Language Model Self-Improvement

Follow this topic Ask about this topic

The field studies how language models can improve their reasoning, instruction following, coding, and other skills by learning from their own attempts and interactions. Common approaches include self-training and self-distillation on generated responses, reinforcement learning and challenger–solver self-play, and test-time adaptation through memory, skills, or evolving prompts. Feedback ranges from exact checks such as code execution and formal proof to model-generated critiques, preferences, and confidence signals, with evaluation spanning math and coding benchmarks as well as agent and multimodal tasks. The central tension is whether these internal signals are reliable enough to produce transferable gains: noisy feedback, narrow curricula, and repeated selection can cause regressions or collapse, while oracle-supervised performance often remains difficult to match.

Latest papers

174 papers26 papers in the last four weeks, up 30% on the four weeks before. 0.2% of all new papers.

Mar 19, 2026

Learning to Self-Evolve

We introduce Learning to Self-Evolve (LSE), a reinforcement learning framework that trains large language models (LLMs) to improve their own contexts at test time. We situate LSE in the setting of test-time self-evolution, where a model iteratively refines its context from feedback on seen problems to perform better on new ones. Existing approaches rely entirely on the inherent reasoning ability of the model and never explicitly train it for this task. LSE reduces the multi-step evolution problem to a single-step RL objective, where each context edit is rewarded by the improvement in downstream performance. We pair this objective with a tree-guided evolution loop. On Text-to-SQL generation (BIRD) and general question answering (MMLU-Redux), a 4B-parameter model trained with LSE outperforms self-evolving policies powered by GPT-5 and Claude Sonnet 4.5, as well as prompt optimization methods including GEPA and TextGrad, and transfers to guide other models without additional training. Our results highlight the effectiveness of treating self-evolution as a learnable skill.

Xiaoyin Chen, Canwen Xu, Yite Wang +3

Mila – Quebec AI Institute · University of Montreal · Snowflake

Language Model Self-Improvement RL for Language Models

Mar 18, 2026

Continually self-improving AI

Modern language model-based AI systems are remarkably powerful, yet their capabilities remain fundamentally capped by their human creators in three key ways. First, although a model's weights can be updated via fine-tuning, acquiring new knowledge from small, specialized corpora after pretraining remains highly data-inefficient. Second, the training of these systems relies heavily on finite, human-generated data from across history. Third, the pipelines used to train AI models are confined by the algorithms that human researchers can discover and explore. This thesis takes a small step toward overcoming these inherent limitations, presenting three chapters aimed at breaking these dependencies to create continually self-improving AI. First, to overcome this data-efficiency barrier in knowledge acquisition, we propose a synthetic data approach that diversifies and amplifies small corpora into rich knowledge representations, enabling a model to effectively update its parameters from limited source material. Second, to reduce reliance on human data, we show that given a fixed amount of such data, the model can self-generate synthetic data to bootstrap its fundamental pretraining capabilities without distillation from any off-the-shelf, instruction-tuned LM. Finally, to transcend human-engineered training paradigms, we demonstrate that by scaling search during test time over the space of algorithms, AI can search over a larger space of learning algorithm configurations than human researchers can explore manually.

Zitong Yang

Department of Statistics, Stanford University

Language Model Self-Improvement Self-Training for Language Models Knowledge Augmentation for Language Models Synthetic Data Generation for Language Models Synthetic Data Pretraining

Mar 17, 2026

Online Experiential Learning for Language Models

The prevailing paradigm for improving large language models relies on offline training with human annotations or simulated environments, leaving the rich experience accumulated during real-world deployment entirely unexploited. We propose Online Experiential Learning (OEL), a framework that enables language models to continuously improve from their own deployment experience. OEL operates in two stages: first, transferable experiential knowledge is extracted and accumulated from interaction trajectories collected on the user side; second, this knowledge is consolidated into model parameters via on-policy context distillation, requiring no access to the user-side environment. The two stages are iterated to form an online learning loop, where the improved model collects higher-quality trajectories that yield richer experiential knowledge for subsequent rounds. We evaluate OEL on text-based game environments across multiple model scales and both thinking and non-thinking variants. OEL achieves consistent improvements over successive iterations, enhancing both task accuracy and token efficiency while preserving out-of-distribution performance. Our analysis further shows that extracted experiential knowledge is significantly more effective than raw trajectories, and that on-policy consistency between the knowledge source and the policy model is critical for effective learning.

Tianzhu Ye, Li Dong, Qingxiu Dong +3

Microsoft Research

Continual Learning for LLMs Language Model Self-Improvement

Mar 16, 2026

GASP: Guided Asymmetric Self-Play For Coding LLMs

Asymmetric self-play has emerged as a promising paradigm for post-training large language models, where a teacher continually generates questions for a student to solve at the edge of the student's learnability. Although these methods promise open-ended data generation bootstrapped from no human data, they suffer from one major problem: not all problems that are hard to solve are interesting or informative to improve the overall capabilities of the model. Current asymmetric self-play methods are goal-agnostic with no real grounding. We propose Guided Asymmetric Self-Play (GASP), where grounding is provided by real-data goalpost questions that are identified to pose a hard exploration challenge to the model. During self-play, the teacher first generates an easier variant of a hard question, and then a harder variant of that easier question, with the goal of gradually closing the gap to the goalpost throughout training. Doing so, we improve pass@20 on LiveCodeBench (LCB) by 2.5% over unguided asymmetric self-play, and through the curriculum constructed by the teacher, we manage to solve hard goalpost questions that remain out of reach for all baselines.

Swadesh Jana, Cansu Sancaktar, Tomáš Daniš +3

University of Tübingen · Max Planck Institute for Intelligent Systems · ELLIS Institute Tübingen +1

Language Model Self-Improvement LLM Code Generation Curriculum Learning

Mar 10, 2026

Maximizing Mutual Information Between Prompt and Response Improves LLM Performance With No Additional Data

While post-training has successfully improved large language models (LLMs) across a variety of domains, these gains heavily rely on human-labeled data or external verifiers. Existing data has already been exploited, and new data is expensive to collect. Moreover, true intelligence goes far beyond verifiable tasks. Therefore, we need self-improvement frameworks that are less dependent on external signals and more broadly applicable to both verifiable and non-verifiable domains. We propose Mutual Information Preference Optimization (MIPO), a contrastive data augmentation method that constructs preference pairs by generating a positive response conditioning on the correct prompt, and a negative response by conditioning on a random, unrelated prompt. We show that using Direct Preference Optimization to learn from this paired data maximizes pointwise mutual information under the base LLM between prompts and model responses. Experiments with with 1-7B parameter Llama and Qwen instruct models show that MIPO achieves 3-16% gains (and 51% increase for Qwen2.5-1.5B-Instruct) on personalization compared to prompting baselines. Surprisingly, MIPO can also be useful in verifiable domains, such as math and multiple-choice question answering, yielding 1-20% gains without any additional data or external supervision. These results suggest a promising direction for self-improvement using intrinsic signals derived from contrastive data pairs.

Hyunji Nam, Haoran Li, Natasha Jaques

Stanford University · Character AI · University of Washington

LLM Personalization Mutual Information Maximization LLM Post-Training Language Model Self-Improvement LLM Training

Mar 6, 2026

A Dynamic Self-Evolving Extraction System

The extraction of structured information from raw text is a fundamental component of many NLP applications, including document retrieval, ranking, and relevance estimation. High-quality extractions often require domain-specific accuracy, up-to-date understanding of specialized taxonomies, and the ability to incorporate emerging jargon and rare outliers. In many domains--such as medical, legal, and HR--the extraction model must also adapt to shifting terminology and benefit from explicit reasoning over structured knowledge. We propose DySECT, a Dynamic Self-Evolving Extraction and Curation Toolkit, which continually improves as it is used. The system incrementally populates a versatile, self-expanding knowledge base (KB) with triples extracted by the LLM. The KB further enriches itself through the integration of probabilistic knowledge and graph-based reasoning, gradually accumulating domain concepts and relationships. The enriched KB then feeds back into the LLM extractor via prompt tuning, sampling of relevant few-shot examples, or fine-tuning using KB-derived synthetic data. As a result, the system forms a symbiotic closed-loop cycle in which extraction continuously improves knowledge, and knowledge continuously improves extraction.

Moin Amin-Naseri, Hannah Kim, Estevam Hruschka

Megagon Labs

Document Information Extraction LLM Information Extraction Knowledge Augmentation for Language Models Language Model Self-Improvement Continual Learning for LLMs

Mar 3, 2026

Preventing Curriculum Collapse in Self-Evolving Reasoning Systems

Self-evolving reasoning frameworks let LLMs improve their reasoning capabilities by iteratively generating and solving problems without external supervision, using verifiable rewards. Ideally, such systems are expected to explore a diverse problem space and propose new challenges of high learning value. While prior work has largely focused on solver-side optimisation and verification, recent evidence suggests that self-evolving systems can exhibit diversity collapse in posing new problems after just a few iterations, even when surface-level variation is preserved. We introduce Prism, a question-centric self-evolution method that directly tackles this collapse. Prism defines a persistent diversity signal over an embedding-induced semantic partition of mathematical problems and uses it to encourage balanced exploration of underrepresented regions across iterations. This coverage signal is combined with a Zone-of-Proximal-Development (ZPD) gate to preserve edge-of-solvability difficulty. Evaluated on seven widely used mathematical reasoning benchmarks against five self-evolving baselines, Prism achieves the highest accuracy on six out of seven tasks, achieving gains of +3.98 absolute points over R-Zero on AMC and +3.68 on Minerva Math. Prism also generates semantically diverse and challenging questions across iterations, resulting in the construction of the Prism-Math dataset comprising 100k mathematical questions. These results demonstrate that cross-iteration semantic coverage is a high-leverage and under-explored axis for building more capable self-evolving reasoners. We release the code, dataset, and models to facilitate further research.

Vaibhav Mishra

Mathematical Problem Generation Language Model Self-Improvement Self-Evolving Agents LLM Mathematical Reasoning

Mar 3, 2026

Test-Time Meta-Adaptation with Self-Synthesis

As strong general reasoners, large language models (LLMs) encounter diverse domains and tasks, where the ability to adapt and self-improve at test time is valuable. We introduce MASS, a meta-learning framework that enables LLMs to self-adapt by generating problem-specific synthetic training data and performing targeted self-updates optimized for downstream performance at inference time. We train this behavior end-to-end via bilevel optimization: an inner loop adapts on self-generated examples while an outer loop meta-learns data-attribution signals and rewards post-update task performance. The synthetic data is optimized with scalable meta-gradients, backpropagating the downstream loss through the inner updates to reward useful generations. Experiments on mathematical reasoning show that MASS learns to synthesize per-instance curricula that yield effective, data-efficient test-time adaptation.

Zeyneb N. Kaya, Nick Rui

Stanford University

Inference-Time Adaptation of Language Models Test-Time Adaptation Language Model Self-Improvement Meta-Learning Synthetic Data Generation for Language Models

Mar 2, 2026

CharacterFlywheel: Scaling Iterative Improvement of Engaging and Steerable LLMs in Production

This report presents CharacterFlywheel, an iterative flywheel process for improving large language models (LLMs) in production social chat applications across Instagram, WhatsApp, and Messenger. Starting from LLaMA 3.1, we refined models across 15 generations using data from both internal and external real-user traffic. Through continuous deployments from July 2024 to April 2025, we conducted controlled 7-day A/B tests showing consistent engagement improvements: 7 of 8 newly deployed models demonstrated positive lift over the baseline, with the strongest performers achieving up to 8.8% improvement in engagement breadth and 19.4% in engagement depth. We also observed substantial gains in steerability, with instruction following increasing from 59.2% to 84.8% and instruction violations decreasing from 26.6% to 5.8%. We detail the CharacterFlywheel process which integrates data curation, reward modeling to estimate and interpolate the landscape of engagement metrics, supervised fine-tuning (SFT), reinforcement learning (RL), and both offline and online evaluation to ensure reliable progress at each optimization step. We also discuss our methods for overfitting prevention and navigating production dynamics at scale. These contributions advance the scientific rigor and understanding of LLMs in social applications serving millions of users.

Yixin Nie, Lin Guan, Zhongyao Ma +19

Meta Superintelligence Labs · OpenAI · xAI +1

LLM Post-Training Conversational Agents Language Model Self-Improvement LLM Evaluation

Mar 2, 2026

Provable and Practical In-Context Policy Optimization for Self-Improvement

We study test-time scaling, where a model improves its answer through multi-round self-reflection at inference. We introduce In-Context Policy Optimization (ICPO), in which an agent optimizes its response in context using self-assessed or externally observed rewards without modifying its parameters. To explain this ICPO process, we theoretically show that with sufficient pretraining under a novel Fisher-weighted logit-matching objective, a single-layer linear self-attention model can provably imitate policy-optimization algorithm for linear bandits. Building on this theory, we propose Minimum-Entropy ICPO (ME-ICPO), a practical algorithm that iteratively uses its response and self-assessed reward to refine its response in-context at inference time. By selecting the responses and their rewards with minimum entropy, ME-ICPO ensures the robustness of the self-assessed rewards via majority voting. Across standard mathematical reasoning tasks, ME-ICPO attains competitive, top-tier performance while keeping inference costs affordable compared with other inference-time algorithms. Overall, ICPO provides a principled understanding of self-reflection in LLMs and yields practical benefits for test-time scaling for mathematical reasoning.

Tianrun Yu, Yuxiao Yang, Zhaoyang Wang +6

Brigham Young University · University of North Carolina at Chapel Hill · Microsoft

Language Model Self-Improvement Test-Time Scaling In-Context RL LLM Reasoning RL for Language Model Reasoning

Feb 25, 2026

Duel-Evolve: Reward-Free Test-Time Scaling via LLM Self-Preferences

Many applications seek to optimize LLM outputs at test time by iteratively proposing, scoring, and refining candidates over a discrete output space. Existing methods use a calibrated scalar evaluator for the target objective to guide search, but for many tasks such scores are unavailable, too sparse, or unreliable. Pairwise comparisons, by contrast, are often easier to elicit, still provide useful signal on improvement directions, and can be obtained from the LLM itself without external supervision. Building on this observation, we introduce Duel-Evolve, an evolutionary optimization algorithm that replaces external scalar rewards with pairwise preferences elicited from the same LLM used to generate candidates. Duel-Evolve aggregates these noisy candidate comparisons via a Bayesian Bradley-Terry model, yielding uncertainty-aware estimates of candidate quality. These quality estimates guide allocation of the comparison budget toward plausible optima using Double Thompson Sampling, as well as selection of high-quality parents to generate improved candidates. We evaluate Duel-Evolve on MathBench, where it achieves 20 percentage points higher accuracy over existing methods and baselines, and on LiveCodeBench, where it improves over comparable iterative methods by over 12 percentage points. Notably, the method requires no reward model, no ground-truth labels during search, and no hand-crafted scoring function. Results show that pairwise self-preferences provide strong optimization signal for test-time improvement over large, discrete output spaces.

Sweta Karlekar, Carolina Zheng, Magnus Saebo +5

Columbia University · Los Alamos National Laboratory

Large Language Model-Guided Optimization Test-Time Scaling Language Model Self-Improvement Evolutionary Optimization Language Model Self-Assessment

Feb 17, 2026

Improving Interactive In-Context Learning from Natural Language Feedback

Adapting one's thought process based on corrective feedback is an essential ability in human learning, particularly in collaborative settings. In contrast, the current large language model training paradigm relies heavily on modeling vast, static corpora. While effective for knowledge acquisition, it overlooks the interactive feedback loops essential for models to adapt dynamically to their context. In this work, we propose a framework that treats this interactive in-context learning ability not as an emergent property, but as a distinct, trainable skill. We introduce a scalable method that transforms single-turn verifiable tasks into multi-turn didactic interactions driven by information asymmetry. We first show that current flagship models struggle to integrate corrective feedback on hard reasoning tasks. We then demonstrate that models trained with our approach dramatically improve the ability to interactively learn from language feedback. More specifically, the multi-turn performance of a smaller model nearly reaches that of a model an order of magnitude larger. We also observe robust out-of-distribution generalization: interactive training on math problems transfers to diverse domains like coding, puzzles and maze navigation. Our qualitative analysis suggests that this improvement is due to an enhanced in-context plasticity. Finally, we show that this paradigm offers a unified path to self-improvement. By training the model to predict the teacher's critiques, effectively modeling the feedback environment, we convert this external signal into an internal capability, allowing the model to self-correct even without a teacher.

Martin Klissarov, Jonathan Cook, Diego Antognini +5

Google DeepMind

In-Context Learning Language Model Self-Improvement

Feb 16, 2026

Evolutionary System Prompt Learning for Reinforcement Learning in LLMs

Building agentic systems that can autonomously self-improve from experience is a longstanding goal of AI. Large language models (LLMs) today primarily self-improve via two mechanisms: self-reflection for context updates, and reinforcement learning (RL) for weight updates. In this work, we propose Evolutionary System Prompt Learning (E-SPL), a method for jointly improving model contexts and model weights. In each RL iteration, E-SPL samples trajectories under multiple system prompts in parallel, then jointly applies RL updates to LLM weights and evolutionary updates to system prompts. System prompts evolve via mutation and crossover, two genetic operators driven by LLM self-reflection; selection is based on relative performance ratings updated across RL iterations. E-SPL encourages a natural division between declarative knowledge encoded in prompts and procedural knowledge encoded in weights, resulting in improved performance across reasoning and agentic tasks. For instance, in an easy-to-hard (AIME

\rightarrow

BeyondAIME) generalization setting, E-SPL improves RL success rate from 38.8%

\rightarrow

45.1% while also outperforming reflective prompt evolution (40.0%). Overall, our results demonstrate that RL and system prompt evolution are deeply synergistic, and combining the two yields consistent gains in sample efficiency and generalization. Code: https://github.com/LunjunZhang/E-SPL

Lunjun Zhang, Ryan Chen, Bradly C. Stadie

Department of Computer Science, University of Toronto · Department of Statistics and Data Science, Northwestern University · Bridgewater AIA Labs

Language Model Self-Improvement RL for Language Models RL for Language Model Reasoning Evolutionary Optimization

Feb 6, 2026

TTSR: Test-Time Self-Evolving via Reflection

Test-time training (TTT) adapts large language models (LLMs) during inference using only unlabeled test inputs. Existing methods, however, face two major bottlenecks on hard reasoning tasks: (1) lack of learnable samples, as self-generated pseudo-labels on difficult questions are often noisy and yield unstable rewards; and (2) inefficient exploration, as performance gains depend on repeatedly sampling many rollouts without explicit diagnosis of why previous attempts fail. We propose TTSR (Test-Time Self-Reflection), a self-evolving framework based on a reflect-then-synthesize paradigm. A single pretrained model alternates between a Student role and a Teacher role: the Student solves test questions and updates, while the Teacher analyzes failed trajectories and synthesizes targeted variant questions closer to the Student's capability frontier. TTSR further maintains a cross-iteration weakness memory and compiles persistent weaknesses into a lightweight strategy note prepended to subsequent Student inputs, so diagnostic knowledge can guide exploration and gradually fade as weaknesses are resolved. Experiments on challenging mathematical reasoning benchmarks show consistent test-time improvements, strong cross-backbone generalization, and transfer to general-domain reasoning tasks.

Haoyang He, Zihua Rong, Yunjia Zhao +3

Beijing University of Posts and Telecommunications · Southwestern University of Finance and Economics · China Unicom Online Information Technology Co., Ltd.

LLM Self-Refinement Test-Time Optimization Test-Time Training Teacher-Student Learning Language Model Self-Improvement

Jan 30, 2026

TTCS: Test-Time Curriculum Synthesis for Self-Evolving

Test-Time Training offers a promising way to improve the reasoning ability of large language models (LLMs) by adapting the model using only the test questions. However, existing methods struggle with difficult reasoning problems for two reasons: raw test questions are often too difficult to yield high-quality pseudo-labels, and the limited size of test sets makes continuous online updates prone to instability. To address these limitations, we propose TTCS, a co-evolving test-time training framework. Specifically, TTCS initializes two policies from the same pretrained model: a question synthesizer and a reasoning solver. These policies evolve through iterative optimization: the synthesizer generates progressively challenging question variants conditioned on the test questions, creating a structured curriculum tailored to the solver's current capability, while the solver updates itself using self-consistency rewards computed from multiple sampled responses on both original test and synthetic questions. Crucially, the solver's feedback guides the synthesizer to generate questions aligned with the model's current capability, and the generated question variants in turn stabilize the solver's test-time training. Experiments show that TTCS consistently strengthens the reasoning ability on challenging mathematical benchmarks and transfers to general-domain tasks across different LLM backbones, highlighting a scalable path towards dynamically constructing test-time curricula for self-evolving. Our code and implementation details are available at https://github.com/XMUDeepLIT/TTCS.

Chengyi Yang, Zhishang Xiang, Yunbo Tang +5

Xiamen University · Washington University in St. Louis · Renmin University of China

Test-Time Training Language Model Self-Improvement Test-Time RL Numerical Reasoning in Language Models

Jan 20, 2026

Self-Improvement as Coherence Optimization: A Theoretical Account

Can language models improve their accuracy without external supervision? Methods such as debate, bootstrap, and internal coherence maximization achieve this surprising feat, even matching golden finetuning performance. Yet why they work remains theoretically unclear. We show that they can all be understood as coherence optimization, the search for a context-to-behavior mapping that is most compressible and jointly predictable, with debate an exact instance and bootstrap and internal coherence maximization closely related to it. We prove that coherence optimization is equivalent to description-length regularization, and that among all such regularization schemes, coherence regularization with a prior derived from a pretrained model optimizes a lower bound of worst-case accuracy for semi-supervised learning. Our theory, supported by preliminary experiments, explains why feedback-free self-improvement works and predicts when it should succeed or fail.

Tianyi Qiu, Ahmed Hani Ismail, Zhonghao He +1

Peking University · University of Oxford · UC Berkeley +1

Language Model Self-Improvement

Nov 1, 2025

OpenSIR: Open-Ended Self-Improving Reasoner

Recent advances in large language model (LLM) reasoning through reinforcement learning rely on annotated datasets for verifiable rewards, which may limit models' ability to surpass human-level performance. While self-play offers a promising alternative, prior methods yield only marginal or even negative gains on post-trained models because they generate problems that cluster around familiar concepts rather than discovering novel ones. We introduce Open-Ended Self-Improving Reasoner (OpenSIR), a self-play framework in which a single LLM alternates teacher and student roles to generate and solve novel problems without external verifiers or annotated data. Starting from a single seed problem, OpenSIR sustains open-ended exploration through diversity rewards that push the model toward unfamiliar concepts and difficulty calibration that keeps problems learnable. Across seven math benchmarks, OpenSIR consistently improves all models, averaging +3.6 points on instruction models and +3.1 on reasoning models, while recent self-play baselines yield marginal or even negative gains; starting from a single trivial seed, it also surpasses GRPO baselines trained on over 7K annotated examples. Despite training only on self-generated math, OpenSIR is the only self-play method that transfers to general reasoning, improving by at least +4.4 points on reasoning models.

Wai-Chung Kwan, Joshua Ong Jun Leang, Pavlos Vougiouklis +3

University of Edinburgh · Imperial College London · Huawei Technologies Research & Development (UK) Limited +1

Self-Play RL RL for Language Model Reasoning Language Model Self-Improvement LLM Mathematical Reasoning Large Reasoning Models

Oct 21, 2025

A Model Can Help Itself: Reward-Free Self-Training for LLM Reasoning

Can a language model improve reasoning by learning from its own imperfect responses, without rewards or teacher-provided solutions? We present Self-evolving Post-Training (SePT), a simple method that alternates temperature-controlled self-generation with next-token likelihood training. Each round uses the updated model to generate new training responses, with one response per prompt by default and no correctness filtering. Across six mathematical benchmarks, SePT improves a temperature-selected no-training baseline by 11.4 and 6.7 AVG points on Qwen2.5-Math-7B and Qwen2.5-7B, respectively, where AVG averages Pass@1, Pass@8 and Pass@32 across benchmarks. We analyze how sampling temperature shapes the learning signal and investigate the value of the resulting responses. Responses from a SePT-trained model improve a student initialized from the original weights, while their reasoning prefixes help an unchanged model complete solutions, even when matched in length to prefixes from a colder initial model. Comparing next-token predictions at identical contexts also reveals changes in token rankings that decoding-temperature adjustment cannot reproduce. Further evaluations across nine starting models, general reasoning and code generation examine broader applicability. Together, these results show that reward-free self-training can improve both a model's predictions and the supervision it provides. Our code is available at https://github.com/ElementQi/SePT.

Mengqi Li, Lei Zhao, Anthony Man-Cho So +2

The Chinese University of Hong Kong, Shenzhen · Shanghai Jiao Tong University · The Chinese University of Hong Kong

Self-Training for Language Models Mathematical Reasoning LLM Post-Training Language Model Self-Improvement LLM Reasoning

Sep 29, 2025

RFG: Self-Improving Diffusion Large Language Models with Reward-Free Guidance

Diffusion Large Language Models (dLLMs) have shown strong reasoning capabilities, yet further improving them typically requires costly post-training with additional data and supervision. We ask whether a post-trained dLLM can improve itself at inference time without additional training, data, or reward models. This requires a guidance signal from the checkpoints alone that is well-defined on the partially masked intermediate states of dLLMs, which existing methods fail to provide. Here we propose Reward-Free Guidance (RFG), a training-free framework for inference-time self-improvement of dLLMs that derives such a signal from the checkpoints themselves. We theoretically demonstrate that a reward signal for a partially masked state can be parameterized by the log-likelihood ratio between a post-trained policy dLLM and its reference model. In practice, this signal can be obtained using off-the-shelf checkpoints alone. Extensive experiments show that RFG consistently improves state-of-the-art post-trained dLLMs by up to 16.1%. Furthermore, despite being entirely training-free, RFG delivers gains that rival or even surpass resource-intensive reinforcement learning techniques.

Tianlang Chen, Minkai Xu, Jure Leskovec +1

Stanford University

Diffusion Guidance Diffusion Language Model Inference Diffusion Language Models Language Model Self-Improvement

Sep 18, 2025

Evolving Language Models without Labels: Majority Drives Selection, Novelty Promotes Variation

Large language models (LLMs) can improve through reinforcement learning with verifiable rewards (RLVR), but many settings lack reference answers or external verifiers. Label-free self-improvement instead constructs training signals from the model's own outputs. Because the same policy generates both the candidate responses and the signals derived from them, each update also changes the candidates available for future training, creating a self-reinforcing feedback loop. We identify a failure mode of this loop, which we call selection-induced diversity collapse: repeated self-selection can progressively narrow the candidate pool and make alternative correct solutions harder to sample. To address this problem, we propose EVOL-RL, which balances selection and variation. Majority voting provides a stable selection signal, while embedding-based novelty refines credit among responses with the same selection status; entropy regularization and asymmetric clipping further support this process by maintaining diverse rollouts and preserving rare, high-advantage updates. Across multiple training distributions, model scales, backbones, and matched label-free baselines, EVOL-RL consistently improves pass@n while generally maintaining or improving pass@1. For example, when trained without labels on AIME24, Qwen3-4B-Base improves over TTRL on held-out AIME25 from 4.6% to 17.1% in pass@1 and from 18.5% to 42.0% in pass@16. Training analyses further show that when the current majority is wrong, EVOL-RL better preserves alternative correct solutions. The code is available at: https://github.com/YujunZhou/EVOL-RL.

Yujun Zhou, Zhenwen Liang, Haolin Liu +7

Tencent AI Lab · University of Notre Dame · University of Virginia

Language Model Self-Improvement RL for Language Models RL for Language Model Reasoning

Sep 16, 2025

Discovering New Theorems via LLMs with In-Context Proof Learning in Lean

Large Language Models (LLMs) have demonstrated significant promise in formal theorem proving. In this study, we investigate the ability of LLMs to discover novel theorems and produce verified proofs. We propose a pipeline called Conjecturing-Proving Loop (CPL), which iteratively generates mathematical conjectures and attempts to prove them in Lean 4. A key feature of CPL is that each iteration conditions the LLM on previously generated theorems and their formal proofs, enabling parameter-free improvement of proof strategies via in-context learning. We provide both theoretical and experimental evidence that CPL increases the discovery rate of hard-to-prove theorems compared to frameworks that generate statements and proofs simultaneously. Moreover, our experiments show that reusing the LLM's own formally verified outputs as context consistently improves subsequent proof success, demonstrating the effectiveness of self-generated in-context learning for neural theorem proving. The source code is available at https://github.com/auto-res/ConjecturingProvingLoop.

Kazumi Kasaura, Naoto Onda, Yuta Oriike +3

OMRON SINIC X Corporation · RIKEN AIP · AutoRes +7

Automated Theorem Proving In-Context Learning Lean Theorem Proving Language Model Self-Improvement Large Language Model-Guided Theorem Proving

Aug 27, 2025

ReST-RL: Reinforcing LLM Reasoning through Unified Self-Training and Value-Guided Search

With respect to improving the reasoning accuracy of LLMs, the representative reinforcement learning (RL) method - Group Relative Policy Optimization (GRPO) - has achieved critical success, yet it still suffers from the issue of insignificant reward signals. This paper introduces ReST-RL, a unified Reinforced Self-Training (ReST) policy-value framework that reconnects policy optimization and value-guided search to improve LLM reasoning ability. Firstly, ReST-GRPO adopts an optimized ReST-style algorithm to reshape the policy-induced trajectory distribution by increasing the reward variance of GRPO sampling and exposing the policy to more informative partial states, thereby improving training efficiency and effectiveness. Then, we further introduce a decoding optimization method, VM-MCTS, which trains a Value Model (VM) from self-collected Monte-Carlo Tree Search (MCTS) targets and deploys it through an adapted MCTS algorithm to provide precise process signals and verification scores, further enhancing reasoning accuracy. These two stages are internally dependent - ReST-GRPO yields higher-quality trajectories for value learning with VM-MCTS, which in turn enables more effective inference-time search. We validate our framework on multiple coding benchmarks (e.g., APPS, BigCodeBench, and HumanEval), where it significantly outperforms other reinforcement training baselines (naive GRPO, DAPO, and ReST-DPO), as well as decoding and verification baselines (e.g., PRM-BoN and ORM-MCTS), indicating its power to strengthen LLM reasoning capability. Moreover, we further evaluate ReST-RL on out-of-domain math and science reasoning tasks, where it achieves improved performance without target-domain tuning and favorable end-to-end efficiency trade-offs, providing preliminary transfer evidence beyond our primary coding domain.

Sining Zhoubian, Dan Zhang, Jie Tang

Department of Computer Science and Technology, Tsinghua University · Work done when interned at Z.ai. · School of Electronics and Computer Scinence, Univeristy of Southampton

RL for Code Generation RL for Language Model Reasoning Monte Carlo Tree Search RL Post-Training Language Model Self-Improvement

Jul 22, 2025

SyncLoop: A Multimodal Dual-Loop Framework for Self-Improving Mathematical Reasoning

Recent advances in multimodal large language models (MLLMs) have shown impressive reasoning capabilities. However, further enhancing existing MLLMs necessitates high-quality vision-language datasets with carefully curated task complexities, which are both costly and challenging to scale. Although recent self-improving models that iteratively refine themselves offer a feasible solution, they still suffer from two core challenges: (i) most existing methods augment visual or textual data separately, resulting in discrepancies in data complexity (e.g., over-simplified diagrams paired with redundant textual descriptions); and (ii) the evolution of data and models is also separated, leading to scenarios where models are exposed to tasks with mismatched difficulty levels. To address these issues, we propose C2-Evo, an automatic, closed-loop self-improving framework that jointly evolves both training data and model capabilities. Specifically, given a base dataset and a base model, C2-Evo enhances them by a cross-modal data evolution loop and a data-model evolution loop. The former loop expands the base dataset by generating complex multimodal problems that combine structured textual sub-problems with iteratively specified geometric diagrams, while the latter loop adaptively selects the generated problems based on the performance of the base model, to conduct supervised fine-tuning and reinforcement learning alternately. Consequently, our method continuously refines its model and training data, and consistently obtains considerable performance gains across multiple mathematical reasoning benchmarks. Our code, models, and datasets will be released.

Xiuwei Chen, Wentao Hu, Hanhui Li +9

Sun Yat-sen University · The Hong Kong Polytechnic University · MBZUAI +2

Synthetic Data Generation Multimodal Large Language Models Multimodal Reasoning Language Model Self-Improvement

Oct 31, 2024

RL-STaR: Theoretical Analysis of Reinforcement Learning Frameworks for Self-Taught Reasoner

The reasoning abilities of large language models (LLMs) have improved with chain-of-thought (CoT) prompting, allowing models to solve complex tasks stepwise. However, training CoT capabilities requires detailed reasoning data, which is often scarce. The self-taught reasoner (STaR) framework addresses this by using reinforcement learning to automatically generate reasoning steps, reducing reliance on human-labeled data. Although STaR and its variants have demonstrated empirical success, a theoretical foundation explaining these improvements is lacking. This work provides a theoretical framework for understanding the effectiveness of reinforcement learning on CoT reasoning and STaR. Our contributions are: (1) criteria for the quality of pre-trained models necessary to initiate effective reasoning improvement; (2) an analysis of policy improvement, showing why LLM reasoning improves iteratively with STaR; (3) conditions for convergence to an optimal reasoning policy; and (4) an examination of STaR's robustness, explaining how it can improve reasoning even when incorporating occasional incorrect steps. We also run RL-STaR on GPT-2, Qwen2.5-0.5B and Phi-3-mini, and the measured return curves follow the ones the analysis predicts. This framework bridges empirical findings with theoretical insights, advancing reinforcement learning approaches for reasoning in LLMs.

Fu-Chieh Chang, Yu-Ting Lee, Hui-Ying Shih +2

MediaTek Research, Taipei, Taiwan · Graduate Institute of Communication Engineering, National Taiwan University, Taipei, Taiwan · Department of Mathematical Sciences, National Chengchi University, Taipei, Taiwan +2

Reinforcement Learning RL for Language Model Reasoning Language Model Self-Improvement LLM Reasoning