Pith. sign in

REVIEW 5 major objections 7 minor 2 cited by

ReCUT: Balancing Reasoning Length and Accuracy in LLMs via Stepwise Trails and Preference Optimization

T0 review · 5 major / 7 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read ReCUT uses stepwise trials and preference learning to cut LLM reasoning length by 30-50% while keeping accuracy.

desk verdict Token reduction is real and useful, but 'maintains accuracy' is not supported by the table; worth a revision and a serious referee. read the letter →

arxiv 2506.10822 v1 pith:6H7OZ3XC submitted 2025-06-12 cs.CL

Zhensheng Jin , Xinze Li , Yifan Ji , Chunyi Peng , Zhenghao Liu , Qi Shi , Yukun Yan , Shuo Wang
show 2 more authors
Furong Peng Ge Yu
This is my paper
classification cs.CL
keywords reasoningcompressionchain-of-thoughtpreferenceoptimizationDPOparameterinterpolationoverthinkingLLMefficiencymathematical
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

ReCUT is a method for making LLMs reason much more tersely without giving up accuracy. The idea is to generate a pool of reasoning paths of very different lengths by stepping through a problem: at each step the model is asked to continue the current best partial solution twice, once with a 'think deeply' instruction and once with a 'be concise' instruction, and the better continuation is chosen using a reward that rewards correct answers and penalizes length. The pool is converted into two preference datasets, one designed to reward accuracy and one to reward shortness, and two models are fine-tuned with DPO. Merging those two models by parameter interpolation gives a single model that keeps the accuracy of the first and the brevity of the second. On math benchmarks with two 7B/8B backbones, the paper reports reasoning lengths roughly 30-50% shorter than a vanilla step-by-step model while Pass@1 accuracy is maintained or improved.

What carries the argument

Three components carry the argument. Long-Short Switched Sampling prompts the model twice at every step, once for a long and once for a short continuation of the current best partial trajectory, and stores both completed paths in a candidate pool. Optimal Reasoning Step Selection scores each completed candidate by $\pm 1/|Y_{[t]}|$ depending on whether its final answer matches the gold answer, and in Eq. (5) takes the first step of the higher-scoring continuation as the next step of the growing optimal path. Gemini-LLM training then builds two DPO datasets from the pool, both with the shortest correct trajectory as the positive sample but with the longest incorrect trajectory as the accuracy-focused negative and the longest correct trajectory as the length-focused negative; the two resulting models are merged with DARE-Ties parameter interpolation to balance the objectives.

What would settle it

Run ReCUT's exploration but replace the Eq. (5) selector with a coin flip between the long and short first steps, keeping the rest of the pipeline identical; if the 30-50% token reduction survives, the stepwise optimal-step mechanism is not what drives compression. Alternatively, inspect the chosen first steps and measure how often they are steps that the final correct trajectory later abandons.

Watch

Extended reading notes

Core claim

The paper's central claim is that stepwise exploration with long-short switching yields a candidate pool diverse enough in both length and correctness that preference optimization alone can compress reasoning by 30-50% without sacrificing accuracy. On the reported benchmarks, ReCUT lowers average generated tokens from 2,233 to 1,298 for Qwen2.5-7B and from 5,331 to 2,058 for Llama-3.1-8B, while the average Pass@1 rises from 43.9 to 45.7 on Qwen and from 27.6 to 29.1 on Llama. The paper also argues that the stepwise selector, not just the length prompting, is responsible: ablations that remove stepwise exploration or the long-short prompts both lose accuracy, and the full model compresses both correct and incorrect trajectories, shrinking the gap in length between them.

Load-bearing premise

The load-bearing premise is the greedy step-selection rule: the first step of whichever full continuation earns the higher reward is assumed to be the right next step, and the paper's own Limitation section concedes that the entire exploration depends on the base model faithfully following the long and short instructions.

Editorial extensions

If this is right

  • On the reported math benchmarks, ReCUT uses roughly half the tokens of vanilla step-by-step prompting while matching or beating its Pass@1, so per-query inference cost drops by about 30-50%.
  • ReCUT compresses incorrect trajectories more than correct ones, so the overthinking pattern in which wrong answers run much longer than right ones is reduced.
  • The method is model-agnostic and preference-based, demonstrated on two different 7B/8B backbones, so it can be layered onto other instruction-following LLMs without architectural changes.
  • Because the final model is a parameter interpolation of an accuracy-optimized and a length-optimized model, the trade-off can in principle be dialed by choosing the interpolation weight rather than retraining.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper, the same stepwise pool construction should transfer to any domain with a cheap correctness signal, such as code generation or constrained QA, where correct and incorrect completions can be scored automatically.
  • Beyond the paper, the reward in Eq. (4) only distinguishes correct from incorrect at a fixed $\pm 1/|Y|$ scale; a graded reward that gives credit for shorter correct paths would likely sharpen the length-accuracy frontier.
  • The authors themselves flag that exploration quality depends on instruction-following and that the DARE-Ties merge parameters are taken from prior practice; both are natural points to vary in follow-up work.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 7 minor

Summary. The paper proposes ReCUT, a method to reduce the reasoning length of LLMs while maintaining accuracy. The method (i) constructs a diverse trajectory pool via stepwise exploration with long/short switching instructions, (ii) trains two DPO models—one favoring accuracy, one favoring shorter reasoning—using preference pairs built from this pool, and (iii) merges them via DARE-Ties parameter interpolation. Experiments on Qwen2.5-7B and Llama-3.1-8B across five math benchmarks report token reductions of roughly 30–50% versus SFT/Direct-DPO/Vanilla/L1 baselines, with comparable or slightly lower Pass@1. The main contributions are the stepwise long-short sampling strategy and the two-model interpolation approach.

Significance. If the claims hold, ReCUT is a practically useful, architecture-agnostic method for reducing inference cost in reasoning LLMs, and the stepwise preference-data construction is a reasonable alternative to whole-trajectory sampling. The paper's strengths are a clear, simple pipeline and direct comparison against several relevant baselines. The length-reduction result is consistently demonstrated. However, the accuracy-maintenance claim is not statistically supported, and a potential training/evaluation data overlap threatens the validity of the absolute accuracy numbers. The method's theoretical grounding is limited—the stepwise greedy selection is a heuristic without analysis. Overall, the idea is promising but the evidence in its current form does not fully support the central 'maintaining accuracy' claim.

major comments (5)
  1. [§5.1, Table 1] The claim that ReCUT 'maintains or even surpasses' baseline accuracy is not supported by the reported results. ReCUT's average Pass@1 is below Direct-DPO on both backbones (Qwen: 45.7 vs 46.4; Llama: 29.1 vs 30.8) and below SFT on Llama (29.1 vs 31.5). For AIME24, AIME25, and AMC23—which contain 30, 30, and 40 problems, respectively—each problem shifts Pass@1 by 2.5–3.3 points, so the observed differences are within the noise floor. No confidence intervals, significance tests, or multiple seeds are reported, making the numerical gaps uninterpretable. The paper should either soften the accuracy claim to 'comparable' with appropriate statistical qualification or provide error bars and significance testing.
  2. [§4, Dataset] The training data are sampled from DeepScaleR-Preview-Dataset, which the paper states comprises AIME, AMC, Omni-Math, and STILL questions. The evaluation sets include AIME24, AIME25, and AMC23. If any of these evaluation problems appear in the 8,000-question training sample, the absolute accuracy numbers for every trained model (including ReCUT and all baselines) are contaminated. The paper must clarify whether the evaluation sets were excluded from the training sample, and if not, rerun the experiments with a disjoint split or report the overlap.
  3. [§3.1, Eq. (5)] The optimal-step selection rule assumes that taking the first step from the continuation with the higher full-trajectory reward is the best next step. This greedy rule is unproven: a longer continuation that corrects an early mistake would receive a high reward, yet its first step may be suboptimal when extended alone. Since Eq. (5) directly determines the partial trajectories used to train the preference models, this heuristic could inject low-quality steps into the candidate pool. The paper should provide an analysis of when this rule is justified, or at least an ablation comparing Eq. (5) against alternatives such as random step selection or selecting based on step-level rewards.
  4. [§5.1, Table 1 (L1 baseline)] The L1 baseline appears to be mis-tuned. On Qwen2.5-7B, L1 achieves only 57.6% Pass@1 on GSM8K, whereas Vanilla reaches 87.4%; similarly low L1 scores appear on MATH500 and AMC23. This suggests the L1 implementation does not match its original paper's hyperparameters or training procedure. As a result, the claim that ReCUT 'significantly outperforms' L1 is not a meaningful comparison. The authors should verify the L1 baseline setup and report the settings used, or replace it with a correctly tuned variant.
  5. [Abstract; §5.1] The abstract's claim of 'reducing reasoning lengths by approximately 30–50%' is over-scoped. While ReCUT does reduce token count substantially versus Vanilla, SFT, Direct-DPO, and L1, it does not hold against the Chain of Draft (CoD) baseline: on Qwen2.5-7B, ReCUT averages 1,298 tokens versus CoD's 239, a 5.4x increase. The paper should either exclude CoD when making the length-reduction claim or state the comparison explicitly with the baselines to which the 30–50% figure applies.
minor comments (7)
  1. [Limitation] The Limitation section contains the typo 'stepwise seasoning trajectory exploration'—presumably 'stepwise sampling trajectory exploration'. This should be corrected.
  2. [§3.2 and §4, Implementation Details] The paper names the two specialized models 'Gemini LLMs', which is potentially confusing given Google's Gemini models. Consider using a different name such as 'twin models' or 'complementary models'.
  3. [§4, Implementation Details] The text states that DARE-Ties density Topx and weight α are set to 0.25, but the Limitation section later says prior work typically uses 0.3–0.4. This inconsistency should be clarified.
  4. [§3.1, Eq. (4)] The reward function in Eq. (4) uses the raw token count |Y[t]|, but the paper does not specify how tokenization is performed (e.g., whitespace, subword units). This should be stated for reproducibility.
  5. [§4, Baseline setup] For the Direct-DPO baseline, the paper says it generates 20 trajectories per question; for L1, no training details are given beyond 'GRPO'. Please provide the full hyperparameter settings for all baselines, including learning rates, LoRA rank, and number of steps, so the comparisons are reproducible.
  6. [§5.3, Figure 4] The paper reports 'reasoning steps' but does not define how a 'step' is counted, particularly for the Vanilla LLM which may not produce explicit [STEP] markers. Please clarify the step-extraction procedure.
  7. [Overall] The paper claims 'All codes and data will be released via https://github.com/NEUIR/ReCUT', but the repository is not yet available. Please either provide a link to a working repository or state the release date.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: ReCUT's preference data are built from gold-answer rewards and external math benchmarks, and its length/accuracy claims are empirical comparisons rather than constructions.

full rationale

ReCUT's derivation chain is self-contained against external correctness signals. Candidate trajectories are sampled from the base model using long/short instructions (Eq. 1-3), but every preference label is assigned by the gold-answer reward in Eq. (4): a trajectory is positive only if its final answer matches o_gold, and negative otherwise, with length used only as a secondary tie-breaker in Eq. (7)-(8). No parameter is fitted to the evaluation benchmarks; DARE-Ties density/weight are fixed at Topx=0.25 and alpha=0.25 from prior empirical practice, as the Limitation section states. The same base model generates both training and evaluation trajectories, but selection is gold-based, so the result is not equivalent to its input by construction. The citation to Jin et al. (2024) supports a peripheral claim about prompt-based methods and is not load-bearing. The central 30-50% length reduction and accuracy-maintenance claims are empirical outcomes reported in Table 1; whether they are statistically robust is a correctness/evidence concern, not a circularity concern.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

ReCUT introduces no new entities; the ledger consists of design assumptions and hyperparameters in the data-construction, training, and merging procedures.

free parameters (5)
  • DPO temperature beta
    Hyperparameter in the DPO loss; no value is reported in the paper.
  • LoRA rank
    Rank used for efficient fine-tuning is not stated.
  • DARE-Ties density Topx = 0.25
    Set to 0.25 based on prior work; the paper states it does not explore other values for weaker models.
  • DARE-Ties interpolation weight alpha = 0.25
    Set to 0.25 based on prior work; no tuning or sensitivity analysis is provided.
  • Maximum exploration steps T = 8 (Qwen2.5-7B), 12 (Llama-3.1-8B)
    Different per backbone; the choice is presented without justification.
assumptions (4)
  • domain assumption The reward function r(Y) = sign(correctness)/|Y| (Eq. 4) appropriately trades off accuracy and length for step selection and preference construction.
    The length-normalized reward is introduced without theoretical or empirical justification; its behavior when comparing trajectories of very different lengths is not analyzed.
  • ad hoc to paper Greedy selection of the next step from the higher-reward full continuation (Eq. 5) yields an optimal or near-optimal partial trajectory.
    The core novelty of the stepwise exploration; the paper provides no proof or analysis that the first step of the better continuation is the best next step.
  • domain assumption DPO on preference data generated by the same base model generalizes and does not overfit.
    Self-training can be fragile; the paper trains only one epoch and does not test with different data scales or seeds.
  • domain assumption DARE-Ties parameter interpolation preserves the complementary strengths of Macc and Mlen without destructive interference.
    The paper relies on prior work (Yu et al., 2024) and admits it did not tune alpha and Topx, which may be inappropriate for weaker base models.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ReCUT: Balancing Reasoning Length and Accuracy in LLMs via Stepwise Trails and Preference Optimization." pith.science (2026). https://pith.science/paper/6H7OZ3XC

@misc{pith2026250610822,
  author       = {Pith},
  title        = {Pith review of: ReCUT: Balancing Reasoning Length and Accuracy in LLMs via Stepwise Trails and Preference Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6H7OZ3XC}},
  note         = {Machine review of arXiv:2506.10822}
}
read the original abstract

Recent advances in Chain-of-Thought (CoT) prompting have substantially improved the reasoning capabilities of Large Language Models (LLMs). However, these methods often suffer from overthinking, leading to unnecessarily lengthy or redundant reasoning traces. Existing approaches attempt to mitigate this issue through curating multiple reasoning chains for training LLMs, but their effectiveness is often constrained by the quality of the generated data and prone to overfitting. To address the challenge, we propose Reasoning Compression ThroUgh Stepwise Trials (ReCUT), a novel method aimed at balancing the accuracy and length of reasoning trajectory. Specifically, ReCUT employs a stepwise exploration mechanism and a long-short switched sampling strategy, enabling LLMs to incrementally generate diverse reasoning paths. These paths are evaluated and used to construct preference pairs to train two specialized models (Gemini LLMs)-one optimized for reasoning accuracy, the other for shorter reasoning. A final integrated model is obtained by interpolating the parameters of these two models. Experimental results across multiple math reasoning datasets and backbone models demonstrate that ReCUT significantly reduces reasoning lengths by approximately 30-50%, while maintaining or improving reasoning accuracy compared to various baselines. All codes and data will be released via https://github.com/NEUIR/ReCUT.

Figures

Figures reproduced from arXiv: 2506.10822 by the authors.

Figure 1
Figure 1. Illustration of Our Reasoning Compression [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The Overview of Our Reasoning Compression Through Stepwise Trials (ReCUT) Model. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Average Lengths of Reasoning Outputs Across [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Performance of Different Models Across Rea [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Case Study. A comparison between the Vanilla LLM and ReCUT (Qwen2.5-7B). The red box marks the [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Performance of Different Models Across Rea [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: The Instruction Used in Our Experiments. [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Contrastive On-Policy Distillation

    cs.CV 2026-07 conditional novelty 6.0 of 10

    Contrasting a teacher's token scores under light- and heavy-thinking prompts yields a training signal that cuts reasoning length roughly in half without hurting accuracy.

  2. Towards Concise and Adaptive Thinking in Large Reasoning Models: A Survey

    cs.AI 2025-07 conditional novelty 3.0 of 10

    A comprehensive review that categorizes methods for shortening and adaptively triggering chain-of-thought reasoning in large language models.

Reference graph

Works this paper leans on

39 extracted references · 9 canonical work pages · cited by 2 Pith papers

  1. [1]

    Pranjal Aggarwal and Sean Welleck. 2025. L1: Controlling how long a reasoning model thinks with reinforcement learning. arXiv preprint arXiv:2503.04697

    arXiv 2025

  2. [2]

    Qiguang Chen, Libo Qin, Jinhao Liu, Dengyun Peng, Jiannan Guan, Peng Wang, Mengkang Hu, Yuhang Zhou, Te Gao, and Wanxiang Che. 2025. Towards reasoning era: A survey of long chain-of-thought for reasoning large language models. arXiv preprint arXiv:2503.09567

    arXiv 2025

  3. [3]

    Xingyu Chen, Jiahao Xu, Tian Liang, Zhiwei He, Jianhui Pang, Dian Yu, Linfeng Song, Qiuzhi Liu, Mengfei Zhou, Zhuosheng Zhang, and 1 others. 2024. Do not think that much for 2+ 3=? on the overthinking of o1-like llms. arXiv preprint arXiv:2412.21187

    arXiv 2024

  4. [4]

    Cheng-Han Chiang and Hung-Yi Lee. 2024. Over-reasoning and redundant calculation of large language models. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 2: Short Papers), pages 161--169

    work page 2024

  5. [5]

    Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, and 1 others. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168

    arXiv 2021

  6. [6]

    Alejandro Cuadron, Dacheng Li, Wenjie Ma, Xingyao Wang, Yichuan Wang, Siyuan Zhuang, Shu Liu, Luis Gaspar Schroeder, Tian Xia, Huanzhi Mao, and 1 others. 2025. The danger of overthinking: Examining the reasoning-action dilemma in agentic tasks. arXiv preprint arXiv:2502.08235

    arXiv 2025

  7. [7]

    Yingqian Cui, Pengfei He, Jingying Zeng, Hui Liu, Xianfeng Tang, Zhenwei Dai, Yan Han, Chen Luo, Jing Huang, Zhen Li, and 1 others. 2025. Stepwise perplexity-guided refinement for efficient chain-of-thought reasoning in large language models. arXiv preprint arXiv:2502.13260

    arXiv 2025

  8. [8]

    DeepSeek-AI, Daya Guo, Dejian Yang, and Haowei Zhang. 2025. https://arxiv.org/abs/2501.12948 Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning . Arxiv preprint, abs/2501.12948

    arXiv 2025

Show all 39 references
  1. [9]

    Bofei Gao, Feifan Song, Zhe Yang, Zefan Cai, Yibo Miao, Chenghao Ma, Shanghaoran Quan, Liang Chen, Qingxiu Dong, Runxin Xu, and 1 others. 2024. Omni-math: A universal olympiad level mathematic benchmark for large language models. In The Thirteenth International Conference on L...

    2024

  2. [10]

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, and 1 others. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783

    2024 arXiv

  3. [11]

    Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. 2021. Measuring mathematical problem solving with the math dataset. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks ...

    2021

  4. [12]

    Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen - Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

    Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen - Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. https://openreview.net/forum?id=nZeVKeeFYf9 Lora: Low-rank adaptation of large language models . In Proceedings of ICLR

    2022

  5. [13]

    Mingyu Jin, Qinkai Yu, Dong Shu, Haiyan Zhao, Wenyue Hua, Yanda Meng, Yongfeng Zhang, and Mengnan Du. 2024. The impact of reasoning step length on large language models. arXiv preprint arXiv:2401.04925

    2024 arXiv

  6. [14]

    Yu Kang, Xianghui Sun, Liangyu Chen, and Wei Zou. 2025. C3ot: Generating shorter chain-of-thought without compromising effectiveness. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 24312--24320

    2025

  7. [15]

    Ayeong Lee, Ethan Che, and Tianyi Peng. 2025. How well do llms compress their own chain-of-thought? a token complexity approach. arXiv preprint arXiv:2503.01141

    2025 arXiv

  8. [16]

    Xiaoxi Li, Guanting Dong, Jiajie Jin, Yuyao Zhang, Yujia Zhou, Yutao Zhu, Peitian Zhang, and Zhicheng Dou. 2025. Search-o1: Agentic search-enhanced large reasoning models. arXiv preprint arXiv:2501.05366

    2025 arXiv

  9. [17]

    Tianqi Liu, Yao Zhao, Rishabh Joshi, Misha Khalman, Mohammad Saleh, Peter J Liu, and Jialu Liu. 2023. Statistical rejection sampling improves preference optimization. arXiv preprint arXiv:2309.06657

    2023 arXiv

  10. [18]

    Michael Luo, Sijun Tan, Justin Wong, Xiaoxiang Shi, William Y Tang, Manan Roongta, Colin Cai, Jeffrey Luo, Tianjun Zhang, Li Erran Li, and 1 others. 2025. Deepscaler: Surpassing o1-preview with a 1.5 b model by scaling rl. Notion Blog

    2025

  11. [19]

    Yun Luo, Zhen Yang, Fandong Meng, Yafu Li, Jie Zhou, and Yue Zhang. 2023. https://arxiv.org/abs/2308.08747 An empirical study of catastrophic forgetting in large language models during continual fine-tuning . ArXiv preprint, abs/2308.08747

    2023 arXiv

  12. [20]

    Yingqian Min, Zhipeng Chen, Jinhao Jiang, Jie Chen, Jia Deng, Yiwen Hu, Yiru Tang, Jiapeng Wang, Xiaoxue Cheng, Huatong Song, and 1 others. 2024. Imitate, explore, and self-improve: A reproduction report on slow-thinking reasoning systems. arXiv preprint arXiv:2412.09413

    2024 arXiv

  13. [21]

    Niklas Muennighoff, Zitong Yang, Weijia Shi, Xiang Lisa Li, Li Fei-Fei, Hannaneh Hajishirzi, Luke Zettlemoyer, Percy Liang, Emmanuel Cand \`e s, and Tatsunori Hashimoto. 2025. s1: Simple test-time scaling. arXiv preprint arXiv:2501.19393

    2025 arXiv

  14. [22]

    OpenAI, Josh Achiam, Steven Adler, and Sandhini Agarwal. 2023. https://arxiv.org/abs/2303.08774 Gpt-4 technical report

    2023 arXiv

  15. [23]

    Manning, Stefano Ermon, and Chelsea Finn

    Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D. Manning, Stefano Ermon, and Chelsea Finn. 2023. http://papers.nips.cc/paper\_files/paper/2023/hash/a85b405ed65c6477a4fe8302b5e06ce7-Abstract-Conference.html Direct preference optimization: Your language model is sec...

    2023

  16. [24]

    David Rein, Betty Li Hou, Asa Cooper Stickland, Jackson Petty, Richard Yuanzhe Pang, Julien Dirani, Julian Michael, and Samuel R Bowman. 2024. Gpqa: A graduate-level google-proof q&a benchmark. In First Conference on Language Modeling

    2024

  17. [25]

    Matthew Renze and Erhan Guven. 2024. The benefits of a concise chain of thought on problem-solving in large language models. In 2024 2nd International Conference on Foundation and Large Language Models (FLLM), pages 476--483. IEEE

    2024

  18. [26]

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. 2017. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347

    2017 arXiv

  19. [27]

    Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Y Wu, and 1 others. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300

    2024 arXiv

  20. [28]

    Charlie Snell, Jaehoon Lee, Kelvin Xu, and Aviral Kumar. 2024. Scaling llm test-time compute optimally can be more effective than scaling model parameters. arXiv preprint arXiv:2408.03314

    2024 arXiv

  21. [29]

    Yang Sui, Yu-Neng Chuang, Guanchu Wang, Jiamu Zhang, Tianyi Zhang, Jiayi Yuan, Hongyi Liu, Andrew Wen, Hanjie Chen, Xia Hu, and 1 others. 2025. Stop overthinking: A survey on efficient reasoning for large language models. arXiv preprint arXiv:2503.16419

    2025 arXiv

  22. [30]

    Kimi Team, Angang Du, Bofei Gao, Bowei Xing, Changjiu Jiang, Cheng Chen, Cheng Li, Chenjun Xiao, Chenzhuang Du, Chonghua Liao, and 1 others. 2025. Kimi k1. 5: Scaling reinforcement learning with llms. arXiv preprint arXiv:2501.12599

    2025 arXiv

  23. [31]

    Qwen Team. 2025. https://qwenlm.github.io/blog/qwq-32b/ Qwq-32b: Embracing the power of reinforcement learning

    2025

  24. [32]

    Siyuan Wang, Enda Zhao, Zhongyu Wei, and Xiang Ren. 2025. Stepwise informativeness search for improving llm reasoning. arXiv preprint arXiv:2502.15335

    2025 arXiv

  25. [33]

    Chi, Quoc V

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. 2022. http://papers.nips.cc/paper\_files/paper/2022/hash/9d5609613524ecf4f15af0f7b31abca4-Abstract-Conference.html Chain-of-thought prompting elicits reasoning...

    2022

  26. [34]

    Yangzhen Wu, Zhiqing Sun, Shanda Li, Sean Welleck, and Yiming Yang. 2024. An empirical analysis of compute-optimal inference for problem-solving with language models

    2024

  27. [35]

    Silei Xu, Wenhao Xie, Lingxiao Zhao, and Pengcheng He. 2025. Chain of draft: Thinking faster by writing less. arXiv preprint arXiv:2502.18600

    2025 arXiv

  28. [36]

    An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, and 1 others. 2024. Qwen2. 5 technical report. arXiv preprint arXiv:2412.15115

    2024 arXiv

  29. [37]

    Le Yu, Bowen Yu, Haiyang Yu, Fei Huang, and Yongbin Li. 2024. Language models are super mario: Absorbing abilities from homologous models as a free lunch. In Forty-first International Conference on Machine Learning

    2024

  30. [38]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  31. [39]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.