Pith. sign in

REVIEW 4 major objections 6 minor 35 references

A*-Decoding: Token-Efficient Inference Scaling

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A*-decoding treats generation as best-first search over partial reasoning paths and claims to match brute-force inference-scaling baselines with up to three times fewer tokens and about 30 percent fewer verifier passes.

desk verdict A useful empirical study of PRM-guided search for test-time scaling, but the A* framing and the headline efficiency numbers are substantially overstated. read the letter →

arxiv 2505.13672 v1 pith:TVLW5ZZ3 submitted 2025-05-19 cs.AI

Giannis Chatziveroglou This is my paper · ORCID
classification cs.AI
keywords A*-decodinginference-timescalingprocessrewardmodelstokenefficiencymathematicalreasoningsmalllanguagebest-firstsearchexactmatchaccuracy
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

This paper proposes A*-decoding, a decoding strategy that spends a fixed inference budget by treating language-model generation as A* search over partial reasoning paths. At each step the policy samples a small set of candidate continuations, a process-reward model scores each partial trajectory, and the decoder expands the state with the best combined score until a complete answer is produced. The paper reports that on MATH500 and AIME 2024 this reaches the accuracy of best-of-N and particle filtering while generating up to three times fewer tokens and calling the process-reward model about 30 percent less often, and that models of 1-8B parameters match or exceed much larger chain-of-thought models on these benchmarks. The reason to care is that if true, smarter search over partial solutions can substitute for brute-force sampling and for parameter scale in math reasoning.

What carries the argument

The carrying mechanism is the A* selection rule $f(s') = g(s') + h(s')$, with heuristic $h(s) = 1 - r(s)$ built from an external process-reward score and step cost $g(s,s') = \max(0, h(s)-h(s'))$. A* is a best-first search that always expands the lowest-estimated-total-cost node; here the heuristic tells the decoder how close a partial trajectory looks to a correct solution, and the cost term turns a drop in the heuristic into measurable progress toward the goal. The breadth cap $b_{\max}$ and candidate count $k$ make the search linear in depth, so the whole mechanism converts a fixed token and verifier budget into focused expansion of promising partial reasoning paths rather than full-length resampling.

What would settle it

On a sample of MATH500 problems, record $f(s')$ for every expanded state; if $h$ decreases monotonically along chosen paths, then $g(s,s') = h(s_0) - h(s')$ and $f(s') = h(s_0)$ is constant, so A* selection reduces to insertion-order tie-breaking and a decoder with the $g$ term removed should reproduce the same accuracy and token counts. Equivalently, rerun the main comparison after randomly permuting PRM scores within each depth: if accuracy does not drop, the heuristic signal is not what drives the gains.

Watch

Extended reading notes

Core claim

The central claim is that autoregressive decoding can be reframed as a best-first search over partial trajectories, and that doing so is more token-efficient than sampling full answers. Each successor state $s'$ from a sampled continuation gets the score $f(s') = g(s') + h(s')$, where $h(s) = 1 - r(s)$ is the complement of a process-reward score $r(s)$ and the step cost $g(s,s') = \max(0, h(s)-h(s'))$ counts only positive gains in the heuristic. The lowest-$f$ state is expanded next, and a per-depth breadth cap $b_{\max}$ bounds the graph to $O(b_{\max} \cdot d)$ nodes instead of $O(k^d)$. On MATH500 and AIME 2024 with Llama and Qwen models between 1B and 8B parameters, the paper reports that this reaches the accuracy of best-of-N and particle filtering while using up to three times fewer generated tokens and about 30 percent fewer PRM passes, including a 1B model matching a 70B chain-of-thought model and a 1.7B model reaching o1-class reasoning accuracy.

Load-bearing premise

The load-bearing premise is that a decrease in the process-reward score from one partial trajectory to its continuation is an honest measure of progress toward a correct answer; if that score is noisy or biased, the search spends its budget following a false gradient and the claimed efficiency gain collapses.

Editorial extensions

If this is right

  • In math reasoning, a small model can reach the accuracy of a much larger chain-of-thought model by spending more search budget on partial solutions, so inference compute can substitute for parameter scale.
  • Token efficiency becomes a design axis: the same accuracy as best-of-N can be reached with roughly a third of the generated tokens, lowering latency and serving cost for deployed systems.
  • PRM passes also drop by about 30 percent, which matters when the verifier is itself a large model and dominates inference cost.
  • The benefit holds across model families and sizes from 1B to 8B, so the strategy does not depend on one base model's strength.

Reading between the lines

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

  • The paper does not isolate what A*'s cost term contributes over a plain PRM-greedy decoder; a head-to-head against greedy depth-limited expansion with the same breadth cap would tell whether the $g$ term is doing work or whether the savings come from the breadth cap and heuristic ranking.
  • Whenever a cheap process-level reward exists — unit tests for code, execution traces for tool use, protocol checks for planning — the same framework should transfer, and the paper only sketches this possibility rather than testing it.
  • The reported 1B-versus-70B parity is on exact-match math accuracy; it says nothing about open-ended writing, knowledge grounding, or instruction-following, so the practical claim is narrower than 'small models match large models' in general.
  • Because the $g$ term collapses to a constant on any path where the PRM score only decreases, the search's behavior in practice may be closer to a PRM-guided beam search than to textbook A*; the effective difference is governed by how noisy and non-monotone the PRM scores are.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. This paper introduces A*-decoding, a search-based decoding strategy that casts autoregressive generation as A* search over partial reasoning trajectories, using a process reward model (PRM) as the heuristic h(s)=1-r(s) and a progress-based cost g(s,s')=max(0,h(s)-h(s')). The search is bounded by a candidate count k and a breadth cap bmax. The authors evaluate exact-match accuracy and average generated tokens on MATH500 and AIME 2024 with Llama-3.2-1B/3.1-8B and Qwen3-1.7B/8B, comparing against best-of-N, self-consistency, and particle filtering. They claim up to 3x fewer tokens and 30% fewer PRM passes under equivalent compute budgets, and report that small models can match or exceed much larger CoT models.

Significance. The core idea, using a learned process verifier as a search heuristic to allocate inference compute more selectively, is timely and could be practically valuable if the efficiency claims are correct. The paper evaluates on standard external benchmarks with exact-match scoring, which avoids the circularity of re-predicting fitted constants, and it includes ablations of temperature, PRM choice, and graph breadth. However, the main quantitative claims are not currently supported by the reported experiments: Table 1 does not show a 3x token reduction, no PRM-pass counts are reported, the compared budgets are not matched, and a hyperparameter is tuned on a subset of the test set. The significance as stated is therefore not established, although the underlying direction remains plausible.

major comments (4)
  1. [Table 1 and Section 4.3] The abstract and Section 4.3 claim that A*-decoding uses 'up to 3x fewer tokens' than strong baselines, but Table 1, the only systematic comparison, does not support this. For Llama-3.2-1B on MATH500, A*-decoding uses 35,365 tokens versus 38,856 for particle filtering (about 9% fewer) and more tokens than best-of-N's 30,783; for Qwen3-1.7B on AIME, A* uses 45,139 tokens versus 94,086 for best-of-N but achieves 16.6% accuracy versus 20.0%, so the comparison is at different accuracy points and is not commensurable. The 'up to 3x' figure appears to be derived from interpolating Figure 1, but the operating points, matching procedure, and variance are not specified. Please report per-configuration token counts and PRM-pass counts with error bars and re-derive any efficiency claim directly from those data.
  2. [Section 4.1 and Table 1] The experimental budgets are not equivalent as claimed. Section 4.1 states that all baselines are run with k=64 model generations while A*-decoding uses k=16 'optimal' candidate continuations, and no PRM-pass counts appear anywhere in Table 1 or Section 4.3. Consequently, the abstract's claims of '30% fewer PRM passes' and 'equivalent compute budgets' are unverifiable from the manuscript. Please report the number of PRM queries per method, state the total inference budget in consistent units, and rerun or re-report the efficiency comparisons under matched budgets.
  3. [Section 4.4 and Figure 3] The sampling temperature 0.8 is selected using a 100-problem subset of MATH500, and the same temperature is then used for the main MATH500 results in Table 1. Since MATH500 is the evaluation set, tuning on a subset of it is a form of test-set selection that can bias the reported comparison. Please use a validation split disjoint from the reported test set, or show that the main conclusions are insensitive to temperature across seeds and that the selected value does not advantage A*-decoding over baselines.
  4. [Section 3.3, Eq. (5)] With the cost function g(s,s')=max(0,h(s)-h(s')), the A* total f(s')=g(s')+h(s') is constant along any path on which h is non-increasing, so selection among such candidates reduces to insertion-order tie-breaking. The algorithm is therefore a PRM-greedy, bounded beam search rather than A* with a meaningful cost-to-go, and Appendix A.2 explicitly permits non-admissible heuristics. Since the paper's framing of 'optimally utilizing' a compute budget relies on the A* transition model, please either provide a cost function that is not defined directly from the heuristic, or present the method as heuristic-guided beam search and adjust the optimality and efficiency claims accordingly.
minor comments (6)
  1. [Section 4.3] There is a typo, 'self-sonsistency' instead of 'self-consistency', and the wording 'up to 3x greater overall efficiency' is inconsistent with the abstract's 'up to 3x fewer tokens'; please align the terminology.
  2. [Figure 1 caption] The caption says points show results for 4-64 sampled generations and that A*-decoding uses up to 16 for Llama and 32 for Qwen, but it does not define what 'sampled generations' means for A*-decoding or how the PRM cost axis is computed; please clarify.
  3. [Table 1] No standard errors or confidence intervals are reported for accuracy or token counts; given the modest benchmark sizes and stochastic sampling, please report variance across seeds.
  4. [Appendix A.2] The consistency condition h(s) <= c(s,s')+h(s') is trivially satisfied when c is defined as max(0,h(s)-h(s')), because the cost is constructed from the heuristic; please state explicitly that this is a tautology and explain what, if anything, it guarantees for the search.
  5. [Algorithm 1] Notation is inconsistent between the algorithm (s0) and the text (s_0); please unify.
  6. [References] Reference [1], Beeching et al., lacks a year, venue, and URL; please complete the citation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the search objective is defined from PRM scores, but all claimed outcomes are external benchmark accuracies and token counts, so the derivation does not reduce to its own inputs.

full rationale

The paper's derivation chain is self-contained against external evaluation targets. The heuristic h(s)=1-r(s) (Eq. 4) and the cost function g(s,s')=max(0,h(s)-h(s')) (Eq. 5) do define the search objective in terms of PRM scores, but the paper's headline claims are exact-match accuracy on MATH500 and AIME 2024 and measured token/PRM usage, which are external quantities not identical to the PRM scores used for guidance. No fitted parameter is renamed as a prediction: temperature, graph breadth, and candidate count are tuned on a 100-problem MATH500 subset, which is a possible selection-overfitting or benchmark-leakage concern, but not a definitional circularity because the reported accuracy is still an independent held-out measurement. The PRM choice is justified by external citations (Puri et al. and Zhang et al.), and the paper contains no load-bearing self-citation chain or uniqueness theorem imported from the author's own prior work. There is an internal overclaim: with c(s,s')=max(0,h(s)-h(s')), f(s') equals h(s0) along monotone-decreasing heuristic paths, so A* selection can degenerate to insertion-order tie-breaking, and the paper concedes in Appendix A.2 that admissibility may be traded for informativeness. That is a correctness and verifiability issue about the claimed optimality and the 'up to 3x' token-efficiency figure, not circularity, because neither the search construction nor the benchmark evaluation re-imports the claimed efficiency result as an input. Therefore the appropriate circularity score is 0.

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

The paper introduces no new physical or formal entities. The load-bearing external input is the pre-trained PRM, treated as a black-box oracle. The method's own hyperparameters (k, bmax, temperature, tau_h) are fitted on a small slice of the evaluation benchmark. The central search f-function is constructed from the same PRM scores that guide it, so the algorithm's behavior is co-determined by the reward model's quality and the chosen cost definition.

free parameters (4)
  • k (candidate continuations per expansion) = 16 (Llama), 32 (Qwen in scaling plots)
    Selected as a hyperparameter; Table 1 uses k=16 for A* and k=64 for baselines, an asymmetric budget that favors A*.
  • bmax (graph breadth cap) = 5
    Chosen from ablation on a 100-problem subset of MATH500 (Section 4.4, Appendix A.4).
  • sampling temperature = 0.8
    Chosen from ablation on a 100-problem subset of MATH500 (Section 4.4).
  • tau_h (pruning threshold) = not specified
    Used in Prune(s', bmax, tau_h) in Algorithm 1 but never defined or ablated.
assumptions (4)
  • domain assumption The PRM reward r(s) is a reliable, noise-limited quality signal for partial reasoning trajectories.
    The entire heuristic h(s)=1-r(s) is built on this; Section 4.1 adopts Qwen2.5-Math-PRM-7B and Section 5 acknowledges reliance on external supervision quality.
  • ad hoc to paper Non-admissible heuristics can still steer search to high-quality solutions within finite budgets.
    Appendix A.2 explicitly relaxes admissibility; no bound or guarantee is given for the resulting suboptimality.
  • domain assumption Hyperparameters tuned on a 100-problem subset of MATH500 transfer to the full MATH500 and AIME 2024.
    Section 4.4 selects temperature and bmax on a subset; main results in Section 4.2 use these settings on the full test sets.
  • standard math Standard A* background from Hart et al. [8] applies to this stochastic transition model.
    Section 3.1 is standard; however the paper's transition model is hybrid (stochastic sampling, deterministic selection), so textbook A* guarantees do not directly apply.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A*-Decoding: Token-Efficient Inference Scaling." pith.science (2026). https://pith.science/paper/TVLW5ZZ3

@misc{pith2026250513672,
  author       = {Pith},
  title        = {Pith review of: A*-Decoding: Token-Efficient Inference Scaling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TVLW5ZZ3}},
  note         = {Machine review of arXiv:2505.13672}
}
read the original abstract

Inference-time scaling has emerged as a powerful alternative to parameter scaling for improving language model performance on complex reasoning tasks. While existing methods have shown strong performance gains under fixed compute budgets, there has been little focus on optimally utilizing that budget during inference. In this work, we introduce A*-decoding, a search-based inference-time strategy that builds on the A* search algorithm to optimally utilize a fixed compute budget by prioritizing high-quality reasoning paths during generation. We frame language model decoding as a structured search in a state space of partial solutions, applying the A* transition model to identify promising continuations guided by an external process supervision signal. In our experiments, A*-decoding reaches the performance levels of strong inference scaling baselines like best-of-N and particle filtering while using up to 3x fewer tokens and 30% fewer PRM passes under equivalent compute budgets. On the MATH500 and AIME 2024 benchmarks, A*-decoding enables Llama-3.2-1B-Instruct to match the performance of the 70x larger Llama-3.1-70B-Instruct, and allows Qwen3-1.7B to reach o1-like reasoning accuracy. These results highlight the power of structured search in decoding, offering an alternative to brute-force sampling or scale-driven gains. Our work demonstrates how thoughtful inference-time strategies can enhance reasoning in SLMs, pointing toward future advances in more efficient and scalable language model deployment.

Figures

Figures reproduced from arXiv: 2505.13672 by the authors.

Figure 1
Figure 1. Top: Results with Llama models. Bottom: Results with Qwen models. Left: [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Illustration of A*-decoding applied to a math question involving function composi [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Left: Llama-3.2-1B-Instruct performance on a 100-problem subset of MATH500 [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Llama-3.2-1B-Instruct performance on MATH500 with different max breadth [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 16 canonical work pages

  1. [1]

    Scaling test-time compute with open models

    Edward Beeching, Lewis Tunstall, and Sasha Rush. Scaling test-time compute with open models

    work page

  2. [2]

    Le, Christopher Ré, and Azalia Mirhoseini

    Bradley Brown, Jordan Juravsky, Ryan Ehrlich, Ronald Clark, Quoc V. Le, Christopher Ré, and Azalia Mirhoseini. Large language monkeys: Scaling inference compute with repeated sampling, 2024

    2024

  3. [3]

    Kcts: Knowledge- constrained tree search decoding with token-level hallucination detection, 2023

    Sehyun Choi, Tianqing Fang, Zhaowei Wang, and Yangqiu Song. Kcts: Knowledge- constrained tree search decoding with token-level hallucination detection, 2023

    work page 2023

  4. [4]

    Empirical analysis of beam search performance degradation in neural sequence models

    Eldan Cohen and Christopher Beck. Empirical analysis of beam search performance degradation in neural sequence models. In Kamalika Chaudhuri and Ruslan Salakhutdi- nov, editors,Proceedings of the 36th International Conference on Machine Learning, volume 97 ofProceedings of Machine Learning Research, pages 1290–1299. PMLR, 09–15 Jun 2019

    work page 2019

  5. [5]

    DeepSeek-AI, Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, Xiaokang Zhang, Xingkai Yu, Yu Wu, Z. F. Wu, Zhibin Gou, Zhihong Shao, Zhuoshu Li, Ziyi Gao, Aixin Liu, Bing Xue, Bingxuan Wang, Bochao Wu, Bei Feng, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai D...

    2025

  6. [6]

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, Arun Rao, Aston Zhang, Aurelien Rodriguez, Austen Gregerson, Ava S...

    work page 2024

  7. [7]

    rstar-math: Small llms can master math reasoning with self-evolved deep thinking, 2025

    Xinyu Guan, Li Lyna Zhang, Yifei Liu, Ning Shang, Youran Sun, Yi Zhu, Fan Yang, and Mao Yang. rstar-math: Small llms can master math reasoning with self-evolved deep thinking, 2025

    work page 2025

  8. [8]

    Hart, Nils J

    Peter E. Hart, Nils J. Nilsson, and Bertram Raphael. A formal basis for the heuristic determination of minimum cost paths.IEEE Transactions on Systems Science and Cybernetics, 4(2):100–107, 1968

    1968

Show all 35 references
  1. [9]

    Skywork-o1 open series

    Jujie He, Tianwen Wei, Rui Yan, Jiacai Liu, Chaojie Wang, Yimeng Gan, Shiwen Tu, Chris Yuhao Liu, Liang Zeng, Xiaokun Wang, Boyang Wang, Yongcong Li, Fuxiang Zhang, Jiacheng Xu, Bo An, Yang Liu, and Yahui Zhou. Skywork-o1 open series. https://huggingface.co/Skywork, November 2024

    2024

  2. [10]

    The curious case of neural text degeneration, 2020

    Ari Holtzman, Jan Buys, Li Du, Maxwell Forbes, and Yejin Choi. The curious case of neural text degeneration, 2020

    2020

  3. [11]

    Audrey Huang, Adam Block, Qinghua Liu, Nan Jiang, Akshay Krishnamurthy, and Dylan J. Foster. Is best-of-n the best of them? coverage, scaling, and optimality in inference-time alignment, 2025

    2025

  4. [12]

    Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei

    Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B. Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. Scaling laws for neural language models, 2020

    2020

  5. [13]

    Math-Verify: Math Verification Library

    Hynek Kydlíček. Math-Verify: Math Verification Library. 12

  6. [14]

    Llms-as-judges: A comprehensive survey on llm-based evaluation methods, 2024

    Haitao Li, Qian Dong, Junjie Chen, Huixue Su, Yujia Zhou, Qingyao Ai, Ziyi Ye, and Yiqun Liu. Llms-as-judges: A comprehensive survey on llm-based evaluation methods, 2024

    2024

  7. [15]

    Let’s verify step by step, 2023

    Hunter Lightman, Vineet Kosaraju, Yura Burda, Harri Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. Let’s verify step by step, 2023

    2023

  8. [16]

    Improve mathematical reasoning in language models by automated process supervision, 2024

    Liangchen Luo, Yinxiao Liu, Rosanne Liu, Samrat Phatale, Meiqi Guo, Harsh Lara, Yunxuan Li, Lei Shu, Yun Zhu, Lei Meng, Jiao Sun, and Abhinav Rastogi. Improve mathematical reasoning in language models by automated process supervision, 2024

    2024

  9. [17]

    Self-refine: Iterative refinement with self-feedback, 2023

    Aman Madaan, Niket Tandon, Prakhar Gupta, Skyler Hallinan, Luyu Gao, Sarah Wiegreffe, Uri Alon, Nouha Dziri, Shrimai Prabhumoye, Yiming Yang, Shashank Gupta, Bodhisattwa Prasad Majumder, Katherine Hermann, Sean Welleck, Amir Yazdanbakhsh, and Peter Clark. Self-refine: Iterativ...

    2023

  10. [18]

    The llama 4 herd: The beginning of a new era of natively multimodal ai innovation, 2025

    Meta AI. The llama 4 herd: The beginning of a new era of natively multimodal ai innovation, 2025. Accessed: 2025-05-10

    2025

  11. [19]

    Introducing openai o1, September 2024

    OpenAI. Introducing openai o1, September 2024. Accessed: 2025-05-10

    2024

  12. [20]

    Introducing gpt-4.5, February 2025

    OpenAI. Introducing gpt-4.5, February 2025. Accessed: 2025-05-10

    2025

  13. [21]

    A probabilistic inference approach to inference-time scaling of llms using particle-based monte carlo methods, 2025

    Isha Puri, Shivchander Sudalairaj, Guangxuan Xu, Kai Xu, and Akash Srivastava. A probabilistic inference approach to inference-time scaling of llms using particle-based monte carlo methods, 2025

    2025

  14. [22]

    The effect of sampling temperature on problem solving in large language models

    Matthew Renze. The effect of sampling temperature on problem solving in large language models. InFindings of the Association for Computational Linguistics: EMNLP 2024, page 7346–7356. Association for Computational Linguistics, 2024

    2024

  15. [23]

    Reflexion: Language agents with verbal reinforcement learning, 2023

    Noah Shinn, Federico Cassano, Edward Berman, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. Reflexion: Language agents with verbal reinforcement learning, 2023

    2023

  16. [24]

    Scaling llm test-time compute optimally can be more effective than scaling model parameters, 2024

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

    2024

  17. [25]

    Qwen3 technical report

    Qwen Team. Qwen3 technical report. https://github.com/QwenLM/Qwen3/blob/ main/Qwen3_Technical_Report.pdf, May 2025

    2025

  18. [26]

    Peiyi Wang, Lei Li, Zhihong Shao, R. X. Xu, Damai Dai, Yifei Li, Deli Chen, Y. Wu, and Zhifang Sui. Math-shepherd: Verify and reinforce llms step-by-step without human annotations, 2024

    2024

  19. [27]

    Self-consistency improves chain of thought reasoning in language models, 2023

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models, 2023

    2023

  20. [28]

    Chain-of-thought prompting elicits reasoning in large language models, 2023

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed Chi, Quoc Le, and Denny Zhou. Chain-of-thought prompting elicits reasoning in large language models, 2023

    2023

  21. [29]

    Self-evaluation guided beam search for reasoning, 2023

    Yuxi Xie, Kenji Kawaguchi, Yiran Zhao, Xu Zhao, Min-Yen Kan, Junxian He, and Qizhe Xie. Self-evaluation guided beam search for reasoning, 2023

    2023

  22. [30]

    Griffiths, Yuan Cao, and Karthik Narasimhan

    Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Thomas L. Griffiths, Yuan Cao, and Karthik Narasimhan. Tree of thoughts: Deliberate problem solving with large language models, 2023

    2023

  23. [31]

    React: Synergizing reasoning and acting in language models, 2023

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. React: Synergizing reasoning and acting in language models, 2023

    2023

  24. [32]

    Tenenbaum, and Chuang Gan

    Shun Zhang, Zhenfang Chen, Yikang Shen, Mingyu Ding, Joshua B. Tenenbaum, and Chuang Gan. Planning with large language models for code generation, 2023. 13

    2023

  25. [33]

    The lessons of developing process reward models in mathematical reasoning, 2025

    Zhenru Zhang, Chujie Zheng, Yangzhen Wu, Beichen Zhang, Runji Lin, Bowen Yu, Dayiheng Liu, Jingren Zhou, and Junyang Lin. The lessons of developing process reward models in mathematical reasoning, 2025

    2025

  26. [34]

    Probabilistic inference in language models via twisted sequential monte carlo, 2024

    Stephen Zhao, Rob Brekelmans, Alireza Makhzani, and Roger Grosse. Probabilistic inference in language models via twisted sequential monte carlo, 2024

    2024

  27. [35]

    optimality

    Andy Zhou, Kai Yan, Michal Shlapentokh-Rothman, Haohan Wang, and Yu-Xiong Wang. Language agent tree search unifies reasoning acting and planning in language models, 2024. 14 A Appendix A.1 Language Model Inference as State Space Search (Extended) State space modelsformalize la...

    2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.