Level Credit On-Policy Self-Evolution Diagnosis

Sep 28, 2026cs.LG

Direct Self-Evolving Optimization: Evolving LLMs without Challenger Training

Self-evolving language models improve by generating tasks and learning from their own feedback, but adapting the task generator often requires a separate challenger-training loop. Can we generate tasks adapted to the current solver without explicitly training a challenger? We introduce \textbf{D}irect Self-\textbf{E}volving \textbf{O}ptimization (DEO), which replaces challenger parameter updates with solver-guided task sampling. The KL-regularized challenger objective defines an exponential tilt of a fixed base task distribution. DEO uses this distribution as a sampling target: a frozen LLM generates and mutates tasks, the solver scores them, and an approximate Metropolis selection rule refines the training pool. Only the solver is trained. Theoretically, for an idealized variant that samples exactly from the tilted distribution, and under regularity, local gradient-dominance, and initialization conditions, we show that DEO learns distributionally robust reasoning ability. In experiments, DEO achieves reasoning performance competitive with R-Zero while using over

50\%

less wall-clock training time, and improves reasoning accuracy over a no-walk ablation. Replacing the task generator with a frozen API-only LLM further improves the local solver, illustrating a capability enabled by removing challenger training.

Yuyang Deng, Yu Wang, Jiayun Wang

Accenture, Center for Advanced AI · Georgia Institute of Technology

On-Policy Self-Evolution LLM Reasoning Strategies Adaptive Sampling Data Generation Competition

Sep 27, 2026cs.AI

COEVO: Co-Evolving Context and Parameters for Recursive Self-Improvement

Recursive self-improvement (RSI) seeks to move large language models beyond static training pipelines toward systems that can participate in improving their own future behavior. Existing approaches largely follow two directions: updating model parameters through online learning, or improving the external context through search, reflection, and prompt optimization. Although both mechanisms can support continued improvement, they are typically studied independently. This separation overlooks an important interaction: the context shapes the experience from which a model learns, while an evolving model may interpret and utilize the same context differently over time. We therefore formulate RSI as a problem of parameter--context co-evolution, where model parameters and the learning context adapt within a shared feedback loop. We introduce COEVO, a framework that updates model parameters from on-policy experience while adapting contextual guidance according to the state of the evolving policy. Policy entropy and prompt-conditioned attention are used as complementary signals to guide this adaptation. Experiments show that COEVO consistently improves task performance over fixed-context reinforcement learning and produces policies that are more robust to changes in system prompts. More broadly, our results suggest that external context should be viewed not merely as a fixed interface to a large language model, but as an adaptive component of recursive self-improvement.

Siwei Chen, Xinping Bao, Xinyu Cai +3

Peking University · Emotional Machine

On-Policy Self-Evolution Recursive Self-Improvement Iterative Co-Training

Sep 21, 2026cs.LG

When Tomorrow Becomes Today: Self-Evolving Policies for Agentic Time-Series Forecasting

Agentic time series forecasting concerns systems whose underlying mechanisms evolve, making the relative effectiveness of numerical models, reasoning strategies, and intervention rules inherently time-varying. Consequently, a time series agent must adapt the forecasts it produces and the orchestration policy that determines which components to trust and how to coordinate them. The deployment process naturally provides supervision for this adaptation as forecast horizons elapse and realized targets reveal the effectiveness of earlier decisions. Committing all numerical expert forecasts and candidate agent paths before target observation allows each realized outcome to evaluate the entire alternative set, providing delayed feedback without additional annotation. However, existing time series agents primarily incorporate prior experience through forecast refinement, reflection, or retrieval, without systematically converting realized outcomes into persistent updates to the joint orchestration policy governing later origins. To exploit this delayed feedback systematically, we introduce TimEvolve, a frozen-backbone time series agent that converts each realized outcome into persistent joint updates of expert trust, agent path selection, and intervention strength. A temporally ordered predict, reveal, and update protocol applies this feedback to subsequent forecasts. Experiments across eight Time-MMD domains show that TimEvolve achieves the best average MSE and MAE ranks among fifteen methods and the lowest errors on both metrics in seven domains. These results demonstrate the value of learning forecasting policies from the futures encountered during deployment.

Yifan Hu, Xilin Dai, Zhiyuan Qu +4

Ant International · Tsinghua University · The Chinese University of Hong Kong

Time Series Forecasting Data-Driven Forecasting On-Policy Self-Evolution Orchestration Expert Demonstrations

Sep 15, 2026cs.AI

EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents

Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment. This limits their ability to adapt how they gather evidence, invoke tools, verify signals, and manage risk under changing market regimes. We introduce EvolveTrade, a self-evolving framework that treats the system prompt of a tool-using trading agent as a text-parameterized policy. After each update interval, a Policy Agent revises this policy using accumulated decision traces and realized portfolio feedback, while keeping the backbone LLM fixed. The updated policy is then used for the next batch of trading decisions, enabling the agent to refine its information-acquisition and portfolio-construction procedure over time. Experiments across multiple market regimes and two LLM backbones show that EvolveTrade often improves Sharpe Ratio and Cumulative Return over fixed-policy LLM baselines, achieving the improved SR and CR in most evaluated settings. Behavioral analyses further show that self-evolved policies increase code-mediated analysis and activate regime-relevant computations; case-level policy-to-return attributions trace how policy-induced allocation changes contribute to realized return differences. These results suggest that adapting the reusable procedure governing tool use is a key direction for building more robust LLM trading agents.

Sehee Kim, Yumin Choi, Minki Kang +1

KAIST

On-Policy Self-Evolution Alphaevolve Refinement

Sep 14, 2026cs.AI

EvoRS: On-Policy Self-Evolution of Reward Systems for Open-Ended Reinforcement Learning

Open-ended reinforcement learning often relies on rubric-based rewards for tasks without directly verifiable answers. Yet the policy and reward system form a dynamic feedback loop: as the policy optimizes the current reward, an initially useful reward system may become unreliable due to reward hacking or reduced response discriminability. The reward system should therefore evolve rather than remain fixed during training. Existing dynamic-rubric methods adapt evaluation criteria, but reward failures can also arise from scoring mechanisms or signal composition. We introduce EvoRS, a self-evolving RL framework that evolves the reward system from on-policy experience, representing it as an executable Reward-DAG. Specifically, an agentic designer updates this system from on-policy rollouts and reward traces to maintain train-time reliability. Across writing and roleplay, EvoRS achieves the best quality under all three judges, outperforming the policy by

2.107

and

4.767

points, respectively, while reducing reward hacking and coverage failures and preserving reward informativeness. Ablations confirm that a comprehensive fixed reward system cannot remain reliable in open-ended tasks and must evolve throughout training.

Weiyuan Li, Aili Chen, Xintao Wang +11

School of Data Science, Fudan University · Shanghai Key Laboratory of Data Science · College of Computer Science and Artificial Intelligence, Fudan University +3

On-Policy Self-Evolution Offline Reinforcement Learning

Sep 8, 2026cs.CL

Experience Funnel: A State-Policy Alternating Loop for Self-Evolving Agents

Autonomous agents powered by large language models (LLMs) continuously accumulate experience through interaction, creating an opportunity to improve future behavior through self-evolution. A fundamental challenge is how to transform abundant, task-specific interaction experience into reusable model competence without sacrificing the ability to adapt rapidly to newly observed evidence. Explicit textual states, such as skills and agent harnesses, provide fast, human-readable and editable adaptation, but incur persistent dependence on external context; parametric policies provide compact and reusable competence, but are substantially slower to update. We present \textit{Experience Funnel}, a self-evolving framework that couples fast state adaptation with slow policy consolidation in an alternating loop. Interaction trajectories are first distilled into an explicit textual state, where newly acquired experience can be rapidly incorporated and validated. The framework then selectively identifies state-enabled behavior that remains useful across state revisions and consolidates it into the policy through transition-aware distillation. The updated state--policy pair subsequently generates new rollouts, providing fresh evidence for the next round of state adaptation and policy consolidation. Experiments across diverse agent benchmarks show that \textit{Experience Funnel} consistently improves agent capability over state-only evolution and policy-internalization approaches, while progressively converting useful explicit experience into autonomous policy competence.

Wenbo Gao, Zhaomou Song, Zhiyuan Ji +7

1The Hong Kong Polytechnic University · 2Huawei · 3Renmin University of China

On-Policy Self-Evolution Self-Evolving Agents Self-Evolution Loop

Sep 8, 2026cs.LG

Environments as Scaffold: Enriching Feedback to Bootstrap Self-Evolving Agents in Long-Horizon Tasks

Large Language Models demonstrate remarkable proficiency in static reasoning, yet training them as autonomous agents through Reinforcement Learning (RL) for long-horizon tasks is often hindered by severe reward sparsity. While conventional \textit{agent-side warming} up via supervised fine-tuning (SFT) can alleviate this, it is frequently limited by data scarcity and constrained exploration. To address this, we propose a paradigm shift to \textit{environment-side adaptation} by constructing \textbf{F}eedback-\textbf{E}nriched \textbf{E}nvironments (\textbf{FEEs}). Through a pilot study, we establish a feedback design strategy that reformulates environments by transitioning from action guidance to observation enrichment during the later stages of both intra-episode exploration and inter-episode evolution. Large-scale experiments on SciWorld and BFCL benchmarks using various Qwen3 model scales and RL algorithms such as GRPO, GSPO, and DAPO demonstrate that FEEs consistently yield performance improvements over standard settings. Furthermore, our analysis reveals that training with FEEs \textbf{(1)} stabilizes training dynamics by reducing entropy volatility, \textbf{(2)} facilitates proactive state-space exploration in difficult tasks, \textbf{(3) }ensures the internalization of environmental guidance into policy weights rather than acting as a mere inference-time prior, and \textbf{(4) }identifies intra-group feedback consistency as a critical boundary for stable optimization.

Hongbang Yuan, Zhuoran Jin, Yixin Cao

1Fudan University · 2CASIA · 3Shanghai Innovation Institute

On-Policy Self-Evolution Self-Evolving Agents Long-Horizon Task Planning Feedback Environment Scaffolds

Sep 1, 2026cs.AI

ARISE-RL: Agentic Rubric-Grounded Iterative Self-Evolution with Reinforcement Learning

Training open-ended agents via reinforcement learning (RL) is hindered by the lack of verifiable gold answers and scalable rubrics. Moreover, even near the model's capability boundary, long-horizon open-ended agentic tasks often yield brittle and unstable rewards, resulting in weak or noisy rollout contrast that obscures fine-grained optimization signals for group-based policy learning. To address these challenges, we propose ARISE-RL, a novel full-cycle self-evolution framework that couples a task/rubric Generator and a reasoning Solver through rubric-mediated co-evolution. The Generator grounds tool-related rubric criteria in real tool observations and is rewarded for producing valid, intermediate-difficulty tasks aligned with the Solver's evolving capability boundary. The Solver, in turn, learns from fine-grained rubric satisfaction signals through multi-step reasoning and tool use. We further introduce Reward-Gated Self-Evolution Distillation (RG-SED), which selectively distills a memory-augmented variant of the same policy back into itself only when the memory yields empirical reward improvement, thereby reducing distribution mismatch and avoiding blind imitation of noisy guidance. Finally, to support rigorous evaluation, we present ECR-Bench, an expert-calibrated rubric benchmark suite covering single-tool deep research and multi-tool travel planning. Extensive experiments demonstrate that ARISE-RL consistently achieves robust and stable overall state-of-the-art performance across all evaluated benchmarks.

Fanrui Zhang, Ruixue Ding, Qiang Zhang +13

Alibaba ATH Token Foundry · Hema, Alibaba Group

On-Policy Self-Evolution Icr-Rl Self-Evolution Agentic Co-Evolution

Aug 10, 2026cs.AI

OpenLoopEvolve: A Verifiable Self-Evolution Framework for Loop Policies in Long-Horizon Complex Tasks

Long-horizon complex tasks require agents to repeatedly observe states, formulate plans, invoke tools, verify results, and recover from failures in continuously changing environments. However, such control experience often remains confined to a single context or a fixed prompt, and is difficult to accumulate and reuse across historical traces. This paper presents OpenLoopEvolve (OLE), a self-evolution framework centered on the Loop Policy. OLE represents an agent's observation, planning, memory, action, verification, recovery, stopping, and budget control as portable policy assets with versions and lineages, and provides online and offline evolution modes that can be selected according to practical needs: the online mode triggers candidate generation based on feedback from continuous operation, whereas the offline mode searches for candidate policies from archived traces and failure evidence. Both modes share an evolution mechanism consisting of autonomous proposals by a large language model, Champion--Challenger paired evaluation, and robust release. Policies released online are activated at a subsequent task boundary, monitored using subsequent feedback, and rolled back to their parent versions when degradation conditions are met. On the simulated business benchmark YC-Bench, both modes improve aggregate task performance, task success rate, and risk metrics relative to a fixed initial Loop Policy. The results indicate that treating the Loop Policy as a governable asset can support the accumulation, comparison, release, and reuse of control experience and improve agent performance on long-horizon complex tasks.

Siqi Wang, Xinlin Li, Zhenglin Li +1

On-Policy Self-Evolution Long-Horizon Agents Self-Evolution Agent Loop Long-Horizon Task Planning

Aug 4, 2026cs.RO

EvoHIL: Self-Evolving Reward and Flow-Matched Policy Optimization for Robust Human-in-the-Loop Reinforcement Learning

Human-in-the-loop reinforcement learning (HIL-RL) enables robots to learn contact-rich manipulation from limited real-world interaction, but deployment exposes three coupled limitations: static visual reward models fail under scene changes; independently sampled actions cause temporally inconsistent motion; and vision-based policies remain sensitive to appearance shifts. We present EvoHIL, a unified framework that adapts the reward model, action generator, and visual do main within a staged human-in-the-loop learning process. First, self-evolving reward (SER) adapts the success classifier from human-confirmed positives and provisional weak negatives. Second, Action Flow Stabilization (AFS) generates temporally coherent action chunks through flow matching, grounding policy updates in executed action prefixes and demonstrated behavior. Third, retention-aware offline fine-tuning replays relit interaction data while anchoring the AFS actor-critic to prior behavior, adapting the visual domain without additional robot interaction. Across six manipulation tasks on Franka FR3 and SO-101 arms under a controlled lighting shift, EvoHIL improves task success, agreement with human-confirmation labels, motion smoothness, and completion time relative to human-in-the-loop and imitation baselines.Project page: https://anonymous4366.github.io/EvoHIL/

Shuoqin Zhang, Tongtong Cheng, Xiru Gao +7

Human-In-The-Loop Reinforcement Learning From Human Feedback On-Policy Self-Evolution Stabilization

Aug 3, 2026cs.LG

RoMeRL: Balancing Feedback Coverage and the Memory-Reward Trap in Self-Evolving Agent Memory via Reduced-Order Utility States

Learning-based memory systems for self-evolving LLM agents face two tightly coupled challenges. First, trajectory-indexed utilities grow with the interaction history, thereby dispersing limited feedback over an ever-expanding state space. Second, because trajectory-level rewards are jointly assigned to co-retrieved memories, irrelevant experiences may receive misleading utility updates and consequently enter the memory-reward trap. To address these challenges, we introduce Reduced-Order Memory Reinforcement Learning (RoMeRL), which represents the growing trajectory-indexed utility space using a fixed-dimensional per-task memory state factorized by outcome polarity and memory dynamics. RoMeRL incorporates new experiences through a fixed set of semantic coordinates whose contents are updated or replaced over time, thereby concentrating feedback over a bounded utility support. Theoretically, we show that this reduced-order parameterization increases the average feedback received by each utility coordinate and characterize the steady-state occupancy of erroneous coordinates under a generic coordinate-transition model. Empirically, across ALFWorld and LifelongAgentBench, RoMeRL improves task performance, reduces the Cold-Q ratio by 80.0%, increases feedback density by approximately 6.0 times, reduces the maintained memory size by 84.4%, and cuts LLM calls by 21.1%. These results show that reduced-order utility states support efficient self-evolving agent memory while limiting persistent reward contamination. Code is available at: https://github.com/YOUNG-fnxm/RoMeRL

Yi Yang, Zhennan Chen, Yihong Zhuang +5

1Nanjing University · 2Xiamen University · 3Zhejiang University

On-Policy Self-Evolution Peak Memory Feedback Reduced-Order Models

Aug 3, 2026cs.CL

Self-Improving Large Language Models via Progressive Experience Evolution

Large language models (LLMs) capable of self-improvement require not only effective policy optimization, but also a principled mechanism for transforming transient interaction experience into persistent model capabilities. Existing self-improvement paradigms remain fragmented: test-time methods can explicitly extract experience but cannot internalize it into model parameters, whereas training-time optimization methods can update model parameters but lack an explicit mechanism for accumulating transferable experience. Bridging these two paradigms requires a critical intermediate stage that remains underexplored, namely \emph{experience distillation}. To address this gap, we propose \textbf{SPEE} (\textbf{S}elf-\textbf{P}rogressive \textbf{E}xperience \textbf{E}volution), a unified post-training framework that sequentially performs explicit experience evolution followed by implicit policy optimization. During explicit experience evolution, SPEE reflects on trajectories collected from multiple interactions to extract, verify, and progressively evolve transferable experience, which is subsequently internalized into the policy through privilege-guided On-Policy Self-Distillation (OPSD). During implicit policy optimization, reward-driven reinforcement learning leverages these internalized priors to explore novel solution strategies. In the experience evolution stage, a continuously evolving global experience pool consolidates knowledge from both successful and failed trajectories, filters out low-utility experience, and mitigates post-hoc rationalization induced by individual trajectories. Experiments on five mathematical reasoning benchmarks demonstrate that SPEE consistently outperforms both test-time and training-time self-evolution baselines across three model scales. The source code is available at https://github.com/rrrsj/SPEE.

Shijie Ren, Xiting Wang, Meng Li +8

Gaoling School of Artificial Intelligence, Renmin University of China · Weixin AI, Tencent Inc, China

Unsupervised On-Policy Self-Distillation On-Policy Self-Evolution

Jul 24, 2026cs.LG

MEMENTO: Memory-Guided Memetic Code-as-Policy Evolution

Long-horizon embodied tasks require policies that execute many dependent actions before task success can be observed. Representing policies as executable control pro- grams (code-as-policy) enables their decision logic to be inspected and revised after rollout evaluation. Revised programs can then be executed and compared by rollout performance, framing policy improvement as execution-guided program search. Evo- lutionary methods driven by large language models (LLMs) provide a natural mecha- nism for this search by generating variants and selecting high-performing candidates. However, existing approaches primarily select among independently generated vari- ants and lack a sequential local improvement phase. We introduce MEMENTO, a memory-guided single-elite memetic framework for code-as-policy evolution. ME- MENTO first evolves a rollout evaluator that maps policy rollouts to scalar fitness and structured feedback metrics. Fitness selects accepted candidates and the next elite, while feedback metrics condition policy proposals generated by memory-guided hill-climbing, macro-mutation, and crossover. We evaluate MEMENTO on two long- horizon embodied domains: Robosuite Franka Tower-of-Hanoi manipulation and AI2- THOR household interaction. MEMENTO outperforms Eureka and REvolve, adapted as code-as-policy evolutionary baselines, in task success and generalization to held- out Robosuite object configurations and unseen AI2-THOR scenes. Ablations show that zero-shot generation and unevolved evaluators fail to solve either domain, and that removing policy-search branches reduces performance. Finally, we deploy the best-evolved Robosuite policy on a physical Franka robot, demonstrating the feasibil- ity of sim-to-real transfer of the evolved code-as-policy. Code, prompts, and videos are available at: https://github.com/sygkounas/MEMENTO.

Alkis Sygkounas, Victor Aregbede, Amy Loutfi +1

Center for Applied Autonomous Sensor Systems (AASS), Örebro University, Sweden · Örebro University, 70281 Örebro, Sweden · Linköing University, 58183 Linköping, Sweden

On-Policy Self-Evolution Robot Policies Memo

Jul 24, 2026cs.LG

Teaching LLMs to Self-Evolve: Cultivating Core Meta-Skills with Reinforcement Learning

Test-time scaling through iterative self-evolution with environment feedback, as demonstrated by AlphaEvolve, shows remarkable performance gains. We hypothesize that the success of such evolution frameworks hinges on meta-skills, such as self-reflection with environment feedback, that enable effective multi-round refinement, yet are largely neglected by traditional post-training. To bridge this gap, we present MetaEvolve, a framework designed to develop these meta-skills via a data synthesis pipeline, evolution-aware reinforcement learning (RL), and inference-time evolutionary search. Concretely, we ground MetaEvolve in coding, where program execution provides natural, continuous reward signals beyond binary correctness. Building on these signals, we synthesize evolution trajectories as training data, each containing a current program, its fitness score (combining correctness and efficiency), and a history of prior attempts, and train the model via RL with verifiable rewards derived from test case execution. By training on large-scale code data, we aim to inspire generalizable domain-agnostic meta-skills that can transfer broadly to open-ended problems where such rich training signals are scarce. Across seven coding benchmarks, MetaEvolve outperforms the strongest baseline by 10.01% absolute on in-distribution tasks and 24.12% on out-of-distribution tasks. On open-ended algorithm optimization problems entirely outside the training domain, it further achieves a 46.9% relative improvement. These results demonstrate that explicitly cultivating self-evolution meta-skills offers a principled path toward more capable and autonomously self-evolving AI.

Shujin Wu, Cheng Qian, Xiusi Chen +1

University of Illinois Urbana-Champaign

On-Policy Self-Evolution Self-Evolution Skill Evolution Alphaevolve

Jul 15, 2026cs.AI

Reward-Free Evolving Agents via Pairwise Validator

A self-evolving agentic loop repeatedly proposes a tweaked version of an agent (its prompt template or program) and accepts or rejects the change based on a per-iteration quality signal. Designing that signal is often the costly part of the project: a reliable scalar reward requires domain expertise and labeled examples that are themselves as expensive to assemble as the agent's underlying task. We propose replacing the scalar at the accept/reject gate with a pairwise validator: a frozen LLM that, given the parent and child candidate, returns a binary verdict on which is better. Pairwise judgment is generally easier and more stable than absolute scoring, due to its contrastive nature, which mitigates the need for strict scale calibration. The validator also requires no training of its own. We integrate the validator into three published self-evolving engines (GEPA, ADRS, ShinkaEvolve) and report two flavors: Adaptive Focus, which retains the engine's existing val-set parent selection, and Soft Elo, which lets the validator's verdicts drive parent selection so that val-set rewards drop as well. Across multiple agents and two artifact substrates (prompt and code), our method matches or exceeds the full-reward baseline on the majority of settings we evaluate, and the pattern survives a cross-family validator swap. The pairwise gate is thus a drop-in replacement for per-step reward design at competitive task accuracy without the labeling cost.

Minghao Liu, Yu Wang, Jiayun Wang +1

Accenture

On-Policy Self-Evolution Self-Evolving Agents Agentic Evaluations Validation

Jul 14, 2026cs.AI

Critic Experience Bank: Self-Evolving Step-Level Confidence Estimation for LLM Agents

LLM agents act in external environments where each action changes the state that later decisions condition on, and where a single wrong step can waste interaction budget or trigger irreversible side effects long before the final failure is observed. Reliable deployment therefore requires \emph{step-level confidence estimation}: a calibrated probability that each proposed action is productive, available \emph{before} the action is executed. Existing LLM confidence estimators are designed to score a response from the given prompt, but agent confidence also depends on execution consequences: whether similar actions in similar situations actually advanced the task after the environment responded. We introduce the \method (\methodshort), a self-evolving critic framework in which an LLM critic accumulates evidence from its own past judgments and their observed consequences. After each trajectory, a hindsight LLM that sees the full execution feedback votes on whether each step was productive. The resulting pseudo-labels populate a memory bank from which related productive and unproductive experiences are retrieved into the critic's prompt whenever a similar step recurs. \methodshort requires no training and uses no ground truth step labels. Across three agent benchmarks and three critic backbones, \methodshort attains the best calibration (ECE and Brier) and ranking (AUC) in every dataset--critic combination, reducing ECE by up to

54\%

relative to the strongest training-free baseline.

Yaopei Zeng, Congchao Wang, JianHang Chen +3

Pennsylvania State University · Virginia Tech · Purdue University +1

Large Language Model Agents Confidence Estimation On-Policy Self-Evolution Confidence Calibration Agentic Benchmarks Critic+1

Jul 2, 2026cs.AI

EvoPolicyGym: Evaluating Autonomous Policy Evolution in Interactive Environments

Autonomous agents are increasingly expected to improve executable policies through feedback, yet existing evaluations often collapse this process into a final score or confound it with open-ended software-engineering progress. We introduce Autonomous Policy Evolution, a controlled evaluation setting in which a harness-model agent repeatedly edits an executable policy system under a fixed interaction budget. We instantiate this setting in EvoPolicyGym, a benchmark built from compact interactive RL environments that evaluates how agents iteratively improve explored policies. On the EvoPolicyGym suite, GPT-5.5 achieves the strongest aggregate rank score and top-two performance on all 16 environments. Beyond leaderboard results, EvoPolicyGym also provides trajectory-level diagnostics that distinguish how agents allocate budget, convert feedback into parametric tuning. These analyses show that strong autonomous policy evolution depends not only on isolated task wins, but on discovering task-appropriate mechanisms and refining policies under bounded feedback.

Zhilin Wang, Han Song, Runzhe Zhan +13

University of Science and Technology of China · The Chinese University of Hong Kong · University of Macau +5

On-Policy Self-Evolution Skillgym Evolutionary Optimization Methods Agent Harness

Jun 29, 2026cs.LG

DRIFT: Difficulty Routing Self-DIstillation with Rhythm-Gated Exploration and Success BuFfer Training

Enabling large language models to achieve stable self-improvement without external expert supervision remains a central challenge in complex reasoning tasks. Existing self-distillation and reinforcement learning methods lack explicit mechanisms for tracking problem-level learning progress and adapting optimization strategies accordingly. Consequently, training may over-optimize easy problems, receive weak supervision from hard problems, and fail to sufficiently explore borderline cases. To resolve these issues, we propose DRIFT, an online self-evolution policy optimization framework for large language models. DRIFT regulates the model's self-improvement process through the joint use of Difficulty Routing and Rhythm Gating. The former identifies the model's learning state at the problem level and dynamically allocates self-distillation and reinforcement learning signals, while the latter refines policy updates at the token level, concentrating exploration on critical reasoning positions. By further incorporating a success buffer and a two-stage curriculum learning strategy, DRIFT preserves high-quality historical experience while progressively guiding the model from reliable behavior acquisition toward stable policy evolution. Evaluated across five benchmarks and three model scales, DRIFT surpasses the peak performance of both GRPO and SDPO across all evaluated metrics. On the average score over the five benchmarks, DRIFT achieves 79.5

\%

, outperforming GRPO by 9.5

\%

and SDPO by 7.5

\%

, establishing a new state-of-the-art result. Notably, on ToolUse, DRIFT reaches an accuracy of 79.2

\%

, improving over GRPO by 13.5

\%

and SDPO by 10.7

\%

, setting a new state-of-the-art and substantially outperforming all concurrent methods.

Haisen Luo, Yiwei Liu, Haoning Wang +13

3Beike Language and Intelligence · 2ENS Paris-Saclay · 1Tsinghua University

Unsupervised On-Policy Self-Distillation Large Language Model Reinforcement Learning On-Policy Self-Evolution Recursive Self-Improvement

Jun 24, 2026cs.LG

EVOM: Agentic Meta-Evolution of Actor-Critic Architectures for Reinforcement Learning

In actor-critic reinforcement learning, network architectures are typically manually designed. Automating this design is challenging because each candidate must be trained before evaluation, and the design space is open-ended. To address these challenges, we introduce EVOM, an agentic meta-evolution framework for discovering high-performance actor-critic architectures. We frame architecture search as a bi-level optimization: an inner loop trains weights via the low-fidelity proximal policy optimization (PPO), while an outer loop drives meta-evolution by iteratively refining architecture programs. Crucially, this outer loop is powered by an LLM-based design agent that operates purely as an architecture designer, completely decoupled from policy execution and environment control. Experiments reveal that EVOM outperforms the manually designed baseline, an LLM-guided random search, and the state-of-the-art LLM-guided programmatic policy search method MLES, delivering superior performance on Ant-v4 and HalfCheetah-v4. Ablation studies validate that both the meta-evolution loop and the LLM Design Agent are indispensable for final performance.

Boyun Zhang, Chao Wang, Kai Wu

Xidian University

On-Policy Self-Evolution Soft Actor-Critic Self-Evolving Agents Proximal Policy Optimization Large Language Model Agents

Jun 22, 2026cs.LG

EvoRubrics: Dynamic Rubrics as Rewards via Adversarial Co-Evolution for LLM Reinforcement Learning

Rubric-based rewards offer interpretable and fine-grained optimization signals for reinforcement learning in open-ended tasks where verifiable answers are unavailable. However, pre-constructed rubrics remain static throughout training, creating a fundamental mismatch with the evolving policy: fixed criteria gradually lose discriminative power as the model improves, leading to reward saturation and potential hacking. Recent dynamic rubric methods partially address this but rely on external frontier models or ground-truth answers, and update rubrics only at coarse granularity. We propose EvoRubrics, a co-evolutionary RL framework where a Policy LLM and a Rubric Generator jointly improve through adversarial interaction within each training step. As the policy improves under the rubric generator's guidance, the rubric generator adapts its criteria to remain discriminative and informative, enabling evaluation to track the policy in real time and naturally inducing an automatic curriculum. Experiments show that EvoRubrics consistently outperforms static and dynamic rubric baselines across benchmarks. The learned Rubric Generator further generalizes as a transferable reward model. Notably, even a fully self-supervised variant without any external supervision achieves meaningful gains, suggesting that co-evolution between generation and evaluation alone can provide sufficiently rich learning signals. Our code is publicly available at https://anonymous.4open.science/r/EvoRubrics-2155/.

Hongxin Ding, Baixiang Huang, Yue Fang +6

National Engineering Research Center of Software Engineering, Peking University, China · School of Computer Science, Peking University, Beijing, China · Key Laboratory of High Confidence Software Technologies, Ministry of Education +2

On-Policy Self-Evolution Large Language Model Reinforcement Learning Co-Evolution Adversarial Robustness

Jun 15, 2026cs.CL

REFLEX: Reflective Evolution from LLM Experience

Large multimodal language models (LLMs) have emerged as powerful tools for guiding evolutionary search toward interpretable programmatic policies. However, existing frameworks rely on a monolithic model call to simultaneously interpret visual behavioral evidence and synthesize corrective code. This diagnosis-repair entanglement creates an opaque feedback loop, obscuring the rationale behind mutations and preventing the retention of algorithmic insights across independent runs. To achieve auditable and efficient policy search, we argue that visual diagnosis must be structurally decoupled from code generation. We present REFLEX, a train-free evolutionary framework that operationalizes this decoupling. In REFLEX, a vision-enabled Critic first distills task-specific behavioral evidence into structured, auditable diagnoses. Subsequently, a text-optimized Actor synthesizes child policies using these diagnoses alongside a persistent, self-evolving Skill Memory of reusable code snippets. This architecture not only provides transparent mutation traces but also enables cross-run programmatic knowledge transfer. Extensive evaluations across control benchmarks (Lunar Lander, Acrobot, Pendulum) and a 36-dimensional antenna array synthesis task demonstrate exceptional sample efficiency. Notably, REFLEX solves Acrobot and Pendulum in under 10 LLM calls and reaches a best Normalized Weighted Score of 1.092 on Lunar Lander, achieving highly competitive final performance while significantly accelerating the early-stage discovery of transparent policies.

Pan Wang

University of Science and Technology of China, China.

Large Language Model Policy Optimization On-Policy Self-Evolution Multimodal Large Language Models Reflections Evolutionary Optimization Methods Visual Evidence

Jun 5, 2026cs.LG

Self-evolving LLM agents with in-distribution Optimization

Large Language Models (LLMs) have recently emerged as powerful controllers for interactive agents in complex environments, yet training them to perform reliable long-horizon decision making remains a fundamental challenge. A key difficulty lies in credit assignment: agents often receive delayed rewards only at the end of episodes. In this paper, we propose Q-Evolve, a self-evolving framework for LLM agents that unifies automatic process-reward labeling and policy learning within a principled in-distribution reinforcement learning paradigm. In each evolving iteration, our method learns an in-distribution critic from a hybrid off-policy dataset that combines expert demonstrations with agent-generated trajectories, stabilizing Bellman backups in sparse-reward settings via a weighted Implicit Q-Learning objective. The learned value function is then used to derive step-wise process rewards through advantage estimation, enabling dense and reliable supervision without environment backtracking or human annotation. Leveraging these signals, we perform behavior-proximal policy optimization that evolves the agent over the data used for process reward labeling, allowing iterative self-improvement without exacerbating distribution shift. We evaluate our method on AlfWorld, WebShop, and ScienceWorld, showing Q-Evolve outperforms strong baselines in sample efficiency, robustness, and overall task performance. Our results demonstrate that stable agent self-evolution is achievable through the co-evolution of process-level supervision and policy, both grounded within a shared in-distribution learning loop.

Yudi Zhang, Meng Fang, Zhenfang Chen +1

Eindhoven University of Technology, Netherlands · University of Liverpool, United Kingdom · MIT-IBM Watson AI Lab.

On-Policy Self-Evolution Self-Evolving Agents Unsupervised On-Policy Self-Distillation Large Language Model Agents Alphaevolve Automatic Labelling+1

Jun 3, 2026cs.CL

Rethinking Continual Experience Internalization for Self-Evolving LLM Agents

Experience internalization converts contextual experience from past interactions into reusable parametric capability, offering a promising path toward continual learning in large language models (LLMs). While prior work has predominantly focused on single-iteration transfer, we discover that under multi-iteration experience learning, existing methods suffer from a progressive capability collapse rather than compounding improvement. We systematically examine this failure through three vital dimensions of experience internalization: (1) Experience Granularity: We find that principle-level experience is more durable than instance-level experience, as it effectively abstracts transferable strategies away from trajectory-specific details. (2) Experience Injection Pattern: Our analysis reveals that step-wise injection significantly outperforms global injection by aligning experience with intermediate decision states, a property that is critical for long-horizon tool use. (3) Internalization Regime: We demonstrate that off-policy context-distillation on high-quality teacher trajectories provides a substantially more stable training signal than on-policy context-distillation, which is inherently limited by local corrections on student-induced flawed states. Together, these insights yield a simple yet robust recipe for stable and sustainable experience internalization, providing concrete guidance for engineering self-evolving and continually learning LLMs.

Jingwen Chen, Wenkai Yang, Shengda Fan +7

1Gaoling School of Artificial Intelligence, Renmin University of China · School of Software, Beihang University · 3Meituan

Internalization On-Policy Self-Evolution Large Language Model Agents

Jun 2, 2026cs.AI

EvoTrainer: Co-Evolving LLM Policies and Training Harnesses for Autonomous Agentic Reinforcement Learning

Autonomous LLM training is often framed as recipe search, which leaves the training harness largely static. This limitation sharpens in agentic RL, where shifting bottlenecks and scalar rewards mask diverse failure modes. We introduce EvoTrainer, an autonomous training framework that co-evolves LLM policies and training-side harnesses through empirical feedback: it diagnoses rollout-level evidence, revises diagnostics, backtests interventions, and accumulates reusable skills. Evaluated on mathematical reasoning, competitive-programming code generation, and repository-level software engineering, EvoTrainer matches or exceeds the human-engineered RL references under the same data, codebase, and evaluation protocol, with the largest gain on long-horizon agentic SWE. Trajectory analyses show that retained strategies diverge across domains, evolving diagnostics prevent invalid high-scoring branches from being promoted, and reusable skills shape later search. Autonomous LLM RL should move beyond recipe search toward joint evolution of policies and the training harnesses that interpret them.

Guhong Chen, Yingcheng Shi, Yongbin Li +6

1Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences · 2Tongyi Lab , Alibaba Group · 3Alibaba Group +1

On-Policy Self-Evolution Agentic Reinforcement Learning Large Language Model Policy Optimization Code Generation

May 14, 2026cs.AI

Learning to Build the Environment: Self-Evolving Reasoning RL via Verifiable Environment Synthesis

We pursue a vision for self-improving language models in which the model does not merely generate problems or traces to imitate, but constructs the environments that train it. In zero-data reasoning RL, this reframes self-improvement from a data-generation loop into an environment-construction loop, where each artifact is a reusable executable object that samples instances, computes references, and scores responses. Whether this vision sustains improvement hinges on a single property: the environments must exhibit stable solve--verify asymmetry, the model must be able to write an oracle once that it cannot reliably execute in natural language on fresh instances. This asymmetry takes two complementary forms. Some tasks are algorithmically hard to reason through but trivial as code: a dynamic program or graph traversal, compiled once, yields unboundedly many calibrated instances. Others are intrinsically hard to solve but easy to verify, like planted subset-sum or constraint satisfaction. Both create a durable gap between proposing and solving that the policy cannot close by gaming the verifier, and it is this gap that keeps reward informative as the learner improves. We instantiate this view in EvoEnv, a single-policy generator, solver method that synthesizes Python environments from ten seeds and admits them only after staged validation, semantic self-review, solver-relative difficulty calibration, and novelty checks. The strongest evidence comes from the already-strong regime: on Qwen3-4B-Thinking, fixed public-data RLVR and fixed hand-crafted environment RLVR reduce the average, while EvoEnv improves it from 72.4 to 74.8, a relative gain of 3.3%. Stable self-improvement, we suggest, depends not on producing more synthetic data, but on models learning to construct worlds whose difficulty stays structurally beyond their own reach.

Yucheng Shi, Zhenwen Liang, Kishan Panaganti +3

Tencent HY LLM

Synthetic Environments On-Policy Self-Evolution Recursive Self-Improvement Offline Reinforcement Learning Alphaevolve Synthesis+1

May 12, 2026cs.AI

On-Policy Self-Evolution via Failure Trajectories for Agentic Safety Alignment

Tool-using LLM agents fail through trajectories rather than only final responses, as they may execute unsafe tool calls, follow injected instructions, comply with harmful requests, or over-refuse benign tasks despite producing a seemingly safe answer. Existing safety-alignment signals are largely response-level or off-policy, and often incur a safety-utility trade-off: improving agent safety comes at the cost of degraded task performance. Such sparse and single-objective rewards severely limit real-world usability. To bridge this gap, we propose FATE, an on-policy self-evolving framework that transforms verifier-scored failures into repair supervision without expert demonstrations. For each failure, the same policy proposes repair candidates, which are then re-scored by verifiers and filtered across security, utility, over-refusal control, and trajectory validity. This dense trajectory-level information is then used as a supervision signal for agent self-evolution. During this process, we further introduce Pareto-Front Policy Optimization (PFPO), combining supervised warmup with Pareto-aware policy optimization to preserve safety-utility trade-offs. Experiments on AgentDojo, AgentHarm, and ATBench show that FATE improves safety across different models and scales while preserving useful behavior. Compared with strong baselines, FATE reduces attack success rate by 33.5%, harmful compliance by 82.6%, and improves external trajectory-safety diagnosis by 6.5%. These results suggest that failed trajectories can provide structured repair supervision for safer self-evolving agents.

Bo Yin, Qi Li, Xinchao Wang

National University of Singapore

On-Policy Self-Evolution Self-Evolving Agents Safety Alignment Repair

May 12, 2026cs.RO

EvoNav: Evolutionary Reward Function Design for Robot Navigation with Large Language Models

Robot navigation is a crucial task with applications to social robots in dynamic human environments. While Reinforcement Learning (RL) has shown great promise for this problem, the policy quality is highly sensitive to the specification of reward functions. Hand-crafted rewards require substantial domain expertise and embed inductive biases that are difficult to audit or adapt, limiting their effectiveness and leading to suboptimal performance. In this paper, we propose EvoNav, an evolutionary framework that automates the design of robot navigation reward functions via large language models (LLMs). To overcome prohibitively costly policy training, EvoNav evaluates each candidate proposal from the LLM via a progressive three-stage warm-up-boost procedure. EvoNav advances from analytical proxies with low-cost surrogates, such as small datasets and analytic rules, to lightweight rollouts and, finally, to full policy training, enabling computationally efficient exploration under effective feedback. Experiment results show that EvoNav produces more effective navigation policies than manually designed RL rewards and state-of-the-art reward design methods.

Zhikai Zhao, Chuanbo Hua, Federico Berto +4

1KAIST · 2Radical Numerics · 3UC Riverside +1

Robot Navigation On-Policy Self-Evolution Robot Policies Evolutionary Optimization Methods

May 11, 2026cs.AI

Evolving-RL: End-to-End Optimization of Experience-Driven Self-Evolving Capability within Agents

Experience-driven self-evolving agents aim to overcome the static nature of large language models by distilling reusable experience from past interactions, thus enabling adaptation to novel tasks at deployment time. This process places substantial demands on the foundation model's capacities for abstraction, generalization, and in-context learning. However, most existing studies focus primarily on system-level design choices, such as how experience is represented and managed, neglecting the inherent capabilities of the underlying model. While some recent works have started to optimize the experience utilization stage via reinforcement learning, they still fail to treat self-evolution as a unified process to be jointly optimized. To this end, we propose Evolving-RL, an efficient algorithmic framework that jointly improves the experience extraction and utilization capabilities required for self-evolution. Specifically, we center the learning process on experience extraction and evaluation, using the two supervisory signals derived from evaluation to optimize the extractor and solver separately and thus enable their coordinated co-evolution. Experiments on ALFWorld and Mind2Web show that Evolving-RL effectively enhances LLMs' ability to extract and reuse experience, leading to strong performance gains on out-of-distribution tasks (up to 98.7% relative improvement over the GRPO baseline on ALFWorld unseen tasks and 35.8% on Mind2Web), and these gains are fully unlocked only through the coordinated co-evolution of experience extraction and utilization. Furthermore, Evolving-RL inherently functions as an experience-augmented RL algorithm. By internalizing reusable experience patterns directly into model parameters, it achieves remarkable performance gains over standard baselines on both seen and unseen tasks, even in the absence of test-time experience accumulation.

Zhiyuan Fan, Wenwei Jin, Feng Zhang +4

1Xiaohongshu Inc. · School of Computer Science, Peking University

On-Policy Self-Evolution Self-Evolving Agents

May 11, 2026cs.AI

Teacher-Aware Evolution of Heuristic Programs from Learned Optimization Policies

LLM-based automatic heuristic design has shown promise for generating executable heuristics for combinatorial optimization, but existing methods mainly rely on delayed endpoint performance. We propose a \emph{teacher-aware evolutionary framework} that uses independently trained learned optimization policies as behavioral teachers. Instead of deploying or imitating the teacher, our method queries it on states visited by candidate heuristic programs and uses its action preferences as local feedback for evolution. The resulting search discovers static executable heuristics guided by both task performance and teacher-derived behavioral signals. Experiments on scheduling, routing, and graph optimization benchmarks show that our method improves over performance-driven LLM heuristic evolution baselines while requiring no neural inference at deployment. These results suggest that learned optimization policies can be repurposed as behavioral feedback sources for automatic heuristic discovery.

Minyu Chen, Song Qin, Ling-I Wu +2

Shenzhen Technology University · Shanghai Polytechnic University · Shanghai Jiao Tong University

Automated Heuristic Design Evolutionary Search Heuristics On-Policy Self-Evolution Evolutionary Optimization Methods Optimal Policies+2

May 11, 2026cs.LG

Metis: Learning to Jailbreak LLMs via Self-Evolving Metacognitive Policy Optimization

Red teaming is critical for uncovering vulnerabilities in Large Language Models (LLMs). While automated methods have improved scalability, existing approaches often rely on static heuristics or stochastic search, rendering them brittle against advanced safety alignment. To address this, we introduce Metis, a framework that reformulates jailbreaking as inference-time policy optimization within an adversarial Partially Observable Markov Decision Process (POMDP). Metis employs a self-evolving metacognitive loop to perform causal diagnosis of a target's defense logic and leverages structured feedback as a semantic gradient to refine its policy, offering enhanced interpretability through transparent reasoning traces. Extensive evaluations across 10 diverse models demonstrate that Metis achieves the strongest average Attack Success Rate (ASR) among compared methods at 89.2%, maintaining high efficacy on resilient frontier models (e.g., 76.0% on O1 and 78.0% on GPT-5-chat) where traditional baselines exhibit substantial performance degradation. By replacing redundant exploration with directed optimization, Metis reduces token costs by an average of 8.2x and up to 11.4x. Our analysis reveals that current defenses remain vulnerable to internally-steered, closed-loop reasoning trajectories under the tested settings, highlighting a critical need for next-generation defenses capable of reasoning about safety dynamically during inference.

Huilin Zhou, Jian Zhao, Yilu Zhong +7

University of Science and Technology of China · Institute of Artificial Intelligence (TeleAI), China Telecom

Large Language Model Jailbreaks On-Policy Self-Evolution Metacognition Inference-Time Steering Self-Evolution

May 9, 2026cs.CL

Breaking the Impasse: Dual-Scale Evolutionary Policy Training for Social Language Agents

While Reinforcement Learning with Verifiable Rewards (RLVR) has proven effective for closed-ended tasks, extending it to open-ended social language games via self-play reveals a critical issue: evolution impasse. Due to the vast strategy space, language agents frequently converge to homogenized behaviors, leading to deterministic match outcomes that eliminate the gradient signals necessary for policy evolution. To tackle this issue, we propose Dual-scale Evolutionary Policy Training (DEPT) for social language games. DEPT introduces a time-scaled evolutionary perception mechanism that detects impasse by quantifying dual-scale value baseline divergence alongside match entropy. Upon perceiving the collapse, it then activates asymmetric advantage reshaping to dynamically modulate the optimization landscape for intervention. Thus, our method effectively restores gradient signals and enforces sustained strategic exploration. Extensive experiments on multiple social language games demonstrate that DEPT outperforms strong baselines, avoiding policy degeneration and driving the continuous evolution of social language agents.

Minzheng Wang, Run Luo, Yanbo Wang +6

MAIS, Institute of Automation, Chinese Academy of Sciences · School of Artificial Intelligence, University of Chinese Academy of Sciences · National University of Singapore +1

On-Policy Self-Evolution

May 9, 2026cs.AI

RewardHarness: Self-Evolving Agentic Post-Training

Evaluating instruction-guided image edits requires rewards that reflect subtle human preferences, yet current reward models typically depend on large-scale preference annotation and additional model training. This creates a data-efficiency gap: humans can often infer the target evaluation criteria from only a few examples, while models are usually trained on hundreds of thousands of comparisons. We present RewardHarness, a self-evolving agentic reward framework that reframes reward modeling as context evolution rather than weight optimization. Instead of learning from large-scale annotations, RewardHarness aligns with human preferences by iteratively evolving a library of tools and skills from as few as 100 preference demonstrations. Given a source image, candidate edited images, and an editing instruction, an Orchestrator selects the most relevant subset of tools and skills from the maintained library, and a frozen Sub-Agent uses them to construct a reasoning chain that produces a preference judgment. By comparing predicted judgments with ground-truth preferences and analyzing successes and failures in the reasoning process, the Orchestrator automatically refines its library of tools and skills without additional human annotation. Using only 0.05% of the EditReward preference data, RewardHarness achieves 47.4% average accuracy on image-editing evaluation benchmarks, surpassing GPT-5 by 5.3 points. When used as a reward signal for GRPO fine-tuning, RL-tuned models achieve 3.52 on ImgEdit-Bench. Project page: https://rewardharness.com.

Yuxuan Zhang, Penghui Du, Bo Li +11

University of British Columbia · Kolors Team, Kuaishou Technology · University of Waterloo +4

Progress Reward Modeling Multi-Image Editing Benchmarks On-Policy Self-Evolution Preference Datasets Post-Training

May 8, 2026cs.CL

SEIF: Self-Evolving Reinforcement Learning for Instruction Following

Instruction following is a fundamental capability of large language models (LLMs), yet continuously improving this capability remains challenging. Existing methods typically rely either on costly external supervision from humans or strong teacher models, or on self-play training with static-difficulty instructions that cannot evolve as the model's capabilities improve. To address these limitations, we propose SEIF (Self-Evolving Reinforcement Learning for Instruction Following), a self-evolving framework for enhancing the instruction-following ability of LLMs. SEIF forms a closed self-evolution loop that improves the model's instruction-following ability, where instruction difficulty evolution and model capability evolution reinforce each other. SEIF consists of four roles: an Instructor that generates increasingly challenging instructions, a Filter that removes conflicting or invalid instructions to ensure data quality, a Follower that learns to follow evolved instructions, and a Judger that provides reward signals for reinforcement learning. The Instructor and Follower are alternately trained and co-evolve throughout the process. Experiments across multiple model scales and architectures show that SEIF consistently improves instruction-following performance, suggesting strong generality. Further analyses reveal the sources of improvement and identify an effective training strategy for self-evolution on open-ended tasks: sufficient early-stage training to build a solid foundation, followed by moderate late-stage training to mitigate overfitting and achieve better final performance. The code and data are publicly available at https://github.com/Rainier-rq1/SEIF.

Qingyu Ren, Qianyu He, Jiajie Zhu +7

Shanghai Key Laboratory of Data Science, College of Computer Science and Artificial Intelligence, Fudan University · School of Data Science, Fudan University · Ant Group

On-Policy Self-Evolution Instruction Instruction-Tuned Models

Apr 20, 2026cs.AI

Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence

Large language models are increasingly expected to serve as general-purpose agents that interact with external, stateful tool environments. The Model Context Protocol (MCP) and broader agent skills offer a unified interface for connecting agents with scalable real-world services, but training robust agents remains limited by the lack of realistic environments and principled mechanisms for life-long learning. In this paper, we present \textbf{Agent-World}, a self-evolving training arena for advancing general agent intelligence through scalable environments. Agent-World has two main components: (1) Agentic Environment-Task Discovery, which autonomously explores topic-aligned databases and executable tool ecosystems from thousands of real-world environment themes and synthesizes verifiable tasks with controllable difficulty; and (2) Continuous Self-Evolving Agent Training, which combines multi-environment reinforcement learning with a self-evolving agent arena that automatically identifies capability gaps through dynamic task synthesis and drives targeted learning, enabling the co-evolution of agent policies and environments. Across 23 challenging agent benchmarks, Agent-World-8B and 14B consistently outperforms strong proprietary models and environment scaling baselines. Further analyses reveal scaling trends in relation to environment diversity and self-evolution rounds, offering insights for building general agent intelligence.

Guanting Dong, Junting Lu, Junjie Huang +17

Renmin University of China · ByteDance Seed

Synthetic Environments Self-Evolving Agents On-Policy Self-Evolution Open-World Synthesis

Apr 20, 2026cs.AI

Training LLM Agents for Spontaneous, Reward-Free Self-Evolution via World Knowledge Exploration

Most agents today ``self-evolve'' by following rewards and rules defined by humans. However, this process remains fundamentally dependent on external supervision; without human guidance, the evolution stops. In this work, we train agents to possess an intrinsic meta-evolution capability to spontaneously learn about unseen environments prior to task execution. To instill this ability, we design an outcome-based reward mechanism that measures how much an agent's self-generated world knowledge improves its success rate on downstream tasks. This reward signal is used exclusively during the training phase to teach the model how to explore and summarize effectively. At inference time, the agent requires no external rewards or human instructions. It spontaneously performs native self-evolution to adapt to unknown environments using its internal parameters. When applied to Qwen3-30B and Seed-OSS-36B, this shift to native evolution yields a 20% performance increase on WebVoyager and WebWalker. Most strikingly, the generated world knowledge even enables a compact 14B Qwen3 model to outperform the unassisted Gemini-2.5-Flash, establishing a new paradigm for truly evolving agents.

Qifan Zhang, Dongyang Ma, Tianqing Fang +5

1Tencent · 2The Hong Kong University of Science and Technology (Guangzhou)

On-Policy Self-Evolution Self-Evolution Large Language Model Agents Exploration

Apr 17, 2026cs.CL

CoEvolve: Training LLM Agents via Agent-Data Mutual Evolution

Reinforcement learning for LLM agents is typically conducted on a static data distribution, which fails to adapt to the agent's evolving behavior and leads to poor coverage of complex environment interactions. To address these challenges, we propose CoEvolve, an agent-data mutual evolution framework that enables LLM agents to improve through closed-loop, interaction-driven training. Specifically, CoEvolve extracts feedback signals such as forgetting and uncertainty from rollout trajectories to identify failure-prone interaction patterns, and utilizes them to guide LLM-based task synthesis. The synthesized tasks are validated through environment interaction and utilized to update the data distribution, enabling joint adaptation of the agent and its data. Extensive experiments on AppWorld and BFCL across Qwen2.5-7B, Qwen3-4B, and Qwen3-30B-A3B demonstrate consistent and significant improvements over strong base models, yielding absolute gains of 19.43%, 15.58%, and 18.14%, respectively.

Shidong Yang, Ziyu Ma, Tongwen Huang +3

AMAP, Alibaba Group

On-Policy Self-Evolution Large Language Model Agents Co-Evolution Synthetic Task Joint Search

Jan 7, 2026cs.LG

Survival Dynamics of Neural and Programmatic Policies in Evolutionary Reinforcement Learning

In evolutionary reinforcement learning tasks (ERL), agent policies are often encoded as small artificial neural networks (NERL). Such representations lack explicit modular structure, limiting behavioral interpretation. We investigate whether programmatic policies (PERL), implemented as soft, differentiable decision lists (SDDL), can match the performance of NERL. To support reproducible evaluation, we provide the first fully specified and open-source reimplementation of the classic 1992 Artificial Life (ALife) ERL testbed. We conduct a rigorous survival analysis across 4000 independent trials utilizing Kaplan-Meier curves and Restricted Mean Survival Time (RMST) metrics absent in the original study. We find a statistically significant difference in survival probability between PERL and NERL. PERL agents survive on average 201.69 steps longer than NERL agents. Moreover, SDDL agents using learning alone (no evolution) survive on average 73.67 steps longer than neural agents using both learning and evaluation. These results demonstrate that programmatic policies can exceed the survival performance of neural policies in ALife.

Anton Roupassov-Ruiz, Yiyang Zuo

Department of Computing Science University of Alberta Edmonton, Alberta, T6G 2E8, Canada

Neural Policies On-Policy Self-Evolution Survival Analysis

Jul 15, 2025cs.AI

Illuminating the Three Dogmas of Reinforcement Learning under Evolutionary Light

Artificial learning systems are graduating from passive learners to increasingly autonomous agents, lending pragmatic urgency to the question of what constitutes agency. Reinforcement learning (RL) offers arguably the most explicit formulation of agent-environment interaction, built on three core tenets: the environment as a Markov decision process, learning as policy optimization, and the agent as a maximizer of scalar reward. Recent work has called to revise these tenets: reconceptualizing learning as adaptation rather than optimization, broadening goals beyond scalar reward, and noting the absence of a formal theory of the agent in a formalism that so heavily emphasizes the environment. We argue that the artificial life community is uniquely positioned to illuminate this critique and concretize an alternative. We draw on open-ended novelty search as a complementary model of adaptation and goal-directed behavior beyond reward optimization, and ground such evolutionary dynamics in thermodynamic theories of origin-of-life and agency, toward a more biologically faithful and formally grounded account of what it is to be an adaptive agent.

Mani Hamidi, Terrence W. Deacon

Department of Computer Science, University of Tübingen, Germany · Department of Anthropology, University of California, Berkeley, USA

On-Policy Self-Evolution Agentic Control Evolutionary Optimization Methods Goal-Directed Behavior