AgentFlare

Scholar · chain-of-thought reasoning language models

Chain-of-Thought Reasoning: Research Digest

10 foundational papers with 2,307 combined citations, plus the latest arXiv preprints.

Foundational papers102,307 citations
Most citedPlan-and-solve prompting: Improving z…812 citations · 2023
arXiv, Sep 2026178 papers-38% vs Oct 2025
Newest preprint2026-10-27

AISA Scholar API · arXiv

arXiv submissions per monthPapers matching abs:"chain-of-thought" · last 12 complete months
monthly arXiv submissions: Oct 2025 289 → Sep 2026 178; high 307 (May 2026), low 178 (Sep 2026)168218267317Oct 2025Mar 2026Sep 2026
402 · Pay-per-Crawl$0.02

You're reading the full report — free for humans and search engines. The same content ships as a structured JSON dataset (normalized facts, sources and citation arrays, no HTML parsing) so AI agents can consume it directly — metered per fetch via the open HTTP 402 payment protocol.

⚡ Agents: fetch .data.json · $0.02 → agents: crawlerAutoPrice=true · verify at https://esa.aisa.one/api/v1/access/verify

What’s new on arXiv

A12026-10-27cs.CV

VideoLVR: Latent Visual Reasoning over Grounded Video Evidence

Shilong Li, Xingyuan Bu, Hao Shi, Yun Xing, Weixin Feng, Qiangpeng Yang, Shilei Wen

Chain-of-thought reasoning has substantially advanced video understanding in multimodal large language models (MLLMs). However, language-based reasoning traces do not provide a reliable mechanism for models to revisit relevant visual evidence, leading to visual forgetting and impairing their ability to reason over video content. Existing approaches maintain visual grounding during reasoning in two main ways: they either refer to evidence using textual coordinates, which specify symbolic locations rather than the visual content itself, or reinsert cropped regions or frames into the multimodal context, preserving visual details at the cost of substantial context expansion. To address these limitations, we introduce VideoLVR, which constructs latent visual evidence from local regions at relevant video moments, thereby providing visual support for intermediate reasoning steps. To supervise latent visual evidence generation, we introduce a visual token prediction objective that formulates visual token reconstruction as prediction over a vocabulary derived from the input video, thereby unifying visual evidence reconstruction and text generation within the same token prediction framework. This shared prediction framework enables MLLMs to reason over interleaved text tokens and latent visual states while integrating evidence from different moments. We further extend VideoLVR to reinforcement learning by treating sampled visual tokens as actions alongside text tokens and jointly optimizing the policy using final-answer rewards. By efficiently revisiting task-relevant visual information during reasoning, VideoLVR improves video understanding without substantially expanding the multimodal context. Experiments on multiple video benchmarks demonstrate its effectiveness, especially on tasks requiring fine-grained visual evidence.

A22026-10-27cs.RO

PDRVLA: A Vision-Language-Action Model with Physical Dynamics Representations

Hanqing Wang, Zhaotian Cai, Ruichuan An, Bohan Zeng, Tao Zhang, Yilin Chen, Xu Jiang, Kaiquan Chen, Chongyang Wu, Haoran Li, Fan Yang, Ming…

Vision--Language--Action (VLA) models generate robot actions from visual observations and language instructions, but their understanding of physical dynamics remains limited. Video world models can capture scene dynamics, yet generating future videos introduces high latency into the control loop. We introduce PDRVLA, whose lightweight Physical Dynamics Module extracts physical tokens from vision--language features to condition action generation without online video generation. The module learns through direct prediction of future RGB, depth, and optical-flow features, with auxiliary supervision through a frozen pretrained video world model. Training relies entirely on publicly available data, expanded through human-to-robot synthesis, subtask annotation, and embodied chain-of-thought (ECoT) annotation. Across four simulation benchmarks, PDRVLA achieves success rates of 99.3% on LIBERO, 89.9% on LIBERO-Plus, 93.6% on RoboTwin 2.0, and 81.5% on SimplerEnv. Ablations show improved future prediction with world-model supervision and evaluate how subtask conditioning, future object-pose supervision, and ECoT training affect action success.

A32026-10-27cs.AI

Correctness Conditioning as Selection in Chain-of-Thought Faithfulness Evaluation

Yueer Pan, Yuexi Pan

Restricting chain-of-thought (CoT) faithfulness evaluation to correctly answered instances changes the population being measured. We study this selection step in a seven-variable structural causal model and derive the exact observed gap Δ= Cov(A, C)/P(C=1). The model distinguishes difficulty confounding from shortcut-induced collider bias and states the conditional-independence assumption needed to express the gap through latent response propensities. A protocol review of 18 papers distinguishes answer-correctness restrictions from agreement-based selection, benchmark construction, and training-data filtering. We evaluate four sub-ceiling deployments on arithmetic tasks and observe positive selection gaps in all four. The primary GLM-5.3 run yields a +7.4 percentage-point gap (n = 1000, p < 0.0001), and pooled Cochran-Mantel-Haenszel analysis gives OR = 1.70 (p < 10^-6). An assumption-free covariance decomposition assigns 42.8% of the GLM-5.3 gap to between-stratum variation in interventionable-operation count, leaving the majority unattributed. When counterfactual compliance is observable for every instance, evaluating all traces directly estimates population faithfulness without correctness filtering.

A42026-10-27cs.IR

Latent Residual Reasoning for Ranking in Generative Recommendation

Haihan Shi, Ruidong Han, Wenjie Wang, Shangyu Chen, Sunhao Dai, Yongqi Li, Yangyang Li, Fei Jiang, Wei Lin, Fuli Feng, Xiangnan He

Generative recommendation (GR) models for ranking typically adopt a fast-thinking direct prediction paradigm with strong practical efficiency. However, ranking requires fine-grained scoring across multiple candidate items, while existing GR models still mainly rely on one-shot prediction. This raises the question whether additional inference-time computation can further improve ranking quality. Recently, inspired by Chain-of-Thought (CoT) reasoning in large language models (LLMs), some work has begun to introduce reasoning into recommendation systems. Some of these methods rely on LLM reasoning frameworks to perform explicit reasoning in textual space, but such designs are not suitable for more widely deployed non-textual recommendation models. A more relevant direction is to introduce latent reasoning into ID-based recommendation models. Although recent work has extended latent reasoning to industrial ranking, how successive reasoning steps should improve fine-grained candidate scores remains less explored. To address this problem, we propose Refiner, a latent residual reasoning framework with parameter-efficient fine-tuning for generative ranking. Our approach introduces three key components: (1) a residual reasoning mechanism that progressively refines ranking scores through cumulative additive corrections, (2) history queries that provide condensed user behavior representations to enhance candidate-specific reasoning, and (3) a parameter-efficient fine-tuning strategy that leverages LoRA modules to adapt pre-trained models for latent reasoning. Extensive experiments on two public benchmarks, an offline industrial dataset, and an online A/B test demonstrate that Refiner consistently improves over direct prediction, validating its effectiveness and practical utility.

A52026-10-27cs.CL

Is Token Confidence Just Chain Length? A Signal-Processing Audit of Reasoning Traces

Pragna Mallappa

Reasoning language models answer math problems by writing long chains of thought, and sampling many chains per problem and taking a majority vote raises accuracy at a large cost in generated text. A popular way to cut that cost scores each chain by the model's confidence in each token it writes, where a token is a word or part of a word. The method averages this confidence over a sliding window of 2048 tokens and keeps or stops chains based on the lowest average. We treat the confidence, recorded token by token, as a signal in the sense of audio or radio, and examine the method with standard signal-processing tools such as spectral analysis and statistical tests for detecting a sudden change. We sampled 16 chains for each of 500 problems from two math benchmarks with a small open reasoning model. The windowed-minimum score turns out to measure mostly how long a chain is. Ranking chains by length alone does as well as or better than the confidence score, and once length is accounted for the score carries no information about correctness. Signal processing explains why: the minimum of a longer stretch of a slowly drifting signal is lower by chance alone, and the confidence signal drifts slowly enough that this effect dominates. The spectrum does hold a small signal that length does not explain, since incorrect chains carry more rapid token-to-token fluctuation, but it is too weak to improve the vote. For stopping chains early, neither the sliding average nor a classical change detector beats the simpler choice of sampling fewer chains.

A62026-10-26cs.CL

EXCISE: Locating and Excising Distracting Sentences in Medical Question Answering

Yerim Hong, Sohyun Son, Heeseung Chung, Mujeen Sung

Language models are increasingly considered for clinical decision support, where their answers should be grounded in clinical evidence and remain robust to irrelevant information. Prior work shows that a single irrelevant sentence can lower the accuracy of medical models, but not where inside the model its effect can be removed, or how to do so without knowing which sentence it is. We find that blocking attention to a known distractor restores the answer, whereas head masking can suppress correct answers. We also show that the model's hidden states mark the distractor as the sentence that least fits the case even when the model follows it. To remove a distractor's effect without knowing which sentence it is, we propose EXCISE, an inference-time method that blocks the sentence most influencing the answer, but only if the model's hidden states mark it as inconsistent with the case. On MedQA-TD (MedQA with Targeted Distractors), a benchmark we construct in which each distractor targets a specific wrong option, EXCISE gains 16.7-22.8 accuracy points at a 1.5-2.1-point cost on the same questions without distractors, outperforming prompting, rewriting, and steering baselines for all three models. Its gains extend to MedDistractQA and BiasMedQA, as well as chain-of-thought generation.

A72026-10-26cs.LG

S3AC: Stabilized, Selective, and Scalable Actor-Critic for LLM Reasoning

Shuo Shen, Guangran Cheng, Chengqi Lyu, Xiangsen Wang, Yanan Sun, Wenwei Zhang

Critic-free methods are widely used in reinforcement learning with verifiable rewards (RLVR) for long chain-of-thought (CoT) reasoning, avoiding difficult and costly critic training. Common alternatives lack token-level credit assignment and typically rely on grouped rollouts to estimate prompt-specific baselines. A critic can address both limitations, but long-CoT critic training is hindered by insufficient coverage of trajectories with diverse lengths and early gradients dominated by the newly initialized value head. We introduce S3AC (Stabilized, Selective, and Scalable Actor-Critic) for LLM reasoning, built around a practical critic-training recipe. The trained critic provides token-specific advantages and prompt-specific baselines. Using the prompt-level baseline to estimate difficulty, we propose Difficulty-Aware Prompt Sampling (DAPS). DAPS combines this estimate with outcome history to select prompts before generation, reducing rollout waste without offline difficulty initialization. S3AC improves group-rollout performance across dense and mixture-of-experts models from 1.5B to 397B parameters, outperforming a modern GRPO variant (GRPO-X). At 35B, it sustains single-rollout training while tested critic-free baselines deteriorate after initial gains.

A82026-10-26cs.CL

When the Persona Contradicts the Line: Diagnosing Plan-to-Synthesis Failures in Chain-of-Thought TTS

Joonyong Park, Jerry Li

Chain-of-thought TTS systems rely on language-model planners to translate persona and context into vocal delivery, but it is unclear where this intent is lost when the requested affect conflicts with lexical sentiment. We introduce a bilingual Japanese--English persona--line conflict benchmark in which each item pairs one line with congruent and conflicting personas and pre-declares the expected acoustic direction. We probe this direction in the planner-generated delivery phrase, the planner-written emotion vector, the QwenEmotion-generated vector, and the rendered waveform. Planner outputs are usually direction-correct, while downstream agreement can be substantially lower. Bypassing QwenEmotion provides only modest gains over the full planner-controlled Base. Supervising QwenEmotion raises the direction rate to 75% in Japanese and 84% in English. A learned text-to-latent interface reaches the highest English rate of 89% and the highest Japanese arousal rate of 88%. These results show that learned repair of the plan-to-synthesis interface can recover counter-textual delivery, although its effect differs across languages and acoustic axes.

Foundational papers

  1. 2023
    Plan-and-solve prompting: Improving zero-shot chain-of-thought reasoning by large language models812 citations
  2. 2023
    Teaching small language models to reason565 citations
  3. 2023
    R3 prompting: review, rephrase and resolve for chain-of-thought reasoning in large language models under noisy context15 citations
  4. 2024
    Enhancing zero-shot chain-of-thought reasoning in large language models through logic114 citations
  5. 2025
    To cot or not to cot? chain-of-thought helps mainly on math and symbolic reasoning418 citations
  6. 2025
    Safechain: Safety of language models with long chain-of-thought reasoning capabilities197 citations
  7. 2026
    Long chain-of-thought reasoning across languages22 citations
  8. 2026
    Not All LLM Reasoning is Visible in the Chain-of-Thought6 citations
  9. 2026-10
    Latest on arXiv: VideoLVR: Latent Visual Reasoning over Grounded Video Evidencecs.CV
Citations per paperCitation counts via the AISA Scholar API
citations: R1 Plan-and-solve prompt… 812, R2 Advancing reasoning i… 121, R3 Interactive reasoning… 37, R4 Enhancing zero-shot c… 114, R5 Teaching small langua… 565, R6 To cot or not to cot?… 418, R7 Safechain: Safety of … 197, R8 R3 prompting: review,… 15, R9 Long chain-of-thought… 22, R10 Not All LLM Reasoning… 6R1 Plan-and-solve prompt…812R2 Advancing reasoning i…121R3 Interactive reasoning…37R4 Enhancing zero-shot c…114R5 Teaching small langua…565R6 To cot or not to cot?…418R7 Safechain: Safety of …197R8 R3 prompting: review,…15R9 Long chain-of-thought…22R10 Not All LLM Reasoning…6
R52023565 citations

Teaching small language models to reason

LC Magister, J Mallinson, J Adamek et al.

… a student model on the chain of thought outputs generated by a larger teacher model. Our … We also finetune on the reasoning steps provided by the datasets. We report the accuracy …

R9202622 citations

Long chain-of-thought reasoning across languages

J Barua, S Eisape, K Yin, A Suhr

… Reasoning has emerged as a central focus of large language model (LLM) research in recent years. By scaling inference-time compute to generate chains-of-thought before answering …

R1020266 citations

Not All LLM Reasoning is Visible in the Chain-of-Thought

V Baherwani, T Goldstein, A Panda

… We evaluate 13 frontier language models across three tasks and find that many … While chain-of-thought reasoning improves the problem solving capabilities of language models, it is …

How this page is made. Foundational papers and citation counts come from academic search via the AISA Scholar API, with abstracts read from each paper’s page; recent preprints and monthly volumes come from the public arXiv API (metadata CC0). The digest is written by a language model from those abstracts only — every theme links the papers it draws on, and figures not in the inputs are removed.

Sources & citations

  1. Plan-and-solve prompting: Improving zero-shot chain-of-thought reasoning by large language models 812 citations
  2. Advancing reasoning in large language models: Promising methods and approaches 121 citations
  3. Interactive reasoning: Visualizing and controlling chain-of-thought reasoning in large language models 37 citations
  4. Enhancing zero-shot chain-of-thought reasoning in large language models through logic 114 citations
  5. Teaching small language models to reason 565 citations
  6. To cot or not to cot? chain-of-thought helps mainly on math and symbolic reasoning 418 citations
  7. Safechain: Safety of language models with long chain-of-thought reasoning capabilities 197 citations
  8. R3 prompting: review, rephrase and resolve for chain-of-thought reasoning in large language models under noisy context 15 citations
  9. Long chain-of-thought reasoning across languages 22 citations
  10. Not All LLM Reasoning is Visible in the Chain-of-Thought 6 citations
  11. VideoLVR: Latent Visual Reasoning over Grounded Video Evidence arXiv · 2026-10-27
  12. PDRVLA: A Vision-Language-Action Model with Physical Dynamics Representations arXiv · 2026-10-27
  13. Correctness Conditioning as Selection in Chain-of-Thought Faithfulness Evaluation arXiv · 2026-10-27
  14. Latent Residual Reasoning for Ranking in Generative Recommendation arXiv · 2026-10-27
  15. Is Token Confidence Just Chain Length? A Signal-Processing Audit of Reasoning Traces arXiv · 2026-10-27
  16. EXCISE: Locating and Excising Distracting Sentences in Medical Question Answering arXiv · 2026-10-26
  17. S3AC: Stabilized, Selective, and Scalable Actor-Critic for LLM Reasoning arXiv · 2026-10-26
  18. When the Persona Contradicts the Line: Diagnosing Plan-to-Synthesis Failures in Chain-of-Thought TTS arXiv · 2026-10-26

Related