LLM Reasoning

LLM: Large Language Model

Ask about this topic

Overview

This field asks how language models can carry out multi-step inference across mathematics, coding, logic, science, and real-world tasks, and how to tell robust reasoning from pattern matching. Common approaches include chain-of-thought prompting, self-consistency and tree search at inference time, tool use and symbolic verification, and post-training with supervised traces, reinforcement learning from verifiable rewards, or process feedback. Researchers are also exploring latent and recurrent computation to reduce reliance on long written traces, alongside benchmarks that test compositionality, robustness, and long-horizon execution. A central tension is that more computation can improve accuracy but raise cost, while plausible explanations and correct final answers do not by themselves establish faithful, reliable reasoning.

Latest papers

CardsList

931 papers105 papers in the last four weeks, up 27% on the four weeks before. 0.6% of all new papers.

Mar 21, 2026

From 50% to Mastery in 3 Days: A Low-Resource SOP for Localizing Graduate-Level AI Tutors via Shadow-RAG

Deploying high-fidelity AI tutors in schools is often blocked by the Resource Curse -- the need for expensive cloud GPUs and massive data engineering. In this practitioner report, we present a replicable Standard Operating Procedure that breaks this barrier. Using a Vision-Language Model data cleaning strategy and a novel Shadow-RAG architecture, we localized a graduate-level Applied Mathematics tutor using only 3 person-days of non-expert labor and open-weights 32B models deployable on a single consumer-grade GPU. Our pilot study on a full graduate-level final exam reveals a striking emergence phenomenon: while both zero-shot baselines and standard retrieval stagnate around 50-60% accuracy across model generations, the Shadow Agent, which provides structured reasoning guidance, triggers a massive capability surge in newer 32B models, boosting performance from 74% (Naive RAG) to mastery level (90%). In contrast, older models see only modest gains (~10%). This suggests that such guidance is the key to unlocking the latent power of modern small language models. This work offers a cost-effective, scientifically grounded blueprint for ubiquitous AI education.

Zonglin Yang, J. -H. Xie, Lining Zhang +2

School of Mechanics and Engineering Science, Peking University, Beijing, 100871, China · Readraft Intelligent Technology Co., Ltd., China · Graduate School of Education, Peking University, Beijing, 100871, China +1

Generative AI in EducationIntelligent Tutoring SystemsRetrieval-Augmented GenerationLLM Reasoning
AskPDF
Mar 20, 2026

Grounded Chess Reasoning in Language Models via Master Distillation

Language models often lack grounded reasoning capabilities in specialized domains where training data is scarce but bespoke systems excel. We introduce a general framework for distilling expert system reasoning into natural language chain-of-thought explanations, enabling compact models to acquire domain expertise and the ability to generate faithful, grounded explanations. Rather than distilling only final outputs, we capture the full reasoning process, transforming opaque expert computations into transparent, step-by-step explanations. We demonstrate this approach in chess, a canonical reasoning domain where language models continue to underperform. Our 4B parameter model, C1, advances from a near-zero baseline to 48.1% accuracy, outperforming all open-source models and most frontier proprietary systems. Notably, C1 surpasses its distillation teacher and generates solutions in two orders of magnitude fewer tokens than baselines. Unlike prior neural chess approaches that predict only best moves, C1 generates explainable solutions revealing strategic reasoning. Our pipeline combines supervised fine-tuning and reinforcement learning with theme-balanced data sampling for comprehensive tactical coverage. Master Distillation demonstrates how to inject expert-level knowledge into compact models for under-optimized domains, offering a recipe for unlocking RLVR where LLMs lack sufficient base capabilities.

Zhenwei Tang, Qianfeng Wen, Seth Grief-Albert +4

Department of Computer Science, University of Toronto · Queen’s University · Coolwei AI Lab

LLM GroundingCoT DistillationLanguage Model DistillationChessLLM Reasoning
AskPDF
Mar 20, 2026

Learning Dynamic Belief Graphs for Theory-of-mind Reasoning

Theory of Mind (ToM) reasoning with Large Language Models (LLMs) requires inferring how people's implicit, evolving beliefs shape what they seek and how they act under uncertainty -- especially in high-stakes settings such as disaster response, emergency medicine, and human-in-the-loop autonomy. Prior approaches either prompt LLMs directly or use latent-state models that treat beliefs as static and independent, often producing incoherent mental models over time and weak reasoning in dynamic contexts. We introduce a structured cognitive trajectory model for LLM-based ToM that represents mental state as a dynamic belief graph, jointly inferring latent beliefs, learning their time-varying dependencies, and linking belief evolution to information seeking and decisions. Our model contributes (i) a novel projection from textualized probabilistic statements to consistent probabilistic graphical model updates, (ii) an energy-based factor graph representation of belief interdependencies, and (iii) an ELBO-based objective that captures belief accumulation and delayed decisions. Across multiple real-world disaster evacuation datasets, our model significantly improves action prediction and recovers interpretable belief trajectories consistent with human reasoning, providing a principled module for augmenting LLMs with ToM in high-uncertainty environment. https://anonymous.4open.science/r/ICML_submission-6373/

Ruxiao Chen, Xilei Zhao, Thomas J. Cova +2

Department of Civil and Systems Engineering, Johns Hopkins University, Baltimore, MD, USA · Civil and Coastal Engineering, University of Florida, Gainesville, FL, USA · School of the Environment, Society & Sustainability, University of Utah, Salt Lake City, UT, USA +1

Computational Cognitive ModelingLLM Reasoning with GraphsTheory of MindProbabilistic Graphical ModelsDisaster ResponseLLM Reasoning
AskPDF
Mar 20, 2026

FIPO: Eliciting Deep Reasoning with Future-KL Influenced Policy Optimization

We present Future-KL Influenced Policy Optimization (FIPO), a reinforcement learning algorithm designed to overcome reasoning bottlenecks in large language models. While GRPO style training scales effectively, it typically relies on outcome-based rewards (ORM) that distribute a global advantage uniformly across every token in a trajectory. We argue that this coarse-grained credit assignment imposes a performance ceiling by failing to distinguish critical logical pivots from trivial tokens. FIPO addresses this by incorporating discounted future-KL divergence into the policy update, creating a dense advantage formulation that re-weights tokens based on their influence on subsequent trajectory behavior. Empirically, FIPO enables models to break through the length stagnation seen in standard baselines. Evaluated on Qwen2.5-32B, FIPO extends the average chain-of-thought length from roughly 4,000 to over 10,000 tokens and increases AIME 2024 Pass@1 accuracy from 50.0% to a peak of 58.0% (converging at approximately 56.0%). This outperforms both DeepSeek-R1-Zero-Math-32B (around 47.0%) and o1-mini (approximately 56.0%). Our results suggest that establishing dense advantage formulations is a vital path for evolving ORM-based algorithms to unlock the full reasoning potential of base models. We open-source our training system, built on the verl framework.

Chiyu Ma, Shuo Yang, Kexin Huang +7

Qwen Pilot Team, Alibaba Group

RL for Language Model ReasoningRL for Language ModelsLLM Reasoning
AskPDF
Mar 20, 2026

One Model, Two Minds: Task-Conditioned Reasoning for Unified Image Quality and Aesthetic Assessment

Unifying Image Quality Assessment (IQA) and Image Aesthetic Assessment (IAA) in a single multimodal large language model is appealing, yet existing methods adopt a task-agnostic recipe that applies the same reasoning strategy and reward to both tasks. We show this is fundamentally misaligned: IQA relies on low-level, objective perceptual cues and benefits from concise distortion-focused reasoning, whereas IAA requires deliberative semantic judgment and is poorly served by point-wise score regression. We identify these as a reasoning mismatch and an optimization mismatch, and provide empirical evidence for both through controlled probes. Motivated by these findings, we propose TATAR (Task-Aware Thinking with Asymmetric Rewards), a unified framework that shares the visual-language backbone while conditioning post-training on each task's nature. TATAR combines three components: fast--slow task-specific reasoning construction that pairs IQA with concise perceptual rationales and IAA with deliberative aesthetic narratives; two-stage SFT+GRPO learning that establishes task-aware behavioral priors before reward-driven refinement; and asymmetric rewards that apply Gaussian score shaping for IQA and Thurstone-style completion ranking for IAA. Extensive experiments across eight benchmarks demonstrate that TATAR consistently outperforms prior unified baselines on both tasks under in-domain and cross-domain settings, remains competitive with task-specific specialized models, and yields more stable training dynamics for aesthetic assessment. Our results establish task-conditioned post-training as a principled paradigm for unified perceptual scoring. Our code is publicly available at https://github.com/yinwen2019/TATAR.

Wen Yin, Cencen Liu, Dingrui Liu +3

University of Electronic Science and Technology of China, Chengdu, Sichuan, China · Jiigan Technology, Chengdu, Sichuan, China · Faculty of Information Technology, Monash University, Melbourne, Australia +1

Image Quality AssessmentUnified Multimodal ModelsLLM Reasoning
AskPDF
Mar 19, 2026

Can LLMs Prove Robotic Path Planning Optimality? A Benchmark for Research-Level Algorithm Verification

Robotic path planning problems are often NP-hard, and practical solutions typically rely on approximation algorithms with provable performance guarantees for general cases. While designing such algorithms is challenging, formally proving their approximation optimality is even more demanding, which requires domain-specific geometric insights and multi-step mathematical reasoning over complex operational constraints. Recent Large Language Models (LLMs) have demonstrated strong performance on mathematical reasoning benchmarks, yet their ability to assist with research-level optimality proofs in robotic path planning remains under-explored. In this work, we introduce the first benchmark for evaluating LLMs on approximation-ratio proofs of robotic path planning algorithms. The benchmark consists of 34 research-grade proof tasks spanning diverse planning problem types and complexity levels, each requiring structured reasoning over algorithm descriptions, problem constraints, and theoretical guarantees. Our evaluation of state-of-the-art proprietary and open-source LLMs reveals that even the strongest models struggle to produce fully valid proofs without external domain knowledge. However, providing LLMs with task-specific in-context lemmas substantially improves reasoning quality, a factor that is more effective than generic chain-of-thought prompting or supplying the ground-truth approximation ratio as posterior knowledge. We further provide fine-grained error analysis to characterize common logical failures and hallucinations, and demonstrate how each error type can be mitigated through targeted context augmentation.

Zhengbang Yang, Md. Tasin Tazwar, Minghan Wei +1

George Mason University · Florida Atlantic University

Mathematical Reasoning BenchmarksAutomated Theorem ProvingApproximation AlgorithmsRobot Motion PlanningLLM Reasoning
AskPDF
Mar 19, 2026

Correct Chains, Wrong Answers: Dissociating Reasoning from Output in LLM Logic

LLMs can execute every step of chain-of-thought reasoning correctly and still produce wrong final answers. We introduce the Novel Operator Test, a benchmark that separates operator logic from operator name, enabling rigorous distinction between genuine reasoning and pattern retrieval. By evaluating Boolean operators under unfamiliar names across depths 1-10 on five models (up to 8,100 problems each), we demonstrate a reasoning-output dissociation that existing benchmarks cannot detect. At Claude Sonnet 4's depth 7, all 31 errors have verifiably correct reasoning yet wrong declared answers; 17/19 errors in mixed-operator chains exhibit the same pattern. The benchmark reveals two failure types: strategy failures at depth 2, where models attempt terse retrieval (+62pp from scaffolding), and content failures at depth 7, where models reason fully but err systematically (+8-30pp, 0/300 errors post-intervention). A Trojan operator (XOR's truth table under a novel name) confirms name alone does not gate reasoning (p >= 0.49), while Llama's novelty gap widens to 28pp at depth 8-9 with the Trojan at 92-100%, isolating genuine difficulty with novel logic from name unfamiliarity.

Abinav Rao, Sujan Rachuri, Nikhil Vemuri

LLM ReasoningLLM EvaluationReasoning Consistency in Language ModelsStructured CoT Reasoning
AskPDF
Mar 19, 2026

Entropy trajectory shape predicts LLM reasoning reliability: A diagnostic study of uncertainty dynamics in chain-of-thought

Understanding uncertainty in chain-of-thought reasoning is critical for reliable deployment of large language models. In this work, we propose a simple yet effective diagnostic approach based on trajectory shape rather than scalar magnitude. We show that this signal is practical, interpretable, and inexpensive to obtain in black-box settings, while remaining robust across models and datasets. Through extensive ablations and cross-domain replications, we demonstrate its utility for selective prediction and triage. Our findings offer a generalizable insight into uncertainty dynamics in reasoning tasks, with particular focus on numeric and discrete-answer settings.

Xinghao Zhao

Huazhong University of Science and Technology

LLM Uncertainty EstimationLLM ReliabilityLLM ReasoningSelective Prediction
AskPDF
Mar 19, 2026

ZEBRAARENA: A Diagnostic Simulation Environment for Studying Reasoning-Action Coupling in Tool-Augmented LLMs

Tool-augmented large language models (LLMs) must tightly couple multi-step reasoning with external actions, yet existing benchmarks often confound this interplay with complex environment dynamics, memorized knowledge or dataset contamination. In this paper, we introduce ZebraArena, a procedurally generated diagnostic environment for studying reasoning-action coupling in tool-augmented LLMs, with controllable difficulty and a knowledge-minimal design, which limits gains from memorization or dataset contamination. Each task in ZebraArena requires a set of critical information which is available only through targeted tool use, yielding an interpretable interface between external information acquisition and deductive reasoning. This design provides deterministic evaluation via unique solutions, and a theoretical optimal query count for measuring efficient tool use. We show that ZebraArena requires a combination of in-depth reasoning and accurate external tool calling, which remains a challenge as frontier reasoning models such as GPT-5 and Gemini 2.5 Pro only achieves 60% accuracy on the hard instances. We also observe a persistent gaps between theoretical optimality and practical tool usage. For example, GPT-5 uses 70-270% more tool calls than the theoretical optimum. We highlight the key findings in our evaluation, and hope ZebraArena stimulates further research on the interplay between internal reasoning and external action.

Wanjia Zhao, Ludwig Schmidt, Yejin Choi +3

Stanford University · Microsoft Research

LLM EvaluationAI Agent BenchmarksLLM Tool UseLLM Reasoning
AskPDF
Mar 19, 2026

TARo: Token-level Adaptive Routing for LLM Test-time Alignment

Large language models (LLMs) exhibit strong reasoning capabilities but typically require expensive post-training to reach high performance. Recent test-time alignment methods offer a lightweight alternative, but have been explored mainly for preference alignment rather than reasoning. To bridge this gap, we propose, Token-level Adaptive Routing (TARo), which steers frozen LLMs toward structured reasoning entirely at inference time. Specifically, we first train reward models on step-wise mathematical traces to capture fine-grained logical consistency signals, then introduce a learnable token-level router that automatically controls the guidance of the reward model to the base model. Extensive experiments show that TARo significantly improves reasoning performance by up to +22.4% over base model and +8.4% over existing token-level test-time alignment methods, while also boosting out-of-distribution clinical reasoning (MedXpertQA) and instruction following (AlpacaEval). Furthermore, TARo also generalizes from small to large backbones without retraining, extending test-time alignment from preference optimization to robust, cross-domain reasoning.

Arushi Rai, Qiang Zhang, Hanqing Zeng +5

Meta · University of Pittsburgh · University of California, Berkeley

Inference-Time Language Model AlignmentLLM ReasoningLLM Routing
AskPDF
Mar 18, 2026

Synthetic Data Generation for Training Diversified Commonsense Reasoning Models

Conversational agents are required to respond to their users not only with high quality (i.e. commonsense bearing) responses, but also considering multiple plausible alternative scenarios, reflecting the diversity in their responses. Despite the growing need to train diverse commonsense generators, the progress of this line of work has been significantly hindered by the lack of large-scale high-quality diverse commonsense training datasets. Due to the high annotation costs, existing Generative Commonsense Reasoning (GCR) datasets are created using a small number of human annotators, covering only a narrow set of commonsense scenarios. To address this training resource gap, we propose a two-stage method to create the first-ever synthetic dataset CommonSyn for diversified (GCR). The model fine-tuned on our synthetic data jointly increase both generation diversity and quality compared with vanilla models and the model fine-tuned on human-crafted dataset across different size Large Language Models (LLMs)

Tianhui Zhang, Bei Peng, Danushka Bollegala

University of Liverpool · University of Sheffield · Amazon

Synthetic Data Generation for Language ModelsSynthetic Data Generation for LLM AgentsDiverse Text GenerationLLM Reasoning
AskPDF
Mar 18, 2026

RAWR: Reward Assignment Without Rollouts in Verifiable Domains

Understanding and evaluating multi-step reasoning in LLMs at the level of individual steps remains a key challenge. Process reward models (PRMs) provide a solution by scoring each step, enabling fine-grained supervision and improved reliability. However, training them requires costly human annotation or computationally intensive rollout-based labeling. To solve this, we introduce MCNIG, a scalable method for automatically labeling the quality of individual reasoning steps in any verifiable domain. Its step score, net information gain (NetIG), improves upon single-reference information gain (IG) by comparing the most-supported correct answer against the most-supported incorrect one, yielding a robust signal even for long and structured outputs like code and SQL, where IG fails. We show that the signal produced by MCNIG correlates with human judgments of step quality, and we apply MCNIG labels to train PRMs that achieve the best average best-of-K accuracy across eight benchmarks spanning mathematics, code generation, text-to-SQL, and scientific QA. Crucially, MCNIG generates no rollouts, cutting labeling complexity to O(N) and making it up to X times cheaper than rollout-based methods at comparable label quality, which makes large-scale process supervision practical.

Corentin Royer, Anna Hedström, Debarun Bhattacharjya +3

International Business Machines · ETH AI Center · Lirio +1

Process Reward ModelsReinforcement Learning with Verifiable RewardsProcess SupervisionStep-Level Credit AssignmentLLM Reasoning
AskPDF
Mar 18, 2026

CoVerRL: Breaking the Consensus Trap in Label-Free Reasoning via Generator-Verifier Co-Evolution

Label-free reinforcement learning enables large language models to improve reasoning capabilities without ground-truth supervision, typically by treating majority-voted answers as pseudo-labels. However, we identify a critical failure mode: as training maximizes self-consistency, output diversity collapses, causing the model to confidently reinforce systematic errors that evade detection. We term this the consensus trap. To escape it, we propose CoVerRL, a framework where a single model alternates between generator and verifier roles, with each capability bootstrapping the other. Majority voting provides noisy but informative supervision for training the verifier, while the improving verifier progressively filters self-consistent errors from pseudo-labels. This co-evolution creates a virtuous cycle that maintains high reward accuracy throughout training. Experiments across Qwen and Llama model families demonstrate that CoVerRL outperforms label-free baselines by 4.7-5.9% on mathematical reasoning benchmarks. Moreover, self-verification accuracy improves from around 55% to over 85%, confirming that both capabilities genuinely co-evolve.

Teng Pan, Yuchen Yan, Zixuan Wang +6

Zhejiang University · Baidu Inc.

Language Model Self-VerificationRL for Language Model ReasoningLLM ReasoningSelf-Training for Language ModelsSelf-Consistency in Language Models
AskPDF
Mar 18, 2026

Temperature-Dependent Performance of Prompting Strategies in Extended Reasoning Large Language Models

Extended reasoning models represent a transformative shift in Large Language Model (LLM) capabilities by enabling explicit test-time computation for complex problem solving. However, the optimal configuration of sampling temperature and prompting strategy for these systems remains largely underexplored. We systematically evaluate chain-of-thought and zero-shot prompting across four temperature settings (0.0, 0.4, 0.7, and 1.0) using Grok-4.1 with extended reasoning on 39 mathematical problems from AMO-Bench, a challenging International Mathematical Olympiad-level benchmark. We find that zero-shot prompting achieves peak performance at moderate temperatures, reaching 59% accuracy at T=0.4 and T=0.7, while chain-of-thought prompting performs best at the temperature extremes. Most notably, the benefit of extended reasoning increases from 6x at T=0.0 to 14.3x at T=1.0. These results suggest that temperature should be optimized jointly with prompting strategy, challenging the common practice of using T=0 for reasoning tasks.

Mousa Salah, Amgad Muneer

Department of Computer Engineering, Gujarat Technological University, India · The University of Texas MD Anderson Cancer Center, Texas, USA

LLM PromptingLLM ReasoningLarge Reasoning ModelsLanguage Model Decoding
AskPDF
Mar 18, 2026

Draft-and-Prune: Improving the Reliability of Auto-formalization for Logical Reasoning

Auto-formalization (AF) translates natural-language reasoning problems into solver-executable programs, enabling symbolic solvers to perform sound logical deduction. In practice, however, AF pipelines are currently brittle: programs may fail to execute, or execute but encode incorrect semantics. While prior work largely mitigates syntactic failures via repairs based on solver feedback, reducing semantics failures remains a major bottleneck. We propose Draft-and-Prune (D&P), an inference-time framework that improves AF-based logical reasoning via diversity and verification. D&P first drafts multiple natural-language plans and conditions program generation on them. It further prunes executable but contradictory or ambiguous formalizations, and aggregates predictions from surviving paths via majority voting. Across four representative benchmarks (AR-LSAT, ProofWriter, PrOntoQA, LogicalDeduction), D&P substantially strengthens AF-based reasoning without extra supervision. On AR-LSAT, in the AF-only setting, D&P achieves 78.43% accuracy with GPT-4 and 78.00% accuracy with GPT-4o, significantly outperforming the strongest AF baselines MAD-LOGIC and CLOVER. D&P then attains near-ceiling performance on the other benchmarks, including 100% on PrOntoQA and LogicalDeduction.

Zhiyu Ni, Zheng Liang, Liangcheng Song +4

University of California, Berkeley · Microsoft

AutoformalizationLLM ReasoningSymbolic Reasoning
AskPDF
Mar 17, 2026

When and Why Does Unsupervised RL Succeed in Mathematical Reasoning? A Manifold Envelopment Perspective

Although outcome-based reinforcement learning (RL) significantly advances the mathematical reasoning capabilities of Large Language Models (LLMs), its reliance on computationally expensive ground-truth annotations imposes a severe scalability bottleneck. Unsupervised RL guided by intrinsic rewards offers a scalable alternative, yet it suffers from opaque training dynamics and catastrophic instability, such as policy collapse and reward hacking. In this paper, we first design and evaluate a suite of intrinsic rewards that explicitly enforce concise and certain generation. Second, to discover the boundaries of this approach, we test base models across a spectrum of intrinsic reasoning capabilities, revealing how a model's foundational logical prior dictates its success or failure. Finally, to demystify why certain configurations stabilize while others collapse, we introduce a novel geometric diagnostic lens, showing that successful cases are enveloped by manifolds. Ultimately, our work goes beyond merely demonstrating that enforcing concise and certain responses successfully boosts mathematical reasoning; we reveal when this unsupervised approach breaks down and geometrically diagnose why.

Zelin Zhang, Fei Cheng, Chenhui Chu

Kyoto University

Unsupervised RLReinforcement LearningRL for Language Model ReasoningLLM ReasoningIntrinsic Reward Methods
AskPDF
Mar 17, 2026

Breaking the Chain: A Causal Analysis of LLM Faithfulness to Intermediate Structures

In schema-guided reasoning (SGR) pipelines, LLMs produce explicit intermediate structures -- rubrics, checklists, or verification queries -- before committing to a final decision. SGR is increasingly adopted because it promises controllability: practitioners expect to inspect, edit, and override these structures to steer the outcome. But does the promise hold? We introduce a causal evaluation protocol to measure it: by selecting tasks where a deterministic function maps intermediate structures to decisions, every controlled edit implies a unique correct output. Across 12 models and 4 benchmarks, models appear self-consistent with their own intermediate structures but fail to update predictions after intervention -- revealing that apparent faithfulness is fragile once the intermediate structure changes. When derivation of the final decision from the structure is delegated to an external tool, this fragility largely disappears; stronger prompting yields only limited improvements, while preference optimization substantially improves intervention faithfulness. Overall, intermediate structures in schema-guided pipelines function as influential context rather than stable causal mediators.

Oleg Somov, Mikhail Chaichuk, Gleb Ershov +4

AIRI · MIPT · HSE University +2

LLM ReasoningCausal Interventions in Language ModelsReasoning Consistency in Language Models
AskPDF
Mar 16, 2026

Mask Is What DLLM Needs: A Masked Data Training Paradigm for Diffusion LLMs

Discrete diffusion models offer global context awareness and flexible parallel generation. However, uniform random noise schedulers in standard DLLM training overlook the highly non-uniform information density inherent in real-world sequences. This wastes optimization resources on low-density structural glues while leaving high-density logical pivot points severely under-optimized. To address this, we propose an Information Density Driven Smart Noise Scheduler. By extracting information-dense hubs and applying Complementary Priority Masking, our method decouples a single training instance into mutually reinforcing reasoning and syntax samples, forcing the model to master both logical deduction and foundational sequence structure. Experiments demonstrate that our approach improves average accuracy by ~4% across four Code and Math reasoning benchmarks, significantly outperforming uniform baselines. Mechanistic analyses further reveal that probabilistic priority masking effectively mitigates contextual collapse during block diffusion training. Overall, this density-aware strategy efficiently unlocks the reasoning potential of diffusion language models at minimal annotation cost, emerging as a promising new masked data training paradigm for Diffusion LLMs. Our processed dataset can be found at https://huggingface.co/datasets/malr07/opc-sft-stage2-dense-extracted.

Linrui Ma, Yufei Cui, Kai Han +1

Noah’s Ark Lab, Huawei Montreal, Canada & Beijing, China

Diffusion Language ModelsBlockwise Diffusion Language ModelsLLM Reasoning
AskPDF
Mar 16, 2026

Can LLMs Model Incorrect Student Reasoning? A Case Study on Distractor Generation

Modeling student misconceptions in a realistic manner is critical for AI in education. In this work, we examine how large language models (LLMs) reason about misconceptions when generating distractor answers for multiple-choice questions (MCQs), a task that requires producing answers that are incorrect, yet plausible. We introduce a taxonomy over reasoning strategies for distractor generation that is grounded in learning-science literature and empirical observation, which we apply to LLM-generated reasoning traces across math and science MCQs. On the math dataset, we find that models follow a misconception-based process with potentially high diagnostic value: they recover the correct solution, articulate student errors, simulate them, and select plausible candidates. On the science dataset, on the other hand, they tend to follow a less robust approach based on semantic similarity to the correct answer. We find the most frequent failure modes to be that the model is unable to generate a correct solution or that it discards plausible distractor candidates when performing selection. Providing the correct solution in the prompt yields a relative improvement of 6.4% in alignment with human-authored distractors, highlighting the critical role of anchoring distractor generation to the correct solution. Together, our findings offer an interpretable view of how LLMs model incorrect student reasoning.

Yanick Zengaffinen, Andreas Opedal, Donya Rooein +3

1ETH Zürich · 2Bocconi University · University of Central Florida

LLM PromptingGenerative AI in EducationUser ModelingText GenerationLLM Reasoning
AskPDF
Mar 16, 2026

Understanding Reasoning in LLMs through Strategic Information Allocation under Uncertainty

LLMs often exhibit Aha moments such as self-correction after tokens like "Wait," yet the underlying mechanism remains unclear. Standard LLMs collapse mainly through silent divergence, where trajectories drift from the correct answer yet remain locally coherent, so no explicit error triggers reactive self-correction. We introduce an information-theoretic framework that separates reasoning into procedural advancement and epistemic verbalization, the token-level externalization of uncertainty, and prove that sporadic verbalization restores convergence toward the correct answer even without explicit error triggers. Empirically, a minimal doubt cue recovers failed trajectories, and small-scale SFT suffices to instill or suppress this capability, suggesting that strong reasoning hinges less on an extraordinary inner mechanism than on the linguistic habit of externalizing uncertainty. Our framework recasts reasoning as strategic information allocation under uncertainty, offering a new lens for understanding and advancing LLM reasoning.

Jeonghye Kim, Xufang Luo, Minbeom Kim +3

Microsoft Research · KAIST · Seoul National University

LLM ReasoningLLM Self-Correction
AskPDF
Mar 16, 2026

SAGE: Multi-Agent Self-Evolution for LLM Reasoning

Reinforcement learning with verifiable rewards improves reasoning in large language models (LLMs), but many methods still rely on large human-labeled datasets. While self-play reduces this dependency, it often lacks explicit planning and strong quality control, limiting stability in long-horizon multi-step reasoning. We present SAGE (Self-evolving Agents for Generalized reasoning Evolution), a closed-loop framework where four agents: Challenger, Planner, Solver, and Critic, co-evolve from a shared LLM backbone using only a small seed set. The Challenger continuously generates increasingly difficult tasks; the Planner converts each task into a structured multi-step plan; and the Solver follows the plan to produce an answer, whose correctness is determined by external verifiers. The Critic scores and filters both generated questions and plans to prevent curriculum drift and maintain training signal quality, enabling stable self-training. Across mathematics and code-generation benchmarks, SAGE delivers consistent gains across model scales, improving the Qwen-2.5-7B model by 8.9% on LiveCodeBench and 10.7% on OlympiadBench.

Yulin Peng, Xinxin Zhu, Chenxing Wei +4

College of Computer Science and Software Engineering, Shenzhen University, China · Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ), China · School of Information Technology, Carleton University, Canada

LLM ReasoningSelf-Training for Language ModelsRL for Language Model ReasoningMulti-Agent LLMsCode Generation
AskPDF
Mar 16, 2026

Why the Valuable Capabilities of LLMs Are Precisely the Unexplainable Ones

This paper proposes and argues for a counterintuitive thesis: the truly valuable capabilities of large language models (LLMs) reside precisely in the part that cannot be fully captured by human-readable discrete rules. The core argument is a proof by contradiction via expert system equivalence: if the full capabilities of an LLM could be described by a complete set of human-readable rules, then that rule set would be functionally equivalent to an expert system; but expert systems have been historically and empirically demonstrated to be strictly weaker than LLMs; therefore, a contradiction arises -- the capabilities of LLMs that exceed those of expert systems are exactly the capabilities that cannot be rule-encoded. This thesis is further supported by the Chinese philosophical concept of Wu (sudden insight through practice), the historical failure of expert systems, and a structural mismatch between human cognitive tools and complex systems. The paper discusses implications for interpretability research, AI safety, and scientific epistemology.

Quan Cheng

Tsinghua University

Explainable Artificial IntelligenceLLM InterpretabilityImplicit Reasoning in Language ModelsLLM Reasoning
AskPDF
Mar 15, 2026

MALicious INTent Dataset and Inoculating LLMs for Enhanced Disinformation Detection

The intentional creation and spread of disinformation poses a significant threat to public discourse. However, existing English datasets and research rarely address the intentionality behind the disinformation. This work presents MALINT, the first human-annotated English corpus developed in collaboration with expert fact-checkers to capture disinformation and its malicious intent. We utilize our novel corpus to benchmark 12 language models, including small language models (SLMs) such as BERT and large language models (LLMs) like Llama 3.3, on binary and multilabel intent classification tasks. Moreover, inspired by inoculation theory from psychology and communication studies, we investigate whether incorporating knowledge of malicious intent can improve disinformation detection. To this end, we propose intent-based inoculation, an intent-augmented reasoning for LLMs that integrates intent analysis to mitigate the persuasive impact of disinformation. Analysis on six disinformation datasets, five LLMs, and seven languages shows that intent-augmented reasoning improves zero-shot disinformation detection. To support research in intent-aware disinformation detection, we release the MALINT dataset with annotations from each annotation step.

Arkadiusz Modzelewski, Witold Sosnowski, Eleni Papadopulos +4

University of Padua, Italy · Polish-Japanese Academy of Information Technology, Poland · NASK National Research Institute, Poland +1

Fake News DetectionIntent ClassificationLLM ReasoningAutomated Fact-Checking
AskPDF
Mar 15, 2026

On the Nature of Attention Sink that Shapes Decoding Strategy in Omni-LLMs

The goal of this paper is to strengthen the reasoning of Omnimodal Large Language Models (Omni-LLMs) at inference time, without additional training. These models jointly process video, audio, and text, and given the large number of tokens they consume, how attention is routed across them is central to their behaviour. We focus specifically on attention sinks, tokens that absorb a disproportionate share of attention mass regardless of their semantic content, to understand how this routing unfolds. To this end, we conduct a systematic analysis of sink behaviour in Omni-LLMs. Our analysis yields two key findings: (i) high sink attention does not solely indicate head redundancy, suggesting that sink value representations play additional functional roles; (ii) the sink value vector acts as a shared bias added to every token's output, serving as a global signal that organises the representation as a whole. Building on this, we propose OutRo, which correspondingly aligns non-sink token representations with the sink in feature space, and relaxes the causal mask for sink tokens at an early layer to sharpen this bias before the rest of decoding proceeds. This design enhances the reasoning process without requiring additional forward passes or access to attention maps. Based on extensive experiments, OutRo consistently improves performance on seven video QA benchmarks and demonstrates strong generalisation, while incurring only a 1.1x decoding overhead.

Suho Yoo, Youngjoon Jang, Joon Son Chung

KAIST · VGG, University of Oxford

Attention SinksMultimodal Large Language ModelsLLM Reasoning
AskPDF
Mar 14, 2026

GRPO and Reflection Reward for Mathematical Reasoning in Large Language Models

The enhancement of reasoning capabilities in large language models (LLMs) has garnered significant attention, with supervised fine-tuning (SFT) and reinforcement learning emerging as dominant paradigms. While recent studies recognize the importance of reflection in reasoning processes, existing methodologies seldom address proactive reflection encouragement during training. This study focuses on mathematical reasoning by proposing a four-stage framework integrating Group Relative Policy Optimization (GRPO) with reflection reward mechanisms to strengthen LLMs' self-reflective capabilities. Besides, this approach incorporates established accuracy and format reward. Experimental results demonstrate GRPO's state-of-the-art performance through reflection-encouraged training, with ablation studies confirming the reflection reward's pivotal role. Comparative evaluations demonstrate full-parameter SFT's superiority over low-rank adaptation (LoRA) despite heightened computational demands. Building on these cumulative findings, this research substantiates GRPO's methodological significance in post-training optimization and envisions its potential to serve as a pivotal enabler for future LLM-based intelligent agents through the synergistic integration of cognitive rewards with dynamic environmental interactions.

Zhijie Wang

School of Computer Science and Engineering, Sun Yat-sen University(SYSU), Guangzhou, China

Self-Reflection in Language Model AgentsLLM ReasoningRL for Language Model ReasoningRL Post-TrainingMetacognition in Language ModelsGroup Relative Policy Optimization
AskPDF
Mar 14, 2026

Domain-Specialized Tree of Thought through Plug-and-Play Predictors

While Large Language Models (LLMs) have advanced complex reasoning, prominent methods like the Tree of Thoughts (ToT) framework face a critical trade-off between exploration depth and computational efficiency. Existing ToT implementations often rely on heavyweight LLM-based self-evaluation or rigid heuristics for branch pruning, making them prohibitively expensive and inflexible for broad application. To address this, we introduce DST, an adaptable, plug-and-play predictor that serves as a lightweight, supervised heuristic to guide the ToT search process. Our predictor enables dynamic, context-aware pruning, allowing the search to proceed with near-greedy efficiency on simpler reasoning steps while adaptively expanding the search beam only when encountering uncertainty or task complexity. We evaluate our approach on a diverse suite of benchmarks spanning mathematical reasoning, general reasoning, and complex logical reasoning. Experimental results demonstrate that our method achieves accuracy competitive with or superior to strong baselines, including standard ToT, while reducing computational overhead by 26-75%. Our work effectively resolves the accuracy-efficiency trade-off in tree-based reasoning, transforming ToT from a resource-intensive technique into a scalable and practical paradigm for complex problem-solving in LLMs.

Xuanqi Gao, Haoyu Wang, Jun Sun +2

Xi’an Jiaotong University, China · Singapore Management University, Singapore · University of Massachusetts at Amherst, United States

LLM ReasoningTree of ThoughtsEfficient Language Model ReasoningLLM Inference
AskPDF
Mar 14, 2026

Can We Trust LLMs on Memristors? Diving into Reasoning Ability under Non-Ideality

Memristor-based analog compute-in-memory (CIM) architectures provide a promising substrate for the efficient deployment of Large Language Models (LLMs), owing to superior energy efficiency and computational density. However, these architectures suffer from precision issues caused by intrinsic non-idealities of memristors. In this paper, we first conduct a comprehensive investigation into the impact of such typical non-idealities on LLM reasoning. Empirical results indicate that reasoning capability decreases significantly but varies for distinct benchmarks. Subsequently, we systematically appraise three training-free strategies, including thinking mode, in-context learning, and module redundancy. We thus summarize valuable guidelines, i.e., shallow layer redundancy is particularly effective for improving robustness, thinking mode performs better under low noise levels but degrades at higher noise, and in-context learning reduces output length with a slight performance trade-off. Our findings offer new insights into LLM reasoning under non-ideality and practical strategies to improve robustness.

Taiqiang Wu, Yuxin Cheng, Chenchen Ding +5

The University of Hong Kong

Language Model RobustnessLLM ReasoningCompute-in-Memory
AskPDF
Mar 13, 2026

When Right Meets Wrong: Bilateral Context Conditioning with Reward-Confidence Correction for GRPO

Group Relative Policy Optimization (GRPO) has emerged as an effective method for training reasoning models. While it computes advantages based on group mean, GRPO treats each output as an independent sample during the optimization and overlooks a vital structural signal: the natural contrast between correct and incorrect solutions within the same group, thus ignoring the rich, comparative data that could be leveraged by explicitly pitting successful reasoning traces against failed ones. To capitalize on this, we present a contrastive reformulation of GRPO, showing that the GRPO objective implicitly maximizes the margin between the policy ratios of correct and incorrect samples. Building on this insight, we propose Bilateral Context Conditioning (BICC), a mechanism that allows the model to cross-reference successful and failed reasoning traces during the optimization, enabling a direct information flow across samples. We further introduce Reward-Confidence Correction (RCC) to stabilize training by dynamically adjusts the advantage baseline in GRPO using reward-confidence covariance derived from the first-order approximation of the variance-minimizing estimator. Both mechanisms require no additional sampling or auxiliary models and can be adapted to all GRPO variants. Experiments on mathematical reasoning benchmarks demonstrate consistent improvements across comprehensive models and algorithms. Code is available at https://github.com/Skylanding/BiCC.

Yu Li, Tian Lan, Zhengling Qi

Department of Electrical and Computer Engineering, George Washington University · School of Business, George Washington University

RL for Language Model ReasoningGroup Relative Policy OptimizationLLM ReasoningContrastive RL
AskPDF
Mar 13, 2026

TDA-RC: Task-Driven Alignment for Knowledge-Based Reasoning Chains in Large Language Models

Enhancing the reasoning capability of large language models (LLMs) remains a core challenge in natural language processing. The Chain-of-Thought (CoT) paradigm dominates practical applications for its single-round efficiency, yet its reasoning chains often exhibit logical gaps. While multi-round paradigms like Graph-of-Thoughts (GoT), Tree-of-Thoughts (ToT), and Atom of Thought (AoT) achieve strong performance and reveal effective reasoning structures, their high cost limits practical use. To address this problem, this paper proposes a topology-based method for optimizing reasoning chains. The framework embeds essential topological patterns of effective reasoning into the lightweight CoT paradigm. Using persistent homology, we map CoT, ToT, and GoT into a unified topological space to quantify their structural features. On this basis, we design a unified optimization system: a Topological Optimization Agent diagnoses deviations in CoT chains from desirable topological characteristics and simultaneously generates targeted strategies to repair these structural deficiencies. Compared with multi-round reasoning methods like ToT and GoT, experiments on multiple datasets show that our approach offers a superior balance between reasoning accuracy and efficiency, showcasing a practical solution to ``single-round generation with multi-round intelligence''.

Jiaquan Zhang, Qigan Sun, Chaoning Zhang +11

School of Information and Software Engineering, University of Electronic Science and Technology of China, Chengdu, China · School of Computing, Kyung Hee University, Yongin-si, South Korea · School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China +3

Efficient Language Model ReasoningLLM ReasoningCoT Reasoning
AskPDF
Mar 13, 2026

FGTR: Fine-Grained Multi-Table Retrieval via Hierarchical LLM Reasoning

With the rapid advancement of large language models (LLMs), growing efforts have been made on LLM-based table retrieval. However, existing studies typically focus on single-table query, and implement it by similarity matching after encoding the entire table. These methods usually result in low accuracy due to their coarse-grained encoding which incorporates much query-irrelated data, and are also inefficient when dealing with large tables, failing to fully utilize the reasoning capabilities of LLM. Further, multi-table query is under-explored in retrieval tasks. To this end, we propose a hierarchical multi-table query method based on LLM: Fine-Grained Multi-Table Retrieval FGTR, a new retrieval paradigm that employs a human-like reasoning strategy. Through hierarchical reasoning, FGTR first identifies relevant schema elements and then retrieves the corresponding cell contents, ultimately constructing a concise and accurate sub-table that aligns with the given query. To comprehensively evaluate the performance of FGTR, we construct two new benchmark datasets based on Spider and BIRD . Experimental results show that FGTR outperforms previous state-of-the-art methods, improving the F_2 metric by 18% on Spider and 21% on BIRD, demonstrating its effectiveness in enhancing fine-grained retrieval and its potential to improve end-to-end performance on table-based downstream tasks.

Chaojie Sun, Bin Cao, Tiantian Li +3

Zhejiang University of Technology Hangzhou, China · University of Aberdeen Aberdeen, UK

Table RetrievalLLM Reasoning
AskPDF