Datasets:

fineset-io
/

test-time-compute-papers

id string	sources list	title string	abstract string	authors list	categories list	fields_of_study list	published_date timestamp[s]	url string	pdf_url string	arxiv_id string	doi string	citation_count int64	influential_citation_count int64	has_code bool	code_url string	venue string	quality_score float64
2890c16b240a2a42e5dbe3574416937ba6a7186999f616397837f13766c96021	[ "arxiv" ]	Universal Image Restoration via Internalized Chain-of-Thought Reasoning	Image restoration seeks to recover high-quality images from degraded inputs but becomes highly ill-posed under complex, mixed degradations. While unified all-in-one models are common, their performance declines as degradation complexity increases. Recent works adopt Chain-of-Thought (CoT) reasoning for multi-round rest...	[ "Yu Guo", "Zhengru Fang", "Shengfeng He", "Senkang Hu", "Yihang Tao", "Phone Lin", "Yuguang Fang" ]	[ "cs.CV" ]	[]	2026-06-16T00:00:00	https://arxiv.org/abs/2606.17557	https://arxiv.org/pdf/2606.17557v1	2606.17557	null	0	0	true	https://github.com/gy65896/CoTIR	null	0.65
aa10d74c9c9366ec63ba58301500ee40162177e78d4d18b48b8db82a0e32da4e	[ "arxiv", "semantic_scholar" ]	Latent Thought Flow: Efficient Latent Reasoning in Large Language Models	Large Language Models (LLMs) increasingly rely on intermediate reasoning, yet explicit Chain-of-Thought (CoT) suffers from a linguistic space bottleneck: each thought must be decoded into tokens, causing high inference overhead. Latent reasoning moves deliberation into continuous space, but existing methods mostly lear...	[ "Xiandong Zou", "Jing Huang", "Jianshu Li", "Pan Zhou" ]	[ "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-06-15T00:00:00	https://arxiv.org/abs/2606.16222	https://arxiv.org/pdf/2606.16222v1	2606.16222	null	0	0	false	null	null	0.35
b8bede8aea3a4641efd49163e98de775abc49de3360427548b5df5eac099c686	[ "arxiv", "semantic_scholar" ]	RoboPIN: Grounded Embodied Reasoning via Pinned Chain-of-Thought	Embodied reasoning requires models to perceive task-relevant objects and spaces in physical environments and maintain consistent visual grounding throughout multi-step reasoning. However, current vision-language models rely on text-only or coordinate-augmented chain-of-thought, where entity references remain implicit a...	[ "Yaoting Huang", "Yifu Yuan", "Linqi Han", "Chengwen Li", "Shuoheng Zhang", "Xianze Yao", "Hongyao Tang", "Yan Zheng", "Jianye Hao" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-14T00:00:00	https://arxiv.org/abs/2606.15753	https://arxiv.org/pdf/2606.15753v1	2606.15753	null	0	0	true	null	null	0.65
78e6d85cf4540c86bcccc2db06d052f84252c174e28c72fd7549d3d190dd9830	[ "arxiv", "semantic_scholar" ]	Beyond the Commitment Boundary: Probing Epiphenomenal Chain-of-Thought in Large Reasoning Models	Chain-of-thought (CoT) reasoning is the dominant paradigm for inference-time scaling in language models, yet the causal influence of individual steps on the final answer poorly understood. We estimate each step's causal importance via early exit and use this measure to study how answers form across the reasoning traces...	[ "Daniel Scalena", "Sara Candussio", "Luca Bortolussi", "Elisabetta Fersini", "Malvina Nissim", "Gabriele Sarti" ]	[ "cs.LG", "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-06-11T00:00:00	https://arxiv.org/abs/2606.13603	https://arxiv.org/pdf/2606.13603v1	2606.13603	null	0	0	false	null	null	0.35
4a1b4dea4dbec85e251a1e996f085f8fd2de1cebee214d614fb5b2825b278221	[ "arxiv", "semantic_scholar" ]	AVIS: Adaptive Test-Time Scaling for Vision-Language Models	Modern Vision-Language Models (VLMs) benefit from chain-of-thought prompting and test-time scaling, but these gains often come with prohibitive inference cost due to large visual contexts and long decoding chains. We view this cost through two coupled axes: Visual Context Scaling (VCS), which controls how much visual e...	[ "Ahmadreza Jeddi", "Minh Ngoc Le", "Amirhossein Kazerouni", "Hakki Can Karaimer", "Hue Nguyen", "Iqbal Mohomed", "Michael Brudno", "Alex Levinshtein", "Konstantinos G. Derpanis", "Babak Taati", "Radek Grzeszczuk" ]	[ "cs.CV", "cs.AI" ]	[ "Computer Science" ]	2026-06-10T00:00:00	https://arxiv.org/abs/2606.11576	https://arxiv.org/pdf/2606.11576v1	2606.11576	null	0	0	false	null	null	0.35
aa3334c4d32fd7c1cc7ccc544cfa0e5a1a255207a16c4eb8e1ca08c33928a946	[ "arxiv", "semantic_scholar" ]	TVI-CoT: Text-Visual Interleaved Chain-of-Thought Reasoning for Multimodal Understanding	Chain-of-thought (CoT) reasoning has proven effective for enhancing problem-solving in large language models. However, when applied to multimodal LLMs (MLLMs), existing CoT approaches suffer from a fundamental limitation: they perform reasoning entirely in text without accessing visual features during the reasoning pro...	[ "Lianyu Hu", "Xiaoyu Ma", "Zeqin Liao", "Yang Liu" ]	[ "cs.CV" ]	[ "Computer Science" ]	2026-06-07T00:00:00	https://arxiv.org/abs/2606.08464	https://arxiv.org/pdf/2606.08464v1	2606.08464	null	0	0	true	https://github.com/hulianyuyy/TVI-CoT	null	0.65
ed6d36c1b4ca3432beff5959ef1c752cc688989dc5485d27e74b69585769a9a5	[ "arxiv", "semantic_scholar" ]	ThinkBooster: A Unified Framework for Seamless Test-Time Scaling of LLM Reasoning	Test-time compute (TTC) scaling has emerged as a powerful paradigm for improving large language model (LLM) reasoning by allocating additional compute during inference, e.g., via multi-sample generation and verifier-based reranking. Existing TTC scaling strategies and reasoning scorers remain fragmented, evaluated unde...	[ "Vladislav Smirnov", "Chieu Nguyen", "Sergey Senichev", "Minh Ngoc Ta", "Ekaterina Fadeeva", "Artem Vazhentsev", "Daria Galimzianova", "Nikolai Rozanov", "Viktor Mazanov", "Jingwei Ni", "Tianyi Wu", "Igor Kiselev", "Mrinmaya Sachan", "Iryna Gurevych", "Preslav Nakov", "Timothy Baldwin"...	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-06-05T00:00:00	https://arxiv.org/abs/2606.06915	https://arxiv.org/pdf/2606.06915v2	2606.06915	null	0	0	true	null	null	0.65
b29ed21a15ec1c1081dac4ebf79b2a09aba73dafef9e52d43d37eb5f23716726	[ "arxiv", "semantic_scholar" ]	VTI-CoT: Visual-Textual Interleaved Chain of Thought for Video Reasoning	Video reasoning aims to understand complex temporal events and causal relationships within videos. Recently, Chain-of-Thought (CoT) has been introduced to this field to enhance reasoning accuracy. However, existing CoT-based video reasoning methods primarily rely on text-only information for logical deduction, overlook...	[ "Shufan Zhang", "Ziyue Lin", "Bairun Wang", "Lei Jin", "Xuanding Ding", "Xinzhu Ma", "Kunlin Yang" ]	[ "cs.CV" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.05736	https://arxiv.org/pdf/2606.05736v1	2606.05736	null	0	0	false	null	null	0.35
604514446b2f6a04aa58fff8b494679bdbe18cd5dfa9b5dcfc5d034cffd4b1f3	[ "arxiv", "semantic_scholar" ]	Closing the Loop on Latent Reasoning via Test-Time Reconstruction	Recent work moves intermediate reasoning from natural-language traces into latent or cache-level representations to reduce token overhead and avoid a discrete communication bottleneck. However, this shift also removes a key advantage of textual reasoning: intermediate states are no longer inspectable, making it difficu...	[ "Xiaopeng Yuan", "Haibo Jin", "Ye Yu", "Peng Kuang", "Lijun Yu", "Yushun Dong", "Haohan Wang" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-06-04T00:00:00	https://arxiv.org/abs/2606.06252	https://arxiv.org/pdf/2606.06252v1	2606.06252	null	0	0	false	null	null	0.35
bffd920975cdc13173cc31903a03cdaf9c70d1ae6cd1e448d6f336c287cbdceb	[ "arxiv", "semantic_scholar" ]	Test-Time Compute Scaling for ASR with Depth-Conditioned Looped Transformers	End-to-end ASR systems typically use fixed-depth acoustic encoders at inference, making it difficult to trade additional test-time computation for improved recognition without training a larger model. A natural approach is to reuse a shared Transformer block recurrently, but we find that naive looping does not fully ex...	[ "Yacouba Kaloga", "Shashi Kumar", "Shakeel A. Sheikh", "Driss Khalil", "Petr Motlicek", "Ina Kodrasi" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-06-03T00:00:00	https://arxiv.org/abs/2606.04678	https://arxiv.org/pdf/2606.04678v2	2606.04678	null	0	0	false	null	null	0.35
14df81b06c0375a689285c596d4ad587399121fd928d5e1bfd23c95de1541d12	[ "arxiv", "semantic_scholar" ]	HybridThinker: Efficient Chain-of-Thought Reasoning via Compressed Memory and Transient Thought Steps	Extended chain-of-thought (CoT) traces improve LLM reasoning but incur substantial computational and memory costs. While existing CoT compression methods mitigate this by condensing thought steps into compact representations via memory tokens and retaining only these representations at inference time, the loss of fine-...	[ "Xin Liu", "Runsong Zhao", "Xinyu Liu", "Junhao Ruan", "Pengcheng Huang", "Shichao Dong", "Chunyang Xiao", "Chenglong Wang", "Changliang Li", "Jingbo Zhu", "Tong Xiao" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.03768	https://arxiv.org/pdf/2606.03768v1	2606.03768	null	0	0	false	null	null	0.35
3334d58c7548bd50708ee71a66f46205f0fec1bc22f88e899b0da4ac4f9c6d0a	[ "arxiv", "semantic_scholar" ]	Agentic Chain-of-Thought Steering for Efficient and Controllable LLM Reasoning	Large language models improve final-answer accuracy through extended chain-of-thought reasoning, but often spend tokens inefficiently and offer little inference-time control. Existing efficient reasoning methods control thinking length by shortening, early-stopping, or compressing traces, leaving how the model thinks i...	[ "Yu Xia", "Zhouhang Xie", "Xin Xu", "Byungkyu Kang", "Prarit Lamba", "Xiang Gao", "Julian McAuley" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-06-02T00:00:00	https://arxiv.org/abs/2606.03965	https://arxiv.org/pdf/2606.03965v1	2606.03965	null	0	0	true	https://github.com/Andree-9/ACTS	null	0.65
877194799d6a291a62d5389c2083cb61ff34c39196b810133a44238a05873085	[ "arxiv", "semantic_scholar" ]	Unveiling the Entropy Dynamics of Chain-of-Thought Reasoning	This paper investigates the entropy dynamics of Chain-of-Thought (CoT) and uncovers a consistent two-phase structure: an Uncertainty Region of exploration transitioning sharply to a Confidence Region of convergence. We demonstrate that the Confidence Region possesses two critical properties: 1) High Reliability -- answ...	[ "Ting Xu", "Xu He", "Yupu Lu", "Jiankai Sun", "Dong Li", "Wai Lam", "Jianye Hao" ]	[ "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-06-01T00:00:00	https://arxiv.org/abs/2606.02020	https://arxiv.org/pdf/2606.02020v1	2606.02020	null	0	0	false	null	null	0.35
4b2836c5545119e8dce572ec418d7525212188a54dde882d7e75f61f888d0e1c	[ "arxiv", "semantic_scholar" ]	Spectral-Progressive Thought Flow for Lightweight Multimodal Reasoning	Multimodal spatial reasoning often relies on long chains of intermediate textual and visual thoughts, where accumulating visual tokens and dense cross-modal attention incur substantial computation and memory overhead. To address this challenge, we propose Spectral-Progressive Thought Flow (SpecFlow), a novel lightweigh...	[ "Yixian Shen", "Zhiheng Yang", "Qi Bi", "Changshuo Wang", "Shuai Wang", "Jia-Hong Huang", "George Floros", "Prayag Tiwari", "Anuj Pathania" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-06-01T00:00:00	https://arxiv.org/abs/2606.02842	https://arxiv.org/pdf/2606.02842v1	2606.02842	null	0	0	false	null	null	0.35
c33f5f383762c8669c600081a7a157e4f1280bbad20e52bdc64954473f2ef74b	[ "arxiv", "semantic_scholar" ]	Echoes within the Reasoning: Stealthy and Effective Watermarking via Chain of Thought	Large Language Models with Chain-of-Thought reasoning capabilities represent valuable intellectual property, yet existing black-box watermarking methods often trade robustness for reasoning fidelity by perturbing final answers or relying on fragile trigger patterns. We propose BiCoT, a watermarking framework that embed...	[ "Jiacheng Lu", "Yiming Li", "Tao Song", "Weijian Wang", "Wenjie Qu", "Haibing Guan", "Jiaheng Zhang" ]	[ "cs.CR", "cs.LG" ]	[ "Computer Science" ]	2026-05-27T00:00:00	https://arxiv.org/abs/2605.28890	https://arxiv.org/pdf/2605.28890v1	2605.28890	null	0	0	false	null	null	0.35
c2d64e9668a6a5023bd78f4679b2644a212d8b6cdf13d420c3a46a00b6644901	[ "arxiv", "semantic_scholar" ]	Boosting Inference with Guided Reasoning: Stochastic Exploration for Recursive Models	Recent work on recursive architectures has shown that tiny neural networks can be surprisingly powerful on structured reasoning tasks. The trick is to model reasoning trajectories with a latent dynamical system. We argue that the inference-time behaviour of these architectures is best understood as approximate inferenc...	[ "Andrew Corbett", "Archit Sood", "Anna Tzatzopoulou", "Sai-Aakash Ramesh", "Tim Dodwell" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-24T00:00:00	https://arxiv.org/abs/2605.25230	https://arxiv.org/pdf/2605.25230v2	2605.25230	null	0	0	false	null	null	0.35
16fb0e7b90437465881ebb2faa0330f209a148002c2992b4225f6aed1fed93e4	[ "arxiv", "semantic_scholar" ]	Pause and Reflect: Conformal Aggregation for Chain-of-Thought Reasoning	Chain-of-thought (CoT) reasoning with self-consistency improves performance by aggregating multiple sampled reasoning paths. In this setting, correctness is no longer tied to a single reasoning trace but to the aggregation rule over a pool of candidate paths, making aggregation uncertainty the central challenge. This i...	[ "Yu Gu", "Zijun Yu", "Vahid Partovi Nia", "Masoud Asgharian" ]	[ "stat.ML", "cs.CL", "cs.LG" ]	[ "Mathematics", "Computer Science" ]	2026-05-13T00:00:00	https://arxiv.org/abs/2605.14098	https://arxiv.org/pdf/2605.14098v1	2605.14098	null	0	0	true	null	null	0.65
420073b4bdbb37ead37537d90ff681b7b79452fbf14c0877a11b1170a485f8c7	[ "arxiv", "semantic_scholar" ]	Drop the Act: Probe-Filtered RL for Faithful Chain-of-Thought Reasoning	Reasoning models post-hoc rationalize answers they have already committed to internally, producing chains of reasoning theater: deliberative-looking steps that contribute nothing to correctness. This wastes inference tokens, pollutes interpretability, and obscures what the model actually computed. We introduce **ProF...	[ "Swapnil Parekh" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-05-12T00:00:00	https://arxiv.org/abs/2605.11467	https://arxiv.org/pdf/2605.11467v1	2605.11467	null	0	0	false	null	null	0.35
a67c077356fcb5fc93431e9d3285cbf424040872c4410515229a86ea512153e2	[ "arxiv", "semantic_scholar" ]	TMAS: Scaling Test-Time Compute via Multi-Agent Synergy	Test-time scaling has become an effective paradigm for improving the reasoning ability of large language models by allocating additional computation during inference. Recent structured approaches have further advanced this paradigm by organizing inference across multiple trajectories, refinement rounds, and verificatio...	[ "George Wu", "Nan Jing", "Qing Yi", "Chuan Hao", "Ming Yang", "Feng Chang", "Yuan Wei", "Jian Yang", "Ran Tao", "Bryan Dai" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-11T00:00:00	https://arxiv.org/abs/2605.10344	https://arxiv.org/pdf/2605.10344v2	2605.10344	null	0	0	true	https://github.com/IQuestLab/tmas	null	0.65
e3060e2c4073895799c97c486c0eefeb178b6dd640de0672ae62ce0c9661c674	[ "arxiv", "semantic_scholar" ]	Latent Chain-of-Thought Improves Structured-Data Transformers	Chain-of-thought and more broadly test-time compute are known to augment the expressive capabilities of language models and have led to major innovations in reasoning. Motivated by this success, this paper explores latent chain-of-thought as well as the impact of depth and looping for time-series and tabular data. We p...	[ "Carson Dudley", "Samet Oymak" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-05-11T00:00:00	https://arxiv.org/abs/2605.11262	https://arxiv.org/pdf/2605.11262v2	2605.11262	null	0	0	true	null	null	0.65
0892fb3b57067ca316c2839e47243fc786bd9064f3689608064faefa9f258e80	[ "arxiv", "semantic_scholar" ]	Rethinking Dense Sequential Chains: Reasoning Language Models Can Extract Answers from Sparse, Order-Shuffling Chain-of-Thoughts	Modern reasoning language models generate dense, sequential chain-of-thought traces implicitly assuming that every token contributes and that steps must be consumed in order. We challenge both assumptions through a systematic intervention pipeline--removal, masking, shuffling, and noise injection--applied to model-gene...	[ "Yi-Chang Chen", "Feng-Ting Liao", "Da-shan Shiu", "Hung-yi Lee" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-05-08T00:00:00	https://arxiv.org/abs/2605.07307	https://arxiv.org/pdf/2605.07307v1	2605.07307	null	1	0	false	null	null	0.35
0a0a80a69160bdc2ced10db021442478c4a639fbd174b5ac4d63b5bb17795f8e	[ "arxiv", "semantic_scholar" ]	Multi-Agent Reasoning Improves Compute Efficiency: Pareto-Optimal Test-Time Scaling	Advances in inference methods have enabled language models to improve their predictions without additional training. These methods often prioritize raw performance over cost-effective compute usage. However, computational efficiency is key for real-world applications with resource constraints. We provide a systematic a...	[ "Florian Valentin Wunderlich", "Lars Benedikt Kaesberg", "Jan Philip Wahle", "Terry Ruas", "Bela Gipp" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-05-02T00:00:00	https://arxiv.org/abs/2605.01566	https://arxiv.org/pdf/2605.01566v1	2605.01566	null	0	0	false	null	null	0.35
54fbf8078843e7f03d3f2db7716cf9ca732718f908a9a3cdb90d8f8fc7ee1202	[ "arxiv", "semantic_scholar" ]	Compute Aligned Training: Optimizing for Test Time Inference	Scaling test-time compute has emerged as a powerful mechanism for enhancing Large Language Model (LLM) performance. However, standard post-training paradigms, Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), optimize the likelihood of individual samples under a base policy, creating a misalignment with tes...	[ "Adam Ousherovitch", "Ambuj Tewari" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-04-27T00:00:00	https://arxiv.org/abs/2604.24957	https://arxiv.org/pdf/2604.24957v2	2604.24957	10.48550/arXiv.2604.24957	0	0	false	null	arXiv.org	0.55
10e315a723d507675407e7cabe22dca1f1ce2b3425fe98cf76442f6089c7bd8e	[ "arxiv", "semantic_scholar" ]	Thoughts-as-Planning: Latent World Models for Chain-of-Thoughts Optimization via Reinforcement Planning	The success of large language models (LLMs) across diverse NLP tasks has elevated the importance of reasoning chain optimization as a critical step in aligning model behavior with task objectives. Existing reasoning chain tuning methods often rely on black-box heuristics or gradient-free search, which lack interpretabi...	[ "Dong Liu", "Yanxuan Yu", "Ying Nian Wu" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science", "Biology" ]	2026-04-27T00:00:00	https://arxiv.org/abs/2605.28842	https://arxiv.org/pdf/2605.28842v1	2605.28842	10.64898/2026.05.10.724161	0	0	true	https://github.com/FastLM/Thoughts-as-Planning	bioRxiv	0.85
7920ca456a052177f62b072204a767269b1a6d90d416b483311c63f74182c9db	[ "arxiv", "semantic_scholar" ]	Ulterior Motives: Detecting Misaligned Reasoning in Continuous Thought Models	Chain-of-Thought (CoT) reasoning has emerged as a key technique for eliciting complex reasoning in Large Language Models (LLMs). Although interpretable, its dependence on natural language limits the model's expressive bandwidth. Continuous thought models address this bottleneck by reasoning in latent space rather than ...	[ "Sharan Ramjee" ]	[ "cs.AI", "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-04-25T00:00:00	https://arxiv.org/abs/2604.23460	https://arxiv.org/pdf/2604.23460v1	2604.23460	10.48550/arXiv.2604.23460	0	0	false	null	arXiv.org	0.55
1d31ea5bdbcef31e58c0aee61b18fd0bc4da1c484108e712eafd8e644c06bf77	[ "arxiv", "semantic_scholar" ]	CAP-CoT: Cycle Adversarial Prompt for Improving Chain of Thoughts in LLM Reasoning	Chain-of-Thought (CoT) prompting has emerged as a simple and effective way to elicit step-by-step solutions from large language models (LLMs). However, CoT reasoning can be unstable across runs on long, multi-step problems, leading to inconsistent answers for unchanged task. Most prior work focuses on improving the for...	[ "Shuxu Chen", "Yitian Zhou", "Jiaquan Zhang", "Haoyu Bian", "Aming Wu", "Sungyoung Lee", "Chaoning Zhang", "Hyundong Shin" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-04-25T00:00:00	https://arxiv.org/abs/2604.23270	https://arxiv.org/pdf/2604.23270v1	2604.23270	10.48550/arXiv.2604.23270	0	0	false	null	arXiv.org	0.55
51dd73db0adae36378ed9798308592414f75c1bec937fa92c3464d09f5e10717	[ "arxiv", "semantic_scholar" ]	Calibration Drift Under Reasoning: How Chain-of-Thought Budgets Induce Overconfidence in Large Language Models	The ability of large language models (LLMs) to express calibrated uncertainty is important for safe deployment. Chain-of-thought (CoT) reasoning is widely used to improve accuracy and reliability, but its effect on calibration is not fully understood. We show that this picture is incomplete: in some settings, increasin...	[ "Prakul Sunil Hiremath", "Harshit R. Hiremath" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-04-24T00:00:00	https://arxiv.org/abs/2606.11211	https://arxiv.org/pdf/2606.11211v1	2606.11211	null	0	0	false	null	null	0.35
08b537b2ae446c36e9101264f6c00b3020b852889a24ab5ff84989581896e72f	[ "arxiv", "semantic_scholar" ]	Thinking Without Words: Efficient Latent Reasoning with Abstract Chain-of-Thought	While long, explicit chains-of-thought (CoT) have proven effective on complex reasoning tasks, they are costly to generate during inference. Non-verbal reasoning methods have emerged with shorter generation lengths by leveraging continuous representations, yet their performance lags behind verbalized CoT. We propose $\...	[ "Keshav Ramji", "Tahira Naseem", "Ramón Fernandez Astudillo" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-04-24T00:00:00	https://arxiv.org/abs/2604.22709	https://arxiv.org/pdf/2604.22709v2	2604.22709	10.48550/arXiv.2604.22709	3	0	false	null	arXiv.org	0.55
1a406f868c12bd57787b2b6a552870075888a0701cd656c49fb458d9184f7bd9	[ "arxiv", "semantic_scholar" ]	Bridging the Reasoning Gap in Vietnamese with Small Language Models via Test-Time Scaling	The democratization of ubiquitous AI hinges on deploying sophisticated reasoning capabilities on resource-constrained devices. However, Small Language Models (SLMs) often face a "reasoning gap", particularly in non-English languages like Vietnamese, where they struggle to maintain coherent chains of thought. This paper...	[ "Bui The Trung", "Do Minh Duc", "Nguyen Van Vinh", "Bui Nguyen Quoc Trinh" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-04-20T00:00:00	https://arxiv.org/abs/2604.17794	https://arxiv.org/pdf/2604.17794v1	2604.17794	10.48550/arXiv.2604.17794	0	0	false	null	arXiv.org	0.55
7ba3c8dac4c812789fba60e24f9a640c641b6e178d4367e44fdb00573c902b45	[ "arxiv", "semantic_scholar" ]	Efficient Test-Time Scaling via Temporal Reasoning Aggregation	Test-time scaling improves the reasoning performance of large language models but often results in token-inefficient overthinking, where models continue reasoning beyond what is necessary for a correct answer. Existing dynamic early-exit methods typically rely on single-step confidence signals, which are often unreliab...	[ "Jiakun Li", "Xingwei He", "Kefan Li", "Hongzheng Chai", "Hongyue Yu", "Yuan Yuan" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-04-19T00:00:00	https://arxiv.org/abs/2604.17304	https://arxiv.org/pdf/2604.17304v1	2604.17304	10.48550/arXiv.2604.17304	0	0	false	null	arXiv.org	0.5489
3f049551d54f083da747e1173326c2334fa9f0a681dacd0b4d0616f82cee9e03	[ "arxiv", "semantic_scholar" ]	LLM Reasoning Is Latent, Not the Chain of Thought	This position paper argues that large language model (LLM) reasoning should be studied as latent-state trajectory formation rather than as faithful surface chain-of-thought (CoT). This matters because claims about faithfulness, interpretability, reasoning benchmarks, and inference-time intervention all depend on what t...	[ "Wenshuo Wang" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-04-17T00:00:00	https://arxiv.org/abs/2604.15726	https://arxiv.org/pdf/2604.15726v1	2604.15726	10.48550/arXiv.2604.15726	0	0	false	null	arXiv.org	0.5466
b7c8cc2bb2a96867aeef5e419ac26c96832dc344a2a09af2fd0e42b8fd3c3fc6	[ "arxiv", "semantic_scholar" ]	Adapting in the Dark: Efficient and Stable Test-Time Adaptation for Black-Box Models	Test-Time Adaptation (TTA) for black-box models accessible only via APIs remains a largely unexplored challenge. Existing approaches such as post-hoc output refinement offer limited adaptive capacity, while Zeroth-Order Optimization (ZOO) enables input-space adaptation but faces high query costs and optimization challe...	[ "Yunbei Zhang", "Shuaicheng Niu", "Chengyi Cai", "Feng Liu", "Jihun Hamm" ]	[ "cs.LG", "cs.CV" ]	[ "Computer Science" ]	2026-04-17T00:00:00	https://arxiv.org/abs/2604.15609	https://arxiv.org/pdf/2604.15609v1	2604.15609	10.48550/arXiv.2604.15609	0	0	false	null	arXiv.org	0.5466
177911204c8c30dcddbd3ef4b09529f145e36d26d4026c13addfdac5e6610579	[ "arxiv", "semantic_scholar" ]	Structured Abductive-Deductive-Inductive Reasoning for LLMs via Algebraic Invariants	Large language models exhibit systematic limitations in structured logical reasoning: they conflate hypothesis generation with verification, cannot distinguish conjecture from validated knowledge, and allow weak reasoning steps to propagate unchecked through inference chains. We present a symbolic reasoning scaffold th...	[ "Sankalp Gilda", "Shlok Gilda" ]	[ "cs.AI", "cs.LG", "cs.LO" ]	[ "Computer Science" ]	2026-04-17T00:00:00	https://arxiv.org/abs/2604.15727	https://arxiv.org/pdf/2604.15727v1	2604.15727	10.48550/arXiv.2604.15727	0	0	false	null	arXiv.org	0.5466
3f4f9817b6710fd89fd891b02c0f418140d51a4f247d8a6f0fa7a969a9b9f5a0	[ "arxiv", "semantic_scholar" ]	ProtoTTA: Prototype-Guided Test-Time Adaptation	Deep networks that rely on prototypes-interpretable representations that can be related to the model input-have gained significant attention for balancing high accuracy with inherent interpretability, which makes them suitable for critical domains such as healthcare. However, these models are limited by their reliance ...	[ "Mohammad Mahdi Abootorabi", "Parvin Mousavi", "Purang Abolmaesumi", "Evan Shelhamer" ]	[ "cs.LG", "cs.CV" ]	[ "Computer Science" ]	2026-04-16T00:00:00	https://arxiv.org/abs/2604.15494	https://arxiv.org/pdf/2604.15494v1	2604.15494	10.48550/arXiv.2604.15494	0	0	true	https://github.com/DeepRCL/ProtoTTA	arXiv.org	0.8429
8076447c2685ddbdf8077674fa06a09d7133f9bdd8fc1deea1fb14c1e4fc6b95	[ "arxiv", "semantic_scholar" ]	LongCoT: Benchmarking Long-Horizon Chain-of-Thought Reasoning	As language models are increasingly deployed for complex autonomous tasks, their ability to reason accurately over longer horizons becomes critical. An essential component of this ability is planning and managing a long, complex chain-of-thought (CoT). We introduce LongCoT, a scalable benchmark of 2,500 expert-designed...	[ "Sumeet Ramesh Motwani", "Daniel Nichols", "Charles London", "Peggy Li", "Fabio Pizzati", "Acer Blake", "Hasan Hammoud", "Tavish McDonald", "Akshat Naik", "Alesia Ivanova", "Vignesh Baskaran", "Ivan Laptev", "Ruben Glatt", "Tal Ben-Nun", "Philip Torr", "Natasha Jaques", "Ameya Prabhu...	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-04-15T00:00:00	https://arxiv.org/abs/2604.14140	https://arxiv.org/pdf/2604.14140v1	2604.14140	10.48550/arXiv.2604.14140	2	0	false	null	arXiv.org	0.5443
5fb881ea6d52354f9ce2cf8c3547a6711f4fc0ec9d9035ad07d3ee343c7e89c1	[ "arxiv", "semantic_scholar" ]	Sample Complexity of Autoregressive Reasoning: Chain-of-Thought vs. End-to-End	Modern large language models generate text autoregressively, producing tokens one at a time. To study the learnability of such systems, Joshi et al. (COLT 2025) introduced a PAC-learning framework for next-token generators, the primitive underlying autoregressive models. In this framework, an unknown next-token generat...	[ "Steve Hanneke", "Idan Mehalel", "Shay Moran" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-04-13T00:00:00	https://arxiv.org/abs/2604.12013	https://arxiv.org/pdf/2604.12013v2	2604.12013	10.48550/arXiv.2604.12013	2	0	false	null	arXiv.org	0.542
83c76090f675531f1ca8e2afffed056d9104f10d7793d4f5f1412e4011be3638	[ "arxiv", "semantic_scholar" ]	When More Thinking Hurts: Overthinking in LLM Test-Time Compute Scaling	Scaling test-time compute through extended chains of thought has become a dominant paradigm for improving large language model reasoning. However, existing research implicitly assumes that longer thinking always yields better results. This assumption remains largely unexamined. We systematically investigate how the mar...	[ "Shu Zhou", "Rui Ling", "Junan Chen", "Xin Wang", "Tao Fan", "Hao Wang" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-04-12T00:00:00	https://arxiv.org/abs/2604.10739	https://arxiv.org/pdf/2604.10739v1	2604.10739	10.48550/arXiv.2604.10739	3	0	false	null	arXiv.org	0.5408
b84b934da1e313fd967f3021bad1618ab33678ce325e58b7fb7479d34090a657	[ "arxiv", "semantic_scholar" ]	FACT-E: Causality-Inspired Evaluation for Trustworthy Chain-of-Thought Reasoning	Chain-of-Thought (CoT) prompting has improved LLM reasoning, but models often generate explanations that appear coherent while containing unfaithful intermediate steps. Existing self-evaluation approaches are prone to inherent biases: the model may confidently endorse coherence even when the step-to-step implication is...	[ "Yuxi Sun", "Aoqi Zuo", "Haotian Xie", "Wei Gao", "Mingming Gong", "Jing Ma" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-04-12T00:00:00	https://arxiv.org/abs/2604.10693	https://arxiv.org/pdf/2604.10693v2	2604.10693	10.48550/arXiv.2604.10693	5	0	false	null	arXiv.org	0.5408
2cc40113a36d0b7e8238c52a4895f8f64cd16a8ca76db21efb05af740b05db94	[ "arxiv", "semantic_scholar" ]	Cognitive Loop of Thought: Reversible Hierarchical Markov Chain for Efficient Mathematical Reasoning	Multi-step Chain-of-Thought (CoT) has significantly advanced the mathematical reasoning capabilities of LLMs by leveraging explicit reasoning steps. However, the widespread adoption of Long CoT often results in sequence lengths that exceed manageable computational limits. While existing approaches attempt to alleviate ...	[ "Jia-Chen Zhang", "Yu-Jie Xiong", "Zheng Zhou" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-04-08T00:00:00	https://arxiv.org/abs/2604.06805	https://arxiv.org/pdf/2604.06805v2	2604.06805	10.48550/arXiv.2604.06805	0	0	false	null	arXiv.org	0.5363
9962d7de9b2e43fa10f41b4d1fe11e55d64f1408698e58e54c18967d8871e3f2	[ "arxiv", "semantic_scholar" ]	Long-Context Reasoning Through Proxy-Based Chain-of-Thought Tuning	Recent large language models support inputs of up to 10 million tokens, yet they perform poorly on long-context tasks that require complex reasoning. Such tasks can be solved using only a subset of the input -- a proxy context -- rather than the full sequence. Despite sharing the same underlying reasoning process, mode...	[ "Miao Li", "Irina Saparina", "Alexander Gurung", "Mirella Lapata" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-04-06T00:00:00	https://arxiv.org/abs/2605.20201	https://arxiv.org/pdf/2605.20201v2	2605.20201	null	0	0	false	null	null	0.3398
921cd8f78d2d1f2caa4a42943d587ad0744bea130005c9e922e8c49d936c6492	[ "arxiv", "semantic_scholar" ]	Hierarchical Chain-of-Thought Prompting: Enhancing LLM Reasoning Performance and Efficiency	Chain-of-Thought (CoT) prompting has significantly improved the reasoning capabilities of large language models (LLMs). However, conventional CoT often relies on unstructured, flat reasoning chains that suffer from redundancy and suboptimal performance. In this work, we introduce Hierarchical Chain-of-Thought (Hi-CoT) ...	[ "Xingshuai Huang", "Derek Li", "Bahareh Nikpour", "Parsa Omidi" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-03-31T00:00:00	https://arxiv.org/abs/2604.00130	https://arxiv.org/pdf/2604.00130v1	2604.00130	10.48550/arXiv.2604.00130	0	0	true	https://github.com/XingshuaiHuang/Hi-CoT	arXiv.org	0.8146
dafc9666d56152aeb3e4fc0b4c6255df90b34b13c0296dd50fcf4b10923f089c	[ "arxiv", "semantic_scholar" ]	CoT2-Meta: Budgeted Metacognitive Control for Test-Time Reasoning	Recent test-time reasoning methods improve performance by generating more candidate chains or searching over larger reasoning trees, but they typically lack explicit control over when to expand, what to prune, how to repair, and when to abstain. We introduce CoT2-Meta, a training-free metacognitive reasoning framework ...	[ "Siyuan Ma", "Bo Gao", "Zikai Xiao", "Hailong Wang", "Xinlei Yu", "Rui Qian", "Jiayu Qian", "Luqi Gong", "Yang Liu" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-03-30T00:00:00	https://arxiv.org/abs/2603.28135	https://arxiv.org/pdf/2603.28135v1	2603.28135	10.48550/arXiv.2603.28135	1	0	false	null	arXiv.org	0.5259
5b2f26528ee86c560c17f9acf5bac05e7029bb072ed334c5b1fb87398d5e9970	[ "arxiv", "semantic_scholar" ]	Enhanced Mycelium of Thought (EMoT): A Bio-Inspired Hierarchical Reasoning Architecture with Strategic Dormancy and Mnemonic Encoding	Current prompting paradigms for large language models (LLMs), including Chain-of-Thought (CoT) and Tree-of-Thoughts (ToT), follow linear or tree-structured reasoning paths that lack persistent memory, strategic dormancy, and cross-domain synthesis. We present the Enhanced Mycelium of Thought (EMoT) framework, a bio-ins...	[ "Florian Odi Stummer" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-03-25T00:00:00	https://arxiv.org/abs/2603.24065	https://arxiv.org/pdf/2603.24065v1	2603.24065	10.48550/arXiv.2603.24065	0	0	false	null	arXiv.org	0.5202
6bb2b49f4817b2a2f5b5b0f6edb669880e002e2b778352834f42a27eaae8ddfc	[ "arxiv", "semantic_scholar" ]	Caterpillar of Thoughts: The Optimal Test-Time Algorithm for Large Language Models	Large language models (LLMs) can often produce substantially better outputs when allowed to use additional test-time computation, such as sampling, chain of thought, backtracking, or revising partial solutions. Despite the growing empirical success of such techniques, there is limited theoretical understanding of how i...	[ "Amir Azarmehr", "Soheil Behnezhad", "Alma Ghafari" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-03-24T00:00:00	https://arxiv.org/abs/2603.22784	https://arxiv.org/pdf/2603.22784v1	2603.22784	10.48550/arXiv.2603.22784	0	0	false	null	arXiv.org	0.5191
8997397b632c2e87d7ca59600c8aed29fb7e8eec1add77c4b3118ff7c7517462	[ "arxiv", "semantic_scholar" ]	Let's Think with Images Efficiently! An Interleaved-Modal Chain-of-Thought Reasoning Framework with Dynamic and Precise Visual Thoughts	Recently, Interleaved-modal Chain-of-Thought (ICoT) reasoning has achieved remarkable success by leveraging both multimodal inputs and outputs, attracting increasing attention. While achieving promising performance, current ICoT methods still suffer from two major limitations: (1) Static Visual Thought Positioning, whi...	[ "Xu Liu", "Yongheng Zhang", "Qiguang Chen", "Yao Li", "Sheng Wang", "Libo Qin" ]	[ "cs.CV", "cs.AI" ]	[ "Computer Science" ]	2026-03-23T00:00:00	https://arxiv.org/abs/2603.21754	https://arxiv.org/pdf/2603.21754v1	2603.21754	10.1609/aaai.v40i38.40494	0	0	false	null	AAAI Conference on Artificial Intelligence	0.5179
76d461eda049cd37ce0a929b31d3d0ebb06afefe939bfe3c11ed475131f1dab4	[ "arxiv", "semantic_scholar" ]	Do Multilingual VLMs Reason Equally? A Cross-Lingual Visual Reasoning Audit for Indian Languages	Vision-language models score well on mathematical, scientific, and spatial reasoning benchmarks, yet these evaluations are overwhelmingly English. I present the first cross-lingual visual reasoning audit for Indian languages. 980 questions from MathVista, ScienceQA, and MMMU are translated into Hindi, Tamil, Telugu, Be...	[ "Swastik R" ]	[ "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-03-23T00:00:00	https://arxiv.org/abs/2603.26742	https://arxiv.org/pdf/2603.26742v1	2603.26742	10.48550/arXiv.2603.26742	0	0	true	https://github.com/QuantumByte-01/multilingual-vlm-reasoning-audit	arXiv.org	0.8004
fe19ca7e418770e0556aaceb5ee7db49a774b7b811556a6ce56b4bc003ebffdf	[ "arxiv", "semantic_scholar" ]	Lie to Me: How Faithful Is Chain-of-Thought Reasoning in Reasoning Models?	Chain-of-thought (CoT) reasoning has been proposed as a transparency mechanism for large language models in safety-critical deployments, yet its effectiveness depends on faithfulness (whether models accurately verbalize the factors that actually influence their outputs), a property that prior evaluations have examined ...	[ "Richard J. Young" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-03-23T00:00:00	https://arxiv.org/abs/2603.22582	https://arxiv.org/pdf/2603.22582v1	2603.22582	10.48550/arXiv.2603.22582	3	1	false	null	arXiv.org	0.5179
cd592e61be487a7e0f1e1e51bc2368b566fe4faae9f543d0e581a5ca9a18bcc0	[ "arxiv", "semantic_scholar" ]	When the Chain Breaks: Interactive Diagnosis of LLM Chain-of-Thought Reasoning Errors	Current Large Language Models (LLMs), especially Large Reasoning Models, can generate Chain-of-Thought (CoT) reasoning traces to illustrate how they produce final outputs, thereby facilitating trust calibration for users. However, these CoT reasoning traces are usually lengthy and tedious, and can contain various issue...	[ "Shiwei Chen", "Niruthikka Sritharan", "Xiaolin Wen", "Chenxi Zhang", "Xingbo Wang", "Yong Wang" ]	[ "cs.HC" ]	[ "Computer Science" ]	2026-03-22T00:00:00	https://arxiv.org/abs/2603.21286	https://arxiv.org/pdf/2603.21286v2	2603.21286	10.1111/cgf.70439	0	0	false	null	null	0.3289
4751df33b16098708d947c17f4f9e8f8568be3b205212ae0f97bc77e2eeb3591	[ "arxiv", "semantic_scholar" ]	Enhancing reasoning accuracy in large language models during inference time	Large Language Models (LLMs) often exhibit strong linguistic abilities while remaining unreliable on multi-step reasoning tasks, particularly when deployed without additional training or fine-tuning. In this work, we study inference-time techniques to improve the reasoning accuracy of LLMs. We systematically evaluate t...	[ "Vinay Sharma", "Manish Jain" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-03-22T00:00:00	https://arxiv.org/abs/2603.21301	https://arxiv.org/pdf/2603.21301v1	2603.21301	10.48550/arXiv.2603.21301	0	0	false	null	arXiv.org	0.5168
d8534aa131519bc69f9d206f8e81ae704e15d57b67cf532e68cc75397bfd3658	[ "arxiv", "semantic_scholar" ]	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 un...	[ "Abinav Rao", "Sujan Rachuri", "Nikhil Vemuri" ]	[ "cs.CL", "cs.AI", "cs.LO" ]	[ "Computer Science" ]	2026-03-19T00:00:00	https://arxiv.org/abs/2604.13065	https://arxiv.org/pdf/2604.13065v1	2604.13065	10.48550/arXiv.2604.13065	0	0	false	null	arXiv.org	0.5133
75f050caf8dc2dffa9a844077350e203896315c238e553187f1c5517c57f2aba	[ "arxiv", "semantic_scholar" ]	Efficient Reasoning on the Edge	Large language models (LLMs) with chain-of-thought reasoning achieve state-of-the-art performance across complex problem-solving tasks, but their verbose reasoning traces and large context requirements make them impractical for edge deployment. These challenges include high token generation costs, large KV-cache footpr...	[ "Yelysei Bondarenko", "Thomas Hehn", "Rob Hesselink", "Romain Lepert", "Fabio Valerio Massoli", "Evgeny Mironov", "Leyla Mirvakhabova", "Tribhuvanesh Orekondy", "Spyridon Stasis", "Andrey Kuzmin", "Anna Kuzina", "Markus Nagel", "Ankita Nayak", "Corrado Rainone", "Ork de Rooij", "Paul N...	[ "cs.LG", "cs.CL" ]	[ "Computer Science" ]	2026-03-17T00:00:00	https://arxiv.org/abs/2603.16867	https://arxiv.org/pdf/2603.16867v2	2603.16867	10.48550/arXiv.2603.16867	0	0	false	null	arXiv.org	0.511
3459d17e911f1be1050b4a7c07782558478a8c1c8da020bf0c3d194606986d20	[ "arxiv", "semantic_scholar" ]	VLA-Thinker: Boosting Vision-Language-Action Models through Thinking-with-Image Reasoning	Vision-Language-Action (VLA) models have shown promising capabilities for embodied intelligence, but most existing approaches rely on text-based chain-of-thought reasoning where visual inputs are treated as static context. This limits the ability of the model to actively revisit the environment and resolve ambiguities ...	[ "Chaoyang Wang", "Wenrui Bao", "Sicheng Gao", "Bingxin Xu", "Yu Tian", "Yogesh S. Rawat", "Yunhao Ge", "Yuzhang Shang" ]	[ "cs.CV", "cs.AI", "cs.RO" ]	[ "Computer Science" ]	2026-03-15T00:00:00	https://arxiv.org/abs/2603.14523	https://arxiv.org/pdf/2603.14523v1	2603.14523	10.48550/arXiv.2603.14523	1	0	false	null	arXiv.org	0.5088
2a779685eddfc77bc5628e0ecd43674294b7d10ca141b335ede0716dfd32c45f	[ "arxiv", "semantic_scholar" ]	TERMINATOR: Learning Optimal Exit Points for Early Stopping in Chain-of-Thought Reasoning	Large Reasoning Models (LRMs) achieve impressive performance on complex reasoning tasks via Chain-of-Thought (CoT) reasoning, which enables them to generate intermediate thinking tokens before arriving at the final answer. However, LRMs often suffer from significant overthinking, spending excessive compute time even af...	[ "Alliot Nagle", "Jakhongir Saydaliev", "Dhia Garbaya", "Michael Gastpar", "Ashok Vardhan Makkuva", "Hyeji Kim" ]	[ "cs.LG", "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-03-13T00:00:00	https://arxiv.org/abs/2603.12529	https://arxiv.org/pdf/2603.12529v2	2603.12529	10.48550/arXiv.2603.12529	2	0	false	null	arXiv.org	0.5065
0eea66db0c5042eb93effe084bc91726263a36248ab125bfb27a0aa1a5cf5b5b	[ "arxiv", "semantic_scholar" ]	EndoCoT: Scaling Endogenous Chain-of-Thought Reasoning in Diffusion Models	Recently, Multimodal Large Language Models (MLLMs) have been widely integrated into diffusion frameworks primarily as text encoders to tackle complex tasks such as spatial reasoning. However, this paradigm suffers from two critical limitations: (i) MLLMs text encoder exhibits insufficient reasoning depth. Single-step e...	[ "Xuanlang Dai", "Yujie Zhou", "Long Xing", "Jiazi Bu", "Xilin Wei", "Yuhong Liu", "Beichen Zhang", "Kai Chen", "Yuhang Zang" ]	[ "cs.CV", "cs.CL" ]	[ "Computer Science" ]	2026-03-12T00:00:00	https://arxiv.org/abs/2603.12252	https://arxiv.org/pdf/2603.12252v4	2603.12252	10.48550/arXiv.2603.12252	1	0	false	null	arXiv.org	0.5053
dd8fecc43a6088400ef2697c117ad0fa02bfeb6ba435b09ca84707997e16ad5c	[ "arxiv", "semantic_scholar" ]	TopoBench: Benchmarking LLMs on Hard Topological Reasoning	Solving topological grid puzzles requires reasoning over global spatial invariants such as connectivity, loop closure, and region symmetry and remains challenging for even the most powerful large language models (LLMs). To study these abilities under controlled settings, we introduce TopoBench, a benchmark of six puzzl...	[ "Mayug Maniparambil", "Nils Hoehing", "Janak Kapuriya", "Arjun Karuvally", "Ellen Rushe", "Anthony Ventresque", "Noel O'Connor", "Fergal Reid" ]	[ "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-03-12T00:00:00	https://arxiv.org/abs/2603.12133	https://arxiv.org/pdf/2603.12133v1	2603.12133	10.48550/arXiv.2603.12133	1	0	true	null	arXiv.org	0.7809
521cd95b0c0f6c04a3427a917b0ae147f042149411f57d15c421a91260a55dd1	[ "arxiv", "semantic_scholar" ]	Ranking Reasoning LLMs under Test-Time Scaling	Test-time scaling evaluates reasoning LLMs by sampling multiple outputs per prompt, but ranking models in this regime remains underexplored. We formalize dense benchmark ranking under test-time scaling and introduce Scorio, a library that implements statistical ranking methods such as paired-comparison models, item res...	[ "Mohsen Hariri", "Michael Hinczewski", "Jing Ma", "Vipin Chaudhary" ]	[ "cs.LG", "math.ST" ]	[ "Computer Science", "Mathematics" ]	2026-03-11T00:00:00	https://arxiv.org/abs/2603.10960	https://arxiv.org/pdf/2603.10960v1	2603.10960	10.48550/arXiv.2603.10960	2	0	true	https://github.com/mohsenhariri/scorio	arXiv.org	0.7792
f7d8989d6f866213a322f399de3956ae349c8bd1d3536e396475c7002e154ebc	[ "arxiv", "semantic_scholar" ]	Efficient Reasoning at Fixed Test-Time Cost via Length-Aware Attention Priors and Gain-Aware Training	We study efficient reasoning under tight compute. We ask how to make structured, correct decisions without increasing test time cost. We add two training only components to small and medium Transformers that also transfer to broader differentiable optimizers. First, a length aware attention prior built via fuzzy regime...	[ "Rian Atri" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-03-10T00:00:00	https://arxiv.org/abs/2603.09253	https://arxiv.org/pdf/2603.09253v1	2603.09253	10.48550/arXiv.2603.09253	0	0	false	null	arXiv.org	0.503
d5600d24a5a80e3123e3fc8d53039758b08eaf99e98c87f678c20fdda4fe6d5e	[ "arxiv", "semantic_scholar" ]	Quantifying the Necessity of Chain of Thought through Opaque Serial Depth	Large language models (LLMs) tend to externalize their reasoning in their chain of thought, making the chain of thought a good target for monitoring. This is partially an inherent feature of the Transformer architecture: sufficiently long serial cognition must pass through the chain of thought (Korbak et al., 2025). We...	[ "Jonah Brown-Cohen", "David Lindner", "Rohin Shah" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-03-10T00:00:00	https://arxiv.org/abs/2603.09786	https://arxiv.org/pdf/2603.09786v1	2603.09786	10.48550/arXiv.2603.09786	0	0	true	null	arXiv.org	0.7774
d1dbff3daecd7a57aabe81629b6abdd81c2d7d049a0f6b051f4a725bd9bf2443	[ "arxiv", "semantic_scholar" ]	Is continuous CoT better suited for multi-lingual reasoning?	We investigate whether performing reasoning in a continuous latent space leads to more robust multilingual capabilities. We compare Continuous Chain-of-Thought (using the CODI framework) against standard supervised fine-tuning across five typologically diverse languages: English, Chinese, German, French, and Urdu. Our ...	[ "Ali Hamza Bashir", "Behzad Shomali", "Markus Frey", "Mehdi Ali", "Rafet Sifa", "David Berghaus" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-03-09T00:00:00	https://arxiv.org/abs/2603.08177	https://arxiv.org/pdf/2603.08177v1	2603.08177	10.48550/arXiv.2603.08177	0	0	false	null	arXiv.org	0.5019
2d71a20be6a548d054b97bf968e779800047b559e3812215bcd6ba36bad55a01	[ "arxiv", "semantic_scholar" ]	Learning When to Sample: Confidence-Aware Selective Sampling for Efficient Chain-of-Thought Reasoning	Large language models (LLMs) can achieve strong reasoning performance through chain-of-thought (CoT) reasoning, yet they often generate unnecessarily long reasoning paths that incur high inference cost. Self-consistency-based approaches push accuracy higher still, but they require sampling and aggregating multiple reas...	[ "Juming Xiong", "Kevin Guo", "Congning Ni", "Wexin Liu", "Chao Yan", "Katherine Brown", "Avinash Baidya", "Xiang Gao", "Bradley Malin", "Zhijun Yin" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-03-09T00:00:00	https://arxiv.org/abs/2603.08999	https://arxiv.org/pdf/2603.08999v3	2603.08999	null	0	0	false	null	null	0.3194
567b5ebc84843a54c715b9ff51e12b5af7e269e2dd68501c32e166c7202ee796	[ "arxiv", "semantic_scholar" ]	FreeFly-Thinking : Aligning Chain-of-Thought Reasoning with Continuous UAV Navigation	Vision-Language Navigation aims to enable agents to understand natural language instructions and carry out appropriate navigation actions in real-world environments. Most work focuses on indoor settings, with little research in complex outdoor scenes. Current UAV Vision-and-Language Navigation models typically act as b...	[ "Jiaxu Zhou", "Shaobo Wang", "Zhiyuan Yang", "Zhenjun Yu", "Tao Li" ]	[ "cs.CV" ]	[ "Computer Science" ]	2026-03-07T00:00:00	https://arxiv.org/abs/2603.07181	https://arxiv.org/pdf/2603.07181v2	2603.07181	10.48550/arXiv.2603.07181	0	0	false	null	arXiv.org	0.4996
ce775f57bdab64bbaa6086fcf405d2ec2ec02a278cad89d9bd8bd54f2895d9ab	[ "arxiv", "semantic_scholar" ]	Improving reasoning at inference time via uncertainty minimisation	Large language models (LLMs) now exhibit strong multi-step reasoning abilities, but existing inference-time scaling methods remain computationally expensive, often relying on extensive sampling or external evaluators. We propose a principled strategy that frames reasoning as uncertainty minimisation and operates at the...	[ "Nicolas Legrand", "Kenneth Enevoldsen", "Márton Kardos", "Kristoffer Nielbo" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-03-07T00:00:00	https://arxiv.org/abs/2603.07159	https://arxiv.org/pdf/2603.07159v1	2603.07159	10.48550/arXiv.2603.07159	0	0	false	null	arXiv.org	0.4996
137010ace5c39f6319afcf24b5436e76155fd711b34a4207594632615d9db313	[ "arxiv", "semantic_scholar" ]	Reasoning Models Struggle to Control their Chains of Thought	Chain-of-thought (CoT) monitoring is a promising tool for detecting misbehaviors and understanding the motivations of modern reasoning models. However, if models can control what they verbalize in their CoT, it could undermine CoT monitorability. To measure this undesirable capability -- CoT controllability -- we intro...	[ "Chen Yueh-Han", "Robert McCarthy", "Bruce W. Lee", "He He", "Ian Kivlichan", "Bowen Baker", "Micah Carroll", "Tomek Korbak" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-03-05T00:00:00	https://arxiv.org/abs/2603.05706	https://arxiv.org/pdf/2603.05706v1	2603.05706	10.48550/arXiv.2603.05706	6	1	false	null	arXiv.org	0.4973
79c22fcb09d4e35448bf438ca5895ca961bf7c2a9e63575591e91d140a1527a6	[ "arxiv", "semantic_scholar" ]	Beyond Test-Time Compute Strategies: Advocating Energy-per-Token in LLM Inference	Large Language Models (LLMs) demonstrate exceptional performance across diverse tasks but come with substantial energy and computational costs, particularly in request-heavy scenarios. In many real-world applications, the full scale and capabilities of LLMs are often unnecessary, as Small Language Models (SLMs) can pro...	[ "Patrick Wilhelm", "Thorsten Wittkopp", "Odej Kao" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-03-04T00:00:00	https://arxiv.org/abs/2603.20224	https://arxiv.org/pdf/2603.20224v1	2603.20224	10.1145/3721146.3721953	19	1	false	null	EuroMLSys2025	0.4961
9ee808f19b0f5f26762fc24c033b7a5c38c7c19a158644494f887dbb1ea70ccd	[ "arxiv", "semantic_scholar" ]	Draft-Thinking: Learning Efficient Reasoning in Long Chain-of-Thought LLMs	Long chain-of-thought~(CoT) has become a dominant paradigm for enhancing the reasoning capability of large reasoning models~(LRMs); however, the performance gains often come with a substantial increase in reasoning budget. Recent studies show that existing CoT paradigms tend to induce systematic overthinking, unnecessa...	[ "Jie Cao", "Tianwei Lin", "Zhenxuan Fan", "Bo Yuan", "Ziyuan Zhao", "Rolan Yan", "Wenqiao Zhang", "Siliang Tang" ]	[ "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-02-28T00:00:00	https://arxiv.org/abs/2603.00578	https://arxiv.org/pdf/2603.00578v1	2603.00578	10.48550/arXiv.2603.00578	0	0	false	null	arXiv.org	0.4916
91da387316ddcc9b9ac8919fb53acbe53f9610ffa0c850d0b5219e7eba943072	[ "arxiv", "semantic_scholar" ]	VisRef: Visual Refocusing while Thinking Improves Test-Time Scaling in Multi-Modal Large Reasoning Models	Advances in large reasoning models have shown strong performance on complex reasoning tasks by scaling test-time compute through extended reasoning. However, recent studies observe that in vision-dependent tasks, extended textual reasoning at inference time can degrade performance as models progressively lose attention...	[ "Soumya Suvra Ghosal", "Youngeun Kim", "Zhuowei Li", "Ritwick Chaudhry", "Linghan Xu", "Hongjing Zhang", "Jakub Zablocki", "Yifan Xing", "Qin Zhang" ]	[ "cs.CV", "cs.AI" ]	[ "Computer Science" ]	2026-02-27T00:00:00	https://arxiv.org/abs/2603.00207	https://arxiv.org/pdf/2603.00207v1	2603.00207	10.48550/arXiv.2603.00207	0	0	false	null	arXiv.org	0.4904
9e2251c99a0b9de6fd5232a99e3761351907cb26f8113adec95c900e7e14a253	[ "arxiv", "semantic_scholar" ]	Learning from Partial Chain-of-Thought via Truncated-Reasoning Self-Distillation	Reasoning-oriented language models achieve strong performance by generating long chain-of-thought traces at inference time. However, this capability comes with substantial and often excessive computational cost, which can materialize in redundant or inefficient reasoning. We study this setting and introduce Truncated-R...	[ "Gianluigi Silvestri", "Edoardo Cetin" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-02-27T00:00:00	https://arxiv.org/abs/2603.13274	https://arxiv.org/pdf/2603.13274v1	2603.13274	10.48550/arXiv.2603.13274	1	0	false	null	arXiv.org	0.4904
b62294fe17f98fb5b0ad0ec2a02499137542d113b06fbd659e3758d885b00edd	[ "arxiv", "semantic_scholar" ]	Stepwise Penalization for Length-Efficient Chain-of-Thought Reasoning	Large reasoning models improve with more test-time computation, but often overthink, producing unnecessarily long chains-of-thought that raise cost without improving accuracy. Prior reinforcement learning approaches typically rely on a single outcome reward with trajectory-level length penalties, which cannot distingui...	[ "Xintong Li", "Sha Li", "Rongmei Lin", "Hongye Jin", "Linwei Li", "Hejie Cui", "Sarah Zhang", "Chia-Yuan Chang", "Kewei Cheng", "Besnik Fetahu", "Priyanka Nigam", "Jingbo Shang", "Bing Yin" ]	[ "cs.CL", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-02-27T00:00:00	https://arxiv.org/abs/2603.00296	https://arxiv.org/pdf/2603.00296v1	2603.00296	10.48550/arXiv.2603.00296	1	0	false	null	arXiv.org	0.4904
a9b8254531790e14a4f1e625155213825a8eacd9a70802c918ea7c2f335a389c	[ "arxiv", "semantic_scholar" ]	TCM-DiffRAG: Personalized Syndrome Differentiation Reasoning Method for Traditional Chinese Medicine based on Knowledge Graph and Chain of Thought	Background: Retrieval augmented generation (RAG) technology can empower large language models (LLMs) to generate more accurate, professional, and timely responses without fine tuning. However, due to the complex reasoning processes and substantial individual differences involved in traditional Chinese medicine (TCM) cl...	[ "Jianmin Li", "Ying Chang", "Su-Kit Tang", "Yujia Liu", "Yanwen Wang", "Shuyuan Lin", "Binkai Ou" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science", "Medicine" ]	2026-02-26T00:00:00	https://arxiv.org/abs/2602.22828	https://arxiv.org/pdf/2602.22828v1	2602.22828	10.3389/fmed.2026.1804478	0	0	false	null	Frontiers in Medicine	0.4893
0bb3ae82830e54c58ea6809aca6b41bf97a9ea84b4fc6c2af218ace9e83a2d26	[ "arxiv", "semantic_scholar" ]	D-COT: Disciplined Chain-of-Thought Learning for Efficient Reasoning in Small Language Models	Chain-of-Thought (CoT) distillation from Large Language Models (LLMs) often induces "overthinking" in Small Language Models (SLMs), leading to performance degradation and excessive token consumption. In this study, we propose Disciplined Chain-of-Thought (D-CoT), a novel framework that enforces a structured reasoning p...	[ "Shunsuke Ubukata" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-02-25T00:00:00	https://arxiv.org/abs/2602.21786	https://arxiv.org/pdf/2602.21786v1	2602.21786	10.48550/arXiv.2602.21786	0	0	true	https://github.com/gitpullpull/DisciplinedChainOfThought	arXiv.org	0.7544
e37d453a9265820365d533c683bd4c97821ea0ed6eb7c87d3b748451d57929a7	[ "arxiv", "semantic_scholar" ]	The Art of Efficient Reasoning: Data, Reward, and Optimization	Large Language Models (LLMs) consistently benefit from scaled Chain-of-Thought (CoT) reasoning, but also suffer from heavy computational overhead. To address this issue, efficient reasoning aims to incentivize short yet accurate thinking trajectories, typically through reward shaping with Reinforcement Learning (RL). I...	[ "Taiqiang Wu", "Zenan Xu", "Bo Zhou", "Ngai Wong" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-02-24T00:00:00	https://arxiv.org/abs/2602.20945	https://arxiv.org/pdf/2602.20945v3	2602.20945	10.48550/arXiv.2602.20945	2	0	false	null	arXiv.org	0.487
fc5a7fe56f7fb18560be34fd6ec26658e61a5948fa4b937e8ea0a670a695fd69	[ "arxiv", "semantic_scholar" ]	To Reason or Not to: Selective Chain-of-Thought in Medical Question Answering	Objective: To improve the efficiency of medical question answering (MedQA) with large language models (LLMs) by avoiding unnecessary reasoning while maintaining accuracy. Methods: We propose Selective Chain-of-Thought (Selective CoT), an inference-time strategy that first predicts whether a question requires reasoning ...	[ "Zaifu Zhan", "Min Zeng", "Shuang Zhou", "Yiran Song", "Xiaoyi Chen", "Yu Hou", "Yifan Wu", "Yang Ruan", "Rui Zhang" ]	[ "cs.CL", "cs.AI" ]	[ "Computer Science" ]	2026-02-23T00:00:00	https://arxiv.org/abs/2602.20130	https://arxiv.org/pdf/2602.20130v1	2602.20130	10.48550/arXiv.2602.20130	0	0	true	null	arXiv.org	0.7508
cecc3fe9230371dc0f84492904b6c929e969df2f9348877e263bb9db0b2b487e	[ "arxiv", "semantic_scholar" ]	Evaluating Chain-of-Thought Reasoning through Reusability and Verifiability	In multi-agent IR pipelines for tasks such as search and ranking, LLM-based agents exchange intermediate reasoning in terms of Chain-of-Thought (CoT) with each other. Current CoT evaluation narrowly focuses on target task accuracy. However, this metric fails to assess the quality or utility of the reasoning process its...	[ "Shashank Aggarwal", "Ram Vikas Mishra", "Amit Awekar" ]	[ "cs.AI", "cs.CL", "cs.IR" ]	[ "Computer Science" ]	2026-02-19T00:00:00	https://arxiv.org/abs/2602.17544	https://arxiv.org/pdf/2602.17544v1	2602.17544	10.48550/arXiv.2602.17544	3	0	false	null	arXiv.org	0.4813
62c4c284364a2abb2c1ca513c64d580beb7f9fb2275350a1b6a7fcede8397b82	[ "arxiv", "semantic_scholar" ]	Framework of Thoughts: A Foundation Framework for Dynamic and Optimized Reasoning based on Chains, Trees, and Graphs	Prompting schemes such as Chain of Thought, Tree of Thoughts, and Graph of Thoughts can significantly enhance the reasoning capabilities of large language models. However, most existing schemes require users to define static, problem-specific reasoning structures that lack adaptability to dynamic or unseen problem type...	[ "Felix Fricke", "Simon Malberg", "Georg Groh" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-02-18T00:00:00	https://arxiv.org/abs/2602.16512	https://arxiv.org/pdf/2602.16512v1	2602.16512	10.48550/arXiv.2602.16512	1	0	false	null	arXiv.org	0.4801
cfcfdc6f8e95f0c7f406a633e639b9ac64e4c28e9b28e973ff6ddb98fb608c79	[ "arxiv", "semantic_scholar" ]	GTS: Inference-Time Scaling of Latent Reasoning with a Learnable Gaussian Thought Sampler	Inference-time scaling (ITS) in latent reasoning models typically relies on heuristic perturbations, such as dropout or fixed Gaussian noise, to generate diverse candidate trajectories. However, we show that stronger perturbations do not necessarily yield better sampling quality: they often induce larger distribution s...	[ "Minghan Wang", "Ye Bai", "Thuy-Trang Vu", "Ehsan Shareghi", "Gholamreza Haffari" ]	[ "cs.CL", "cs.LG" ]	[ "Computer Science" ]	2026-02-15T00:00:00	https://arxiv.org/abs/2602.14077	https://arxiv.org/pdf/2602.14077v2	2602.14077	10.48550/arXiv.2602.14077	0	0	false	null	arXiv.org	0.4767
ded16176fd7022ea5f21054aed51471a6010ae6151ac6eae53361a1a0056e31d	[ "arxiv", "semantic_scholar" ]	The Interspeech 2026 Audio Reasoning Challenge: Evaluating Reasoning Process Quality for Audio Reasoning Models and Agents	Recent Large Audio Language Models (LALMs) excel in understanding but often lack transparent reasoning. To address this "black-box" limitation, we organized the Audio Reasoning Challenge at Interspeech 2026, the first shared task dedicated to evaluating Chain-of-Thought (CoT) quality in the audio domain. The challenge ...	[ "Ziyang Ma", "Ruiyang Xu", "Yinghao Ma", "Chao-Han Huck Yang", "Bohan Li", "Jaeyeon Kim", "Jin Xu", "Jinyu Li", "Carlos Busso", "Kai Yu", "Eng Siong Chng", "Xie Chen" ]	[ "cs.SD", "cs.CL", "cs.MM" ]	[ "Computer Science" ]	2026-02-15T00:00:00	https://arxiv.org/abs/2602.14224	https://arxiv.org/pdf/2602.14224v1	2602.14224	10.48550/arXiv.2602.14224	9	1	false	null	arXiv.org	0.4767
c9e1417ccbbf6bb42215bdc992e88b7880857b40e461add2daa0206b5bd2980f	[ "arxiv", "semantic_scholar" ]	Diagnosing Pathological Chain-of-Thought in Reasoning Models	Chain-of-thought (CoT) reasoning is fundamental to modern LLM architectures and represents a critical intervention point for AI safety. However, CoT reasoning may exhibit failure modes that we note as pathologies, which prevent it from being useful for monitoring. Prior work has identified three distinct pathologies: p...	[ "Manqing Liu", "David Williams-King", "Ida Caspary", "Linh Le", "Hannes Whittingham", "Puria Radmard", "Cameron Tice", "Edward James Young" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-02-14T00:00:00	https://arxiv.org/abs/2602.13904	https://arxiv.org/pdf/2602.13904v1	2602.13904	10.48550/arXiv.2602.13904	1	0	false	null	arXiv.org	0.4755
72ba21d1644158a658cbe0f366a77a750b9ac123bfb46774a65c9ad839e3fe3c	[ "arxiv", "semantic_scholar" ]	UniT: Unified Multimodal Chain-of-Thought Test-time Scaling	Unified models can handle both multimodal understanding and generation within a single architecture, yet they typically operate in a single pass without iteratively refining their outputs. Many multimodal tasks, especially those involving complex spatial compositions, multiple interacting objects, or evolving instructi...	[ "Leon Liangyu Chen", "Haoyu Ma", "Zhipeng Fan", "Ziqi Huang", "Animesh Sinha", "Xiaoliang Dai", "Jialiang Wang", "Zecheng He", "Jianwei Yang", "Chunyuan Li", "Junzhe Sun", "Chu Wang", "Serena Yeung-Levy", "Felix Juefei-Xu" ]	[ "cs.CV", "cs.AI", "cs.LG" ]	[ "Computer Science" ]	2026-02-12T00:00:00	https://arxiv.org/abs/2602.12279	https://arxiv.org/pdf/2602.12279v2	2602.12279	10.48550/arXiv.2602.12279	3	0	false	null	arXiv.org	0.4732
e984ecf62b7b268bb62c552659350cd01f388a8cf8bb4f0f30f44d4e075b0279	[ "arxiv", "semantic_scholar" ]	Characterizing, Evaluating, and Optimizing Complex Reasoning	Large Reasoning Models (LRMs) increasingly rely on reasoning traces with complex internal structures. However, existing work lacks a unified answer to three fundamental questions: (1) what defines high-quality reasoning, (2) how to reliably evaluate long, implicitly structured reasoning traces, and (3) how to use such ...	[ "Haoran Zhang", "Yafu Li", "Zhi Wang", "Zhilin Wang", "Shunkai Zhang", "Xiaoye Qu", "Yu Cheng" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-02-09T00:00:00	https://arxiv.org/abs/2602.08498	https://arxiv.org/pdf/2602.08498v2	2602.08498	10.48550/arXiv.2602.08498	1	0	true	https://github.com/Simplified-Reasoning/TRM	arXiv.org	0.726
33c73e02de489b03686fdd51a186597428b0d0f2d22a5d304fc170ff07fad601	[ "arxiv", "semantic_scholar" ]	Towards Efficient Large Language Reasoning Models via Extreme-Ratio Chain-of-Thought Compression	Chain-of-Thought (CoT) reasoning successfully enhances the reasoning capabilities of Large Language Models (LLMs), yet it incurs substantial computational overhead for inference. Existing CoT compression methods often suffer from a critical loss of logical fidelity at high compression ratios, resulting in significant p...	[ "Yuntian Tang", "Bohan Jia", "Wenxuan Huang", "Lianyue Zhang", "Jiao Xie", "Wenxi Li", "Wei Li", "Jie Hu", "Xinghao Chen Rongrong Ji", "Shaohui Lin" ]	[ "cs.LG" ]	[ "Computer Science" ]	2026-02-09T00:00:00	https://arxiv.org/abs/2602.08324	https://arxiv.org/pdf/2602.08324v5	2602.08324	10.48550/arXiv.2602.08324	3	0	true	https://github.com/Mwie1024/Extra-CoT	arXiv.org	0.726
f4f2cdbb1f79f602792fe318baeb232e641041a4847d7fc9cae7f6bffc167061	[ "arxiv", "semantic_scholar" ]	Recurrent-Depth VLA: Implicit Test-Time Compute Scaling of Vision-Language-Action Models via Latent Iterative Reasoning	Current Vision-Language-Action (VLA) models rely on fixed computational depth, expending the same amount of compute on simple adjustments and complex multi-step manipulation. While Chain-of-Thought (CoT) prompting enables variable computation, it scales memory linearly and is ill-suited for continuous action spaces. We...	[ "Yalcin Tur", "Jalal Naghiyev", "Haoquan Fang", "Wei-Chuan Tsai", "Jiafei Duan", "Dieter Fox", "Ranjay Krishna" ]	[ "cs.RO" ]	[ "Computer Science" ]	2026-02-08T00:00:00	https://arxiv.org/abs/2602.07845	https://arxiv.org/pdf/2602.07845v1	2602.07845	10.48550/arXiv.2602.07845	6	1	false	null	arXiv.org	0.4686
5301e5ea0613803d9fe21d848abd13832061ca0eda4e7fea13da288f1f41a82e	[ "arxiv", "semantic_scholar" ]	Surprisal-Guided Selection: Compute-Optimal Test-Time Strategies for Execution-Grounded Code Generation	Test-time training (TTT) adapts language models through gradient-based updates at inference. But is adaptation the right strategy? We study compute-optimal test-time strategies for verifiable execution-grounded (VEG) tasks, domains like GPU kernel optimization where a deterministic evaluator provides dense, continuous ...	[ "Jarrod Barnes" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-02-07T00:00:00	https://arxiv.org/abs/2602.07670	https://arxiv.org/pdf/2602.07670v1	2602.07670	10.48550/arXiv.2602.07670	0	0	true	https://github.com/jbarnes850/test-time-training	arXiv.org	0.7225
cba375976102edb252d9f26ea2a4383828973bfa5accef3441c18e5b415114bd	[ "arxiv", "semantic_scholar" ]	Are Reasoning LLMs Robust to Interventions on Their Chain-of-Thought?	Reasoning LLMs (RLLMs) generate step-by-step chains of thought (CoTs) before giving an answer, which improves performance on complex tasks and makes reasoning more transparent. But how robust are these reasoning traces to disruptions that occur within them? To address this question, we introduce a controlled evaluation...	[ "Alexander von Recum", "Leander Girrbach", "Zeynep Akata" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-02-07T00:00:00	https://arxiv.org/abs/2602.07470	https://arxiv.org/pdf/2602.07470v1	2602.07470	10.48550/arXiv.2602.07470	4	0	false	null	arXiv.org	0.4675
a46b0d3436f4e1c619b67d8784861348153b3d306cf87ad2306d234bbe6a645c	[ "arxiv", "semantic_scholar" ]	Inference-Time Rethinking with Latent Thought Vectors for Math Reasoning	Standard chain-of-thought reasoning generates a solution in a single forward pass, committing irrevocably to each token and lacking a mechanism to recover from early errors. We introduce Inference-Time Rethinking, a generative framework that enables iterative self-correction by decoupling declarative latent thought vec...	[ "Deqian Kong", "Minglu Zhao", "Aoyang Qin", "Bo Pang", "Chenxin Tao", "David Hartmann", "Edouardo Honig", "Dehong Xu", "Amit Kumar", "Matt Sarte", "Chuan Li", "Jianwen Xie", "Ying Nian Wu" ]	[ "cs.CL", "cs.LG", "stat.ML" ]	[ "Computer Science", "Mathematics" ]	2026-02-06T00:00:00	https://arxiv.org/abs/2602.06584	https://arxiv.org/pdf/2602.06584v1	2602.06584	10.48550/arXiv.2602.06584	0	0	false	null	arXiv.org	0.4664
fb5d3bc6cb961121f789a9f6f7d37778684d989e4c48783c0b4a20377cf5ba93	[ "arxiv", "semantic_scholar" ]	Intrinsic Stability Limits of Autoregressive Reasoning: Structural Consequences for Long-Horizon Execution	Large language models (LLMs) demonstrate remarkable reasoning capabilities, yet their performance often deteriorates sharply in long-horizon tasks, exhibiting systematic breakdown beyond certain scales. Conventional explanations primarily attribute this phenomenon to task complexity, such as combinatorial search explos...	[ "Hsien-Jyh Liao" ]	[ "cs.AI" ]	[ "Computer Science" ]	2026-02-06T00:00:00	https://arxiv.org/abs/2602.06413	https://arxiv.org/pdf/2602.06413v1	2602.06413	10.48550/arXiv.2602.06413	0	0	false	null	arXiv.org	0.4664
51c699147a3a07ed163b45f7ec63cb145ce6db768edee681ed14fd8285bd5a5f	[ "arxiv", "semantic_scholar" ]	SPARC: Separating Perception And Reasoning Circuits for Test-time Scaling of VLMs	Despite recent successes, test-time scaling - i.e., dynamically expanding the token budget during inference as needed - remains brittle for vision-language models (VLMs): unstructured chains-of-thought about images entangle perception and reasoning, leading to long, disorganized contexts where small perceptual mistakes...	[ "Niccolo Avogaro", "Nayanika Debnath", "Li Mi", "Thomas Frick", "Junling Wang", "Zexue He", "Hang Hua", "Konrad Schindler", "Mattia Rigotti" ]	[ "cs.CV", "cs.AI", "cs.CL" ]	[ "Computer Science" ]	2026-02-06T00:00:00	https://arxiv.org/abs/2602.06566	https://arxiv.org/pdf/2602.06566v2	2602.06566	10.48550/arXiv.2602.06566	2	0	false	null	arXiv.org	0.4664
3187fc6060ef24251b5b9863eca5a08632148f06a903bd75fe1490c01fa262dc	[ "arxiv", "semantic_scholar" ]	ORACL: Optimized Reasoning for Autoscaling via Chain of Thought with LLMs for Microservices	Applications are moving away from monolithic designs to microservice and serverless architectures, where fleets of lightweight and independently deployable components run on public clouds. Autoscaling serves as the primary control mechanism for balancing resource utilization and quality of service, yet existing policie...	[ "Haoyu Bai", "Muhammed Tawfiqul Islam", "Minxian Xu", "Rajkumar Buyya" ]	[ "cs.DC" ]	[ "Computer Science" ]	2026-02-05T00:00:00	https://arxiv.org/abs/2602.05292	https://arxiv.org/pdf/2602.05292v1	2602.05292	10.48550/arXiv.2602.05292	0	0	true	null	arXiv.org	0.719
1b52ecb90ed4cb210c9a0c246ed398ec880899651646a3087227467bbf1b07b4	[ "arxiv", "semantic_scholar" ]	MentorCollab: Selective Large-to-Small Inference-Time Guidance for Efficient Reasoning	Large reasoning models (LRMs) achieve strong performance by producing long chains of thought, but their inference costs are high and often generate redundant reasoning. Small language models (SLMs) are far more efficient, yet struggle on multi-step reasoning tasks. A natural idea is to let a large model guide a small o...	[ "Haojin Wang", "Yike Wang", "Shangbin Feng", "Hannaneh Hajishirzi", "Yulia Tsvetkov" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-02-05T00:00:00	https://arxiv.org/abs/2602.05307	https://arxiv.org/pdf/2602.05307v1	2602.05307	10.48550/arXiv.2602.05307	1	0	false	null	arXiv.org	0.4652
aa61f69d9c8363b807291e91e122676db6a66791a437f87f819482f2bf11d01c	[ "arxiv", "semantic_scholar" ]	Mechanistic Evidence for Faithfulness Decay in Chain-of-Thought Reasoning	Chain-of-Thought (CoT) explanations are widely used to interpret how language models solve complex problems, yet it remains unclear whether these step-by-step explanations reflect how the model actually reaches its answer, or merely post-hoc justifications. We propose Normalized Logit Difference Decay (NLDD), a metric ...	[ "Donald Ye", "Max Loffgren", "Om Kotadia", "Linus Wong", "Jonas Rohweder" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-02-04T00:00:00	https://arxiv.org/abs/2602.11201	https://arxiv.org/pdf/2602.11201v2	2602.11201	10.48550/arXiv.2602.11201	7	0	true	https://github.com/donald-ye/NLDD	arXiv.org	0.7172
fa27c13ef3621408f2d2754559968a7bc0eaee5355af301a371c82a6d9cf7835	[ "arxiv", "semantic_scholar" ]	CodeScaler: Scaling Code LLM Training and Test-Time Inference via Reward Models	Reinforcement Learning from Verifiable Rewards (RLVR) has driven recent progress in code large language models by leveraging execution-based feedback from unit tests, but its scalability is fundamentally constrained by the availability and reliability of high-quality test cases. We propose CodeScaler, a reward model de...	[ "Xiao Zhu", "Xinyu Zhou", "Boyu Zhu", "Hanxu Hu", "Mingzhe Du", "Haotian Zhang", "Huiming Wang", "Zhijiang Guo" ]	[ "cs.LG", "cs.AI" ]	[ "Computer Science" ]	2026-02-04T00:00:00	https://arxiv.org/abs/2602.17684	https://arxiv.org/pdf/2602.17684v2	2602.17684	null	0	0	false	null	null	0.2953
dbd9c25063c503c2066ba9710c68b71d7305b6a6cc48cad35e3783fb4366b5ca	[ "arxiv", "semantic_scholar" ]	Reasoning about Reasoning: BAPO Bounds on Chain-of-Thought Token Complexity in LLMs	Inference-time scaling via chain-of-thought (CoT) reasoning is a major driver of state-of-the-art LLM performance, but it comes with substantial latency and compute costs. We address a fundamental theoretical question: how many reasoning tokens are required to solve a problem as input size grows? By extending the bound...	[ "Kiran Tomlinson", "Tobias Schnabel", "Adith Swaminathan", "Jennifer Neville" ]	[ "cs.AI", "cs.FL", "cs.LG" ]	[ "Computer Science" ]	2026-02-02T00:00:00	https://arxiv.org/abs/2602.02909	https://arxiv.org/pdf/2602.02909v2	2602.02909	10.48550/arXiv.2602.02909	2	0	false	null	arXiv.org	0.4618
3164079b38a725273ffa17b38f5bb58be1494c8549d6ad7684df576ed2dbe69f	[ "arxiv", "semantic_scholar" ]	Chronos: Learning Temporal Dynamics of Reasoning Chains for Test-Time Scaling	Test-Time Scaling (TTS) has emerged as an effective paradigm for improving the reasoning performance of large language models (LLMs). However, existing methods -- most notably majority voting and heuristic token-level scoring -- treat reasoning traces or tokens equally, thereby being susceptible to substantial variatio...	[ "Kai Zhang", "Jiayi Liao", "Chengpeng Li", "Ziyuan Xie", "Sihang Li", "Xiang Wang" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-02-01T00:00:00	https://arxiv.org/abs/2602.01208	https://arxiv.org/pdf/2602.01208v1	2602.01208	10.48550/arXiv.2602.01208	0	0	false	null	arXiv.org	0.4606
c03cf2c3f228634f6d5217ac8e571d8609586891459a5b4dfd486913816418f2	[ "arxiv", "semantic_scholar" ]	ReGuLaR: Variational Latent Reasoning Guided by Rendered Chain-of-Thought	While Chain-of-Thought (CoT) significantly enhances the performance of Large Language Models (LLMs), explicit reasoning chains introduce substantial computational redundancy. Recent latent reasoning methods attempt to mitigate this by compressing reasoning processes into latent space, but often suffer from severe perfo...	[ "Fanmeng Wang", "Haotian Liu", "Guojiang Zhao", "Hongteng Xu", "Zhifeng Gao" ]	[ "cs.CL" ]	[ "Computer Science" ]	2026-01-30T00:00:00	https://arxiv.org/abs/2601.23184	https://arxiv.org/pdf/2601.23184v1	2601.23184	10.48550/arXiv.2601.23184	3	0	true	https://github.com/FanmengWang/ReGuLaR	arXiv.org	0.7083
f0dcaddcd026d86677d5f3fdf76f6ed356701c813521b393cb3daab9fa55d682	[ "arxiv", "semantic_scholar" ]	ImgCoT: Compressing Long Chain of Thought into Compact Visual Tokens for Efficient Reasoning of Large Language Model	Compressing long chains of thought (CoT) into compact latent tokens is crucial for efficient reasoning with large language models (LLMs). Recent studies employ autoencoders to achieve this by reconstructing textual CoT from latent tokens, thus encoding CoT semantics. However, treating textual CoT as the reconstruction ...	[ "Xiaoshu Chen", "Sihang Zhou", "Ke Liang", "Taichun Zhou", "Xinwang Liu" ]	[ "cs.CV", "cs.AI" ]	[ "Computer Science" ]	2026-01-30T00:00:00	https://arxiv.org/abs/2601.22730	https://arxiv.org/pdf/2601.22730v1	2601.22730	10.48550/arXiv.2601.22730	1	0	false	null	arXiv.org	0.4583

End of preview. Expand in Data Studio

Test-Time Compute & Reasoning Scaling Papers — FineSet

A research-paper dataset on Test-Time Compute & Reasoning Scaling Papers, assembled, deduplicated, and quality-scored by FineSet from arXiv and Semantic Scholar.

📸 This is a dated snapshot — generated 2026-06-19. It is not auto-updated. Research on Test-Time Compute & Reasoning Scaling Papers moves fast — new papers land on arXiv every week. Want this same dataset refreshed daily, on a topic you choose? See the bottom. ↓

Why this dataset

Quality-scored: quality_score float (0–1), blends citations with recency + code/venue signals — filter out the noise
Papers with code: 151 flagged via has_code — find reproducible work fast
Deduplicated: arXiv + Semantic Scholar cross-referenced, duplicate records merged
Clean JSONL: 480 records, one per line, normalized fields — no encoding garbage

Dataset details

Records: 480
Date range: 2022–2026
Snapshot date: 2026-06-19 (frozen — see note above)
Sources: arXiv, Semantic Scholar (cross-referenced, duplicates merged)
arXiv categories: cs.LG, cs.CL, cs.AI
Quality scoring: citations + recency + code/venue blend, 0–1 (p50=0.389, p90=0.651)
Format: JSONL, one record per line

Fields

Field	Type	Description
id	string	Deterministic SHA256 record id
sources	list	Which sources contributed (`arxiv`, `semantic_scholar`)
title	string	Paper title
abstract	string	Full abstract
authors	list	Author names
categories	list	arXiv category codes
fields_of_study	list	Semantic Scholar field tags
published_date	string	ISO 8601 date
url	string	arXiv abstract URL
pdf_url	string\|null	Open-access PDF if available
arxiv_id	string\|null	arXiv identifier
doi	string\|null	DOI if available
citation_count	int	Citation count (Semantic Scholar)
influential_citation_count	int	Influential citations (Semantic Scholar)
has_code	bool	Code repo detected in the arXiv comment
code_url	string\|null	GitHub URL if detected
venue	string\|null	Publication venue
quality_score	float	0–1, blended (citations + recency + code/venue)

Quality score methodology

quality_score = max(impact, freshness), clamped to [0, 1], where:

impact = max( log10(citations+1)/4 , log10(influential_citations+1)/2 ) — realized impact (0.5 at 100 citations, ~0.75 at 1,000, 1.0 at 10,000+).
freshness = recency × (0.35 + 0.30·has_code + 0.20·has_venue) — a baseline for recent papers (so a strong paper published this week isn't scored 0 just for lacking citations), where recency is 1.0 for papers ≤60 days old and decays linearly to 0 by ~18 months.

Old highly-cited papers score on impact; brand-new papers score on freshness; old uncited papers score ~0. Useful for filtering training data by quality, not just age.

👉 Want this on YOUR topic, updated daily?

This snapshot is frozen at 2026-06-19. The live FineSet pipeline keeps a dataset like this refreshed every day on whatever topic you describe — new papers in, dedup and quality scoring automatic, export as JSONL/Parquet or push straight to the Hub.

Tell me the topic you'd want and I'll run the pipeline on it — open a discussion on this dataset, it's free and it's how I decide what to build next.

→ fineset.io — describe what you want to train on, get a dataset. Early-access waitlist open (referral skip available).

Downloads last month: 34