Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Multilingual Test-Time Scaling via Initial Thought Transfer

T0 review · 3 major / 5 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 DeepSeek-R1-Distill-Qwen-7B's multilingual test-time scaling is restored by fine-tuning on English-only initial reasoning prefixes.

desk verdict The descriptive multilingual test-time scaling study is new, but MITT's headline gains are uninterpretable due to train/eval overlap on the same 30 questions. read the letter →

arxiv 2505.15508 v1 pith:CLGMRKHV submitted 2025-05-21 cs.CL

Prasoon Bajpai , Tanmoy Chakraborty This is my paper
classification cs.CL
keywords test-timescalingmultilingualreasoninginitialthoughttransferprefixtuninglow-resourcelanguagescross-lingualchain-of-thoughtDeepSeek-R1
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

Test-time scaling does not benefit all languages equally, and this paper argues the gap has a concrete cause in the earliest reasoning thoughts. The paper evaluates DeepSeek-R1-Distill-Llama-8B and DeepSeek-R1-Distill-Qwen-7B on challenging mathematics in English, Italian, German, Portuguese, Vietnamese, and Tagalog, finding that high-resource languages show clearer accuracy gains as generation length grows, while low-resource languages, especially for Qwen-7B, scale erratically or not at all. It also finds that models switch to English mid-reasoning under strict monolingual prompts, and that low-resource initial reasoning prefixes are less similar to English and less consistent across repeated generations. Building on that diagnosis, the paper proposes MITT, an unsupervised prefix-tuning method that fine-tunes on English or high-resource-language reasoning prefixes and restores steady test-time scaling across languages. If correct, English-only, unlabeled reasoning prefixes can act as transferable inductive priors, improving low-resource reasoning without target-language training data.

What carries the argument

The carrying mechanism is the reasoning prefix: the first 32 tokens of a chain of thought, treated as a portable inductive prior. MITT collects 100 such prefixes per question in English, or in a pooled set of high-resource languages, and fine-tunes DeepSeek-R1-Distill-Qwen-7B with 4-bit quantized LoRA adapters on the query and value projections. The evaluation protocol that makes the claim visible appends an answer-extraction prompt every 32 generated tokens and inserts a language-specific wait prompt whenever the model attempts to answer on its own, so accuracy can be plotted against reasoning length. The paper's key observation is that after the English-prefix tune, this accuracy-versus-length curve turns from flat to clearly increasing for low-resource languages, which supports the transfer claim.

What would settle it

Rerun the DeepSeek-R1-Distill-Qwen-7B evaluation with wait prompts removed entirely or held constant in English for every language, and check whether low-resource scaling curves still stay flat; if the gap disappears, the reported language disparity comes from the prompt mechanics rather than from model reasoning. A second decisive comparison would fine-tune the model on low-resource-language prefixes of equal length and show whether MITT's English prefixes beat them.

Watch

Extended reading notes

Core claim

The central discovery is that multilingual test-time scaling is deeply uneven and that the unevenness tracks the stability and similarity of the model's earliest reasoning. Test-time scaling gains are robust for high-resource languages but weak or absent for low-resource ones in DeepSeek-R1-Distill-Qwen-7B, while DeepSeek-R1-Distill-Llama-8B scales more steadily across languages. Low-resource languages produce initial reasoning prefixes, the first 32 tokens of a chain of thought, that diverge sharply from English and are more variable across repeated generations, which the paper interprets as less stable inductive reasoning priors. The proposed intervention, Multilingual Initial Thought Transfer (MITT), fine-tunes the model with LoRA on English-only reasoning prefixes and reports that this single unsupervised English tune raises accuracy at every tested token budget and turns a flat accuracy-versus-length curve into a clearly increasing one across high- and low-resource languages.

Load-bearing premise

The comparison assumes that inserting an answer-extraction prompt every 32 tokens and manually placed language-specific wait prompts measures reasoning quality fairly and equivalently in all six languages; if these prompts help or hinder different languages differently, the cross-lingual scaling curves are not comparable and the 'uneven scaling' finding collapses.

Editorial extensions

If this is right

  • If English prefix tuning provides transferable reasoning priors, an unsupervised English-only fine-tune can substitute for supervised multilingual reasoning data when improving low-resource test-time scaling.
  • Test-time scaling evaluations should report per-language curves, because language-averaged curves can hide the flat or erratic scaling of low-resource languages.
  • Mid-reasoning switches to English under monolingual prompts mean language fidelity is a separate failure mode that multilingual reasoning benchmarks should track alongside accuracy.
  • Models with code-heavy, English-skewed pretraining mixes may need targeted prefix adaptation before test-time scaling can help non-English languages.
  • The first-32-token prefix consistency of a language could serve as a cheap diagnostic for whether a model will benefit from longer reasoning in that language.

Reading between the lines

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

  • An untested extension is whether the same English-prefix transfer works for non-Latin scripts such as Arabic, Hindi, or Chinese; the paper only studies Latin-script languages, so the mechanism could be tied to shared tokenization or Roman-script similarity.
  • A sharper causal test would compare English-prefix tuning with prefix tuning on the same number of low-resource-language prefixes; if low-resource prefixes work equally well, the real bottleneck is consistency of early reasoning, not the English language itself.
  • If the effect is about structuring the start of a chain of thought rather than about English content, then even non-linguistic structured prefixes, such as a fixed planning template, might reproduce part of the gain—this is my inference, not the paper's claim.
  • The periodic answer-extraction and wait-prompt protocol itself may interact with the fine-tune; a practitioner should check whether MITT's gain persists when wait prompts are omitted before attributing it purely to reasoning quality.
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

3 major / 5 minor

Summary. The paper presents a multilingual study of test-time scaling on AIME 2025, translated into six Latin-script languages, using DeepSeek-R1-Distill-Llama-8B and DeepSeek-R1-Distill-Qwen-7B. It reports that test-time scaling gains are uneven across languages and models, that models often switch to English mid-reasoning under monolingual prompts, and that low-resource languages have less similar and less consistent initial reasoning prefixes relative to English. The paper then proposes MITT, an unsupervised prefix-tuning method that fine-tunes DeepSeek-R1-Distill-Qwen-7B with LoRA on the first 32 tokens of reasoning prefixes sampled in English (E-3) or in four high-resource languages (H-1), and claims that this improves accuracy and test-time scaling across all six languages, especially low-resource ones. The central contribution, however, is evaluated on the same 30 questions used to collect the training prefixes, with no held-out split, no error bars, and an underspecified inference-time answer-detection protocol.

Significance. The descriptive findings (uneven scaling, language switching, prefix dissimilarity and consistency) are a useful contribution to a largely English-centric literature, and the paper extends AIME 2025 with Vietnamese and Tagalog translations. The two-method similarity analysis (multilingual embeddings and translation-plus-English-embeddings) is a reasonable attempt to control for embedding bias. That said, the paper's headline method, MITT, is not supported by the reported experiment: the training prefixes are drawn from the same questions used for evaluation, and accuracy differences are not quantified with uncertainty. The transferability claim in the abstract and Section 4.5 therefore requires a proper held-out evaluation before the method can be assessed. The paper does not ship machine-checked proofs or reproducible code in a verifiable form; the GitHub link is a placeholder.

major comments (3)
  1. [§3.6, §4.5, Table 1] MITT is fine-tuned on initial reasoning prefixes sampled from the same AIME 2025 questions that are used for evaluation. Section 3.6 states that "we extract all 100 initial reasoning prefixes (i.e., the first 32 tokens) sampled per question," and Section 4.5 describes training on "each of 100 English reasoning prefixes for each question for both datasets," while Table 1 reports accuracy on those same questions in six languages. No held-out split is described anywhere in the paper, including Section 7, which lists limitations but does not mention this overlap. Improvements in Table 1 could therefore be explained by memorization of question-specific reasoning starts rather than by transferable inductive priors. The claim in Section 4.5 that "English prefix tuning provides broadly transferable inductive priors for reasoning across the multilingual spectrum" is not supported by the reported experiment. The authors should retrain on a disjoint set of questions (or a disjoint reasoning dataset) and evaluate on the same test questions, and also report an ablation that trains on prefixes from one set of questions and evaluates on another.
  2. [§3.2, Appendix B] The test-time scaling protocol is underspecified at a load-bearing point. Section 3.2 says a language-specific wait prompt is inserted "whenever M tries to produce an answer on its own," but the manuscript never states how this event is detected during inference. Appendix B describes how wait prompts were placed manually in the 1-shot demonstration, and the main text does not say whether the same mechanism is used for all six languages and for all strategies (Base, E-3, H-1). If the trigger or the placement of wait prompts differs across languages or runs, the cross-lingual scaling curves in Figure 2 and the MITT comparisons in Table 1 are not comparable. The authors should specify the automatic detection rule (for example, detecting an answer-like pattern or a stop token), and verify that it is applied identically across conditions.
  3. [Table 1, Figure 1] The reported accuracies are based on 30 questions per language (two AIME subsets of 15) and are presented without error bars, raw counts, or significance tests. Several differences that support the headline claim are differences of one to three questions; for example, English Base at 2000 tokens is 0.164 and E-3 is 0.257, and Tagalog Base at 2000 is 0.043 versus E-3 at 0.077. Conversely, the claim of a "clear upward trajectory" is not uniform across Table 1: German H-1 drops to 0.000 at 6000 and 8000 tokens, and English H-1 at 2000 and 4000 is below E-3 and sometimes below Base. Bootstrap confidence intervals and per-question results are needed before "significantly boosts" (abstract) can be assessed.
minor comments (5)
  1. [Table 1] Several numeric entries in Table 1 are concatenated (e.g., "0.2540.251", "0.3430.271", "0.2390.235", "0.1760.349", "0.0780.1080.1930.147", "0.2200.1930.237 0.209"), making the table unreliable as printed.
  2. [§3.1, Figure 2] The model name is inconsistently capitalized as "DeepSeek-R1-Distill-LLama-8B" rather than "Llama" in several places, and Section 4.1 contains a typo "Disill" for "Distill".
  3. [Figure 2 caption] The caption is incomplete: it ends with a colon followed by a period, with no sentence following the colon.
  4. [Introduction] The GitHub repository link is a placeholder ("Link"), so the claimed release of dataset and code cannot be verified.
  5. [Appendix B] Appendix B says wait prompts are manually placed in the demonstration; the main text should clarify whether the same prompts are also injected automatically during evaluation, and if so, how the trigger is detected.

Circularity Check

1 steps flagged · score 7.0 of 10

MITT's reported gains are computed on the same AIME 2025 questions used to collect its prefix-tuning data, so the claimed cross-lingual transfer is not independently tested.

  1. fitted input called prediction [Section 3.6 (MITT training) and Section 4.5 / Table 1 (claimed transfer)]
    "Specifically, we extract all 100 initial reasoning prefixes (i.e., the first 32 tokens) sampled per question and fine-tune the model using LoRA adapters applied to the query and value matrices. We then evaluate the downstream effects on test-time scaling performance across both low-resource languages and high-resource counterparts. ... suggesting that English prefix tuning provides broadly transferable inductive priors for reasoning across the multilingual spectrum."

    The dataset is the multilingual AIME 2025 set, and Table 1 reports accuracy at varying token limits on the same questions; the paper never introduces a held-out question split. The LoRA adapters are therefore fitted on English reasoning prefixes sampled from the very questions on which the E-3 and H-1 rows are later reported. Under this protocol, any improvement on those questions can be question-specific prefix memorization rather than an independently demonstrated transfer mechanism. The Section 4.5 inference that English prefix tuning provides broadly transferable inductive priors is not tested by the experiment: the transfer claim is derived from same-question evaluation, so the prediction is forced by training on the test items by construction.

full rationale

The central claimed contribution is MITT as a transferable, language-general intervention. The paper's own protocol trains on 'all 100 initial reasoning prefixes ... sampled per question' from the same AIME 2025 questions later used for evaluation, with no held-out split described anywhere in the main text or appendix. Consequently, the accuracy gains in Table 1 and Figure 1 are compatible with the model memorizing question-specific reasoning starts, and the leap to 'broadly transferable inductive priors' is unsupported by the experimental design. The descriptive findings about cross-lingual test-time scaling, language drift, and prefix consistency are not circular because they involve no training on the evaluation items; they may still suffer from protocol issues (e.g., manually inserted wait prompts and answer extraction), but those are correctness risks rather than circularity. One specific circular step is identified: the prediction of improved multilingual performance reduces to fitting on the evaluation question set.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central experiment has no fitted numerical parameters, but it depends on several hand-chosen hyperparameters and domain assumptions. The decisive issue is the overlap between MITT training prefixes and the evaluation questions, documented under red flags.

free parameters (7)
  • prefix_length = 32 tokens
    Hand-chosen threshold for what counts as an initial thought; motivated by Ji et al. (2025a) but not swept or justified for multilingual use (Section 3.6).
  • loRA_rank = 8
    LoRA rank used in MITT fine-tuning (Appendix A); no ablation.
  • lora_alpha = 32
    LoRA scaling factor (Appendix A); no ablation.
  • dropout = 0.05
    LoRA dropout (Appendix A); no ablation.
  • english_epochs = 3
    Epochs for E-3 training; no validation set or early stopping (Section 3.6).
  • multilingual_epochs = 1
    Epochs for H-1 training; no validation set or early stopping (Section 3.6).
  • samples_per_question = 100
    Number of reasoning prefixes sampled per question for training and consistency analysis (Sections 3.5, 3.6).
assumptions (5)
  • domain assumption Test-time scaling as measured by periodic answer extraction reflects genuine reasoning quality.
    The paper assumes that appending an answer prompt to a partial trace and requiring exact match is a valid measure; Section 3.2.
  • domain assumption The langid classifier correctly labels 32-token windows.
    Section 3.3; short windows and code-mixed text may be misclassified.
  • domain assumption English is a suitable universal source of reasoning prefixes for transfer.
    MITT assumes English reasoning structure transfers; Section 3.6.
  • domain assumption The 'first few tokens are all you need' hypothesis (Ji et al. 2025a) holds in cross-lingual settings.
    Cited as motivation for MITT; Section 3.6.
  • domain assumption GPT-4o and Gemini translations are faithful.
    Dataset creation and similarity analysis rely on machine translation quality; Sections 3.1 and 3.5.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multilingual Test-Time Scaling via Initial Thought Transfer." pith.science (2026). https://pith.science/paper/CLGMRKHV

@misc{pith2026250515508,
  author       = {Pith},
  title        = {Pith review of: Multilingual Test-Time Scaling via Initial Thought Transfer},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CLGMRKHV}},
  note         = {Machine review of arXiv:2505.15508}
}
read the original abstract

Test-time scaling has emerged as a widely adopted inference-time strategy for boosting reasoning performance. However, its effectiveness has been studied almost exclusively in English, leaving its behavior in other languages largely unexplored. We present the first systematic study of test-time scaling in multilingual settings, evaluating DeepSeek-R1-Distill-LLama-8B and DeepSeek-R1-Distill-Qwen-7B across both high- and low-resource Latin-script languages. Our findings reveal that the relative gains from test-time scaling vary significantly across languages. Additionally, models frequently switch to English mid-reasoning, even when operating under strictly monolingual prompts. We further show that low-resource languages not only produce initial reasoning thoughts that differ significantly from English but also have lower internal consistency across generations in their early reasoning. Building on our findings, we introduce MITT (Multilingual Initial Thought Transfer), an unsupervised and lightweight reasoning prefix-tuning approach that transfers high-resource reasoning prefixes to enhance test-time scaling across all languages, addressing inconsistencies in multilingual reasoning performance. MITT significantly boosts DeepSeek-R1-Distill-Qwen-7B's reasoning performance, especially for underrepresented languages.

Figures

Figures reproduced from arXiv: 2505.15508 by the authors.

Figure 1
Figure 1. Effect of Multilingual Initial Thought Transfer [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Test-time scaling trends for (a) + (b) DeepSeek-R1-Distill-LLama-8B and (c) + (d) DeepSeek-R1- Distill-Qwen-7B: . (a) and (c) display overall trends in test-time scaling across all languages, while (b) and (d) present average gains separately for low-resource and high-resource language groups. The results reveal a consistent pattern: both models demonstrate stronger test-time scaling in high-resource languages compa… view at source ↗
Figure 3
Figure 3. Visualization of the sentence-level language trajectory across the generated reasoning stream. Each row [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Comparison of the similarity of initial reason [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Distribution of similarity scores across all questions, for initial reasoning segments (first 32 tokens) [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Training loss per training step for training on [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Panel (a) provides the full prompt template that we use for inference. Panel (b) gives an overview of our [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Figure shows pairwise similarity scores between initial reasoning segments (first 32 tokens) generated in [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Each row represents the average success score in target language production per token generation segment. [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

  1. EfficientXLang: Towards Improving Token Efficiency Through Cross-Lingual Reasoning

    cs.CL 2025-06 conditional novelty 5.0 of 10

    Reasoning in non-English languages reduces thinking tokens by 20-40% while largely preserving math accuracy, with savings persisting after translation to English.

Reference graph

Works this paper leans on

41 extracted references · 3 canonical work pages · cited by 1 Pith paper

  1. [1]

    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 INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    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 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    AIME: A dataset on complex numerical problems

    2025. AIME: A dataset on complex numerical problems . https://huggingface.co/datasets/fedric95/AIME2025-Multilingual

    work page 2025

  4. [4]

    Jinze Bai, Shuai Bai, Yunfei Chu, and Zeyu Cui et al. 2023. https://arxiv.org/abs/2309.16609 Qwen technical report . Preprint, arXiv:2309.16609

    arXiv 2023

  5. [5]

    Sébastien Bubeck, Varun Chandrasekaran, Ronen Eldan, Johannes Gehrke, Eric Horvitz, Ece Kamar, Peter Lee, Yin Tat Lee, Yuanzhi Li, Scott Lundberg, Harsha Nori, Hamid Palangi, Marco Tulio Ribeiro, and Yi Zhang. 2023. https://arxiv.org/abs/2303.12712 Sparks of artificial general intelligence: Early experiments with gpt-4 . Preprint, arXiv:2303.12712

    arXiv 2023

  6. [6]

    Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. https://arxiv.org/abs/2110.14168 Training verifiers to solve math word problems . Preprint, arXiv:2110.14168

    arXiv 2021

  7. [7]

    Ben Cottier, Robi Rahman, Loredana Fattorini, Nestor Maslej, Tamay Besiroglu, and David Owen. 2025. https://arxiv.org/abs/2405.21015 The rising costs of training frontier ai models . Preprint, arXiv:2405.21015

    arXiv 2025

  8. [8]

    DeepSeek-AI, Daya Guo, Dejian Yang, Haowei Zhang, and Junxiao Song et al. 2025. https://arxiv.org/abs/2501.12948 Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning . Preprint, arXiv:2501.12948

    arXiv 2025

Show all 41 references
  1. [9]

    Mor Geva, Daniel Khashabi, Elad Segal, Tushar Khot, Dan Roth, and Jonathan Berant. 2021. https://arxiv.org/abs/2101.02235 Did aristotle use a laptop? a question answering benchmark with implicit reasoning strategies . Preprint, arXiv:2101.02235

    2021 arXiv

  2. [10]

    Akash Ghosh, Debayan Datta, Sriparna Saha, and Chirag Agarwal. 2025. https://arxiv.org/abs/2502.09457 The multilingual mind : A survey of multilingual reasoning in language models . Preprint, arXiv:2502.09457

    2025

  3. [11]

    Aaron Grattafiori, Abhimanyu Dubey, and Abhinav Jauhri et al. 2024. https://arxiv.org/abs/2407.21783 The llama 3 herd of models . Preprint, arXiv:2407.21783

    2024 arXiv

  4. [12]

    Ke Ji, Jiahao Xu , Tian Liang, Qiuzhi Liu , Zhiwei He , Xingyu Chen , Xiaoyuan Liu , Zhijie Wang , Junying Chen , Benyou Wang , Zhaopeng Tu , Haitao Mi , and Dong Yu. 2025 a . https://doi.org/10.13140/RG.2.2.33772.07043 The first few tokens are all you need: An efficient and e...

    2025

  5. [13]

    Yixin Ji, Juntao Li, Hai Ye, Kaixin Wu, Kai Yao, Jia Xu, Linjian Mo, and Min Zhang. 2025 b . https://arxiv.org/abs/2501.02497 Test-time compute: from system-1 thinking to system-2 thinking . Preprint, arXiv:2501.02497

    2025 arXiv

  6. [14]

    Jones and Benjamin K

    Cameron R. Jones and Benjamin K. Bergen. 2025. https://arxiv.org/abs/2503.23674 Large language models pass the turing test . Preprint, arXiv:2503.23674

    2025 arXiv

  7. [15]

    Ryo Kamoi, Yusen Zhang, Nan Zhang, Jiawei Han, and Rui Zhang. 2024. https://doi.org/10.1162/tacl_a_00713 When can LLM s actually correct their own mistakes? a critical survey of self-correction of LLM s . Transactions of the Association for Computational Linguistics, 12:1417--1440

    2024 doi

  8. [16]

    Vid Kocijan, Thomas Lukasiewicz, Ernest Davis, Gary Marcus, and Leora Morgenstern. 2020. https://arxiv.org/abs/2004.13831 A review of winograd schema challenge datasets and approaches . Preprint, arXiv:2004.13831

    2020 arXiv

  9. [17]

    Aviral Kumar, Vincent Zhuang, Rishabh Agarwal, Yi Su, John D Co-Reyes, Avi Singh, Kate Baumli, Shariq Iqbal, Colton Bishop, Rebecca Roelofs, Lei M Zhang, Kay McKinney, Disha Shrivastava, Cosmin Paduraru, George Tucker, Doina Precup, Feryal Behbahani, and Aleksandra Faust. 2024...

    2024 arXiv

  10. [18]

    Patil, Matei Zaharia, Joseph E

    Dacheng Li, Shiyi Cao, Tyler Griggs, Shu Liu, Xiangxi Mo, Eric Tang, Sumanth Hegde, Kourosh Hakhamaneshi, Shishir G. Patil, Matei Zaharia, Joseph E. Gonzalez, and Ion Stoica. 2025 a . https://arxiv.org/abs/2502.07374 Llms can easily learn to reason from demonstrations structur...

    2025 arXiv

  11. [19]

    Yafu Li, Zhilin Wang, Tingchen Fu, Ganqu Cui, Sen Yang, and Yu Cheng. 2025 b . https://arxiv.org/abs/2501.11877 From drafts to answers: Unlocking llm potential via aggregation fine-tuning . Preprint, arXiv:2501.11877

    2025 arXiv

  12. [20]

    Zhong-Zhi Li, Duzhen Zhang, Ming-Liang Zhang, Jiaxin Zhang, Zengyan Liu, Yuxuan Yao, Haotian Xu, Junhao Zheng, Pei-Jie Wang, Xiuyi Chen, Yingying Zhang, Fei Yin, Jiahua Dong, Zhiwei Li, Bao-Long Bi, Ling-Rui Mei, Junfeng Fang, Zhijiang Guo, Le Song, and Cheng-Lin Liu. 2025 c ....

    2025 arXiv

  13. [21]

    McIlraith, and Yilun Du

    Shalev Lifshitz, Sheila A. McIlraith, and Yilun Du. 2025. https://openreview.net/forum?id=H22e93wnMe Multi-agent verification: Scaling test-time compute with goal verifiers . In Workshop on Reasoning and Planning for Large Language Models

    2025

  14. [22]

    Runze Liu, Junqi Gao, Jian Zhao, Kaiyan Zhang, Xiu Li, Biqing Qi, Wanli Ouyang, and Bowen Zhou. 2025. https://arxiv.org/abs/2502.06703 Can 1b llm surpass 405b llm? rethinking compute-optimal test-time scaling . Preprint, arXiv:2502.06703

    2025 arXiv

  15. [23]

    Marco Lui and Timothy Baldwin. 2011. https://aclanthology.org/I11-1062/ Cross-domain feature selection for language identification . In Proceedings of 5th International Joint Conference on Natural Language Processing, pages 553--561, Chiang Mai, Thailand. Asian Federation of N...

    2011

  16. [24]

    Niklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts, Stella Biderman, Teven Le Scao, M Saiful Bari, Sheng Shen, Zheng-Xin Yong, Hailey Schoelkopf, Xiangru Tang, Dragomir Radev, Alham Fikri Aji, Khalid Almubarak, Samuel Albanie, Zaid Alyafeai, Albert Webson, Edward ...

    2023 arXiv

  17. [25]

    Niklas Muennighoff, Zitong Yang, Weijia Shi, Xiang Lisa Li, Li Fei-Fei, Hannaneh Hajishirzi, Luke Zettlemoyer, Percy Liang, Emmanuel Candès, and Tatsunori Hashimoto. 2025. https://arxiv.org/abs/2501.19393 s1: Simple test-time scaling . Preprint, arXiv:2501.19393

    2025 arXiv

  18. [26]

    Goucher et al

    OpenAI, :, Aaron Hurst, Adam Lerer, and Adam P. Goucher et al. 2024 a . https://arxiv.org/abs/2410.21276 Gpt-4o system card . Preprint, arXiv:2410.21276

    2024 arXiv

  19. [27]

    OpenAI, :, Aaron Jaech, Adam Kalai, Adam Lerer, and Adam Richardson et al. 2024 b . https://arxiv.org/abs/2412.16720 Openai o1 system card . Preprint, arXiv:2412.16720

    2024 arXiv

  20. [28]

    Aske Plaat, Annie Wong, Suzan Verberne, Joost Broekens, Niki van Stein, and Thomas Back. 2024. https://arxiv.org/abs/2407.11511 Reasoning with large language models, a survey . Preprint, arXiv:2407.11511

    2024

  21. [29]

    Edoardo Maria Ponti, Goran Glava s , Olga Majewska, Qianchu Liu, Ivan Vuli \'c , and Anna Korhonen. 2020. https://doi.org/10.18653/v1/2020.emnlp-main.185 XCOPA : A multilingual dataset for causal commonsense reasoning . In Proceedings of the 2020 Conference on Empirical Method...

    2020 doi

  22. [30]

    Libo Qin, Qiguang Chen, Fuxuan Wei, Shijue Huang, and Wanxiang Che. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.163 Cross-lingual prompting: Improving zero-shot chain-of-thought reasoning across languages . In Proceedings of the 2023 Conference on Empirical Methods in Na...

    2023 doi

  23. [31]

    Nils Reimers and Iryna Gurevych. 2019. https://arxiv.org/abs/1908.10084 Sentence-bert: Sentence embeddings using siamese bert-networks . In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics

    2019 arXiv

  24. [32]

    Nils Reimers and Iryna Gurevych. 2020. https://arxiv.org/abs/2004.09813 Making monolingual sentence embeddings multilingual using knowledge distillation . In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing. Association for Computational L...

    2020 arXiv

  25. [33]

    Nikunj Saunshi, Nishanth Dikkala, Zhiyuan Li, Sanjiv Kumar, and Sashank J. Reddi. 2025. https://arxiv.org/abs/2502.17416 Reasoning with latent thoughts: On the power of looped transformers . Preprint, arXiv:2502.17416

    2025 arXiv

  26. [34]

    Freda Shi, Mirac Suzgun, Markus Freitag, Xuezhi Wang, Suraj Srivats, Soroush Vosoughi, Hyung Won Chung, Yi Tay, Sebastian Ruder, Denny Zhou, Dipanjan Das, and Jason Wei. 2022. https://arxiv.org/abs/2210.03057 Language models are multilingual chain-of-thought reasoners . Prepri...

    2022 arXiv

  27. [35]

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

    2024 arXiv

  28. [36]

    Fengwei Teng, Zhaoyang Yu, Quan Shi, Jiayi Zhang, Chenglin Wu, and Yuyu Luo. 2025. https://arxiv.org/abs/2502.12018 Atom of thoughts for markov llm test-time scaling . Preprint, arXiv:2502.12018

    2025

  29. [37]

    Pablo Villalobos, Anson Ho, Jaime Sevilla, Tamay Besiroglu, Lennart Heim, and Marius Hobbhahn. 2024. https://arxiv.org/abs/2211.04325 Will we run out of data? limits of llm scaling based on human-generated data . Preprint, arXiv:2211.04325

    2024 arXiv

  30. [38]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed Chi, Quoc Le, and Denny Zhou. 2023. https://arxiv.org/abs/2201.11903 Chain-of-thought prompting elicits reasoning in large language models . Preprint, arXiv:2201.11903

    2023 arXiv

  31. [39]

    Tian Xie, Zitian Gao, Qingnan Ren, Haoming Luo, Yuqian Hong, Bryan Dai, Joey Zhou, Kai Qiu, Zhirong Wu, and Chong Luo. 2025. https://arxiv.org/abs/2502.14768 Logic-rl: Unleashing llm reasoning with rule-based reinforcement learning . Preprint, arXiv:2502.14768

    2025 arXiv

  32. [40]

    Wenkai Yang, Shuming Ma, Yankai Lin, and Furu Wei. 2025. https://arxiv.org/abs/2502.18080 Towards thinking-optimal scaling of test-time compute for llm reasoning . Preprint, arXiv:2502.18080

    2025

  33. [41]

    Qiyuan Zhang, Fuyuan Lyu, Zexu Sun, Lei Wang, Weixu Zhang, Wenyue Hua, Haolun Wu, Zhihan Guo, Yufei Wang, Niklas Muennighoff, Irwin King, Xue Liu, and Chen Ma. 2025. https://arxiv.org/abs/2503.24235 A survey on test-time scaling in large language models: What, how, where, and ...

    2025 arXiv

Pith tools

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