Pith. sign in

REVIEW 3 major objections 6 minor 41 references

Stabilizing Reasoning in Medical LLMs with Continued Pretraining and Reasoning Preference Optimization

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Fine-tuning Qwen2.5-72B with continued pretraining followed by reasoning preference optimization produces a Japanese medical model that holds a 0.868 accuracy whether or not it is prompted to explain its answers, while baselines fall by…

desk verdict Useful, honest ablation showing RPO beats DPO for keeping accuracy under explanation prompts in Japanese medical QA, but the 'stable reasoning' claim is not backed by any check on the explanations themselves. read the letter →

arxiv 2504.18080 v1 pith:5J3NGKAF submitted 2025-04-25 cs.CL cs.AIcs.LG

Wataru Kawakami , Keita Suzuki , Junichiro Iwasawa This is my paper · ORCID
classification cs.CLcs.AIcs.LG
keywords JapanesemedicalLLMreasoningpreferenceoptimizationcontinuedpretrainingIgakuQAexplanationstabilityquestionansweringlicensingexam
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

Large language models used in medicine often answer accurately when asked a direct question, but their accuracy drops when asked to show their reasoning first; that instability is a barrier to clinical trust. This paper proposes a two-stage fine-tuning recipe — continued pretraining on Japanese medical exam material followed by reasoning preference optimization (RPO) — applied to a 72-billion-parameter model. The resulting Preferred-MedLLM-Qwen-72B scores 0.868 on the IgakuQA Japanese medical licensing benchmark and, unlike the base model and a continued-pretraining-only variant, keeps exactly that score when prompted to explain its answers. The paper takes this as evidence that RPO stabilizes reasoning generation while preserving accuracy, arguing that specialized high-stakes domains need optimization for reliable explanations, not just correct answers.

What carries the argument

The load-bearing mechanism is Reasoning Preference Optimization (RPO), a DPO variant that adds a weighted negative log-likelihood term on the chosen response to the standard DPO preference loss. Trained on preference triples built from Japanese Medical Licensing Exam questions, where a curated ground-truth explanation is preferred over a model-generated correct answer and that over a wrong answer, RPO pushes the model toward high-quality reasoning pathways while keeping the chosen response's likelihood high. It is applied with QLoRA after a continued-pretraining stage on the same Japanese medical corpus, and the paper's ablation shows that RPO, not DPO, is the component that erases the explanation-prompt accuracy drop.

What would settle it

Ask two independent clinicians to judge the medical accuracy and logical coherence of the explanations the final model produces on, say, 100 randomly sampled IgakuQA questions. If the fraction of explanations judged medically sound is clearly below the 86.8% accuracy rate, or if correct answers are regularly paired with confidently wrong rationales, then the paper's claim that RPO stabilizes reasoning is not supported and what is stabilized is only answer selection.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that adding Reasoning Preference Optimization after Continued Pretraining removes the accuracy collapse that otherwise occurs when a Japanese medical model is prompted to generate step-by-step explanations. Concretely, Preferred-MedLLM-Qwen-72B scores 0.868 on IgakuQA under standard three-shot prompting and the identical 0.868 when the prompt requires an explanation, while the continued-pretraining-only model drops from 0.867 to 0.834 and the base Qwen2.5-72B drops from 0.802 to 0.710. The paper also reports that the CPT+RPO model edges out GPT-4o's 0.866 in the standard setting and, unlike a CPT+DPO configuration, shows no drop under explanation prompts. The authors read this as evidence that RPO stabilizes reasoning generation, and they frame the contribution as optimizing for reliable explanations alongside answer accuracy.

Load-bearing premise

The load-bearing premise is that answer accuracy under an explanation-requesting prompt is a valid proxy for reliable and stable reasoning, because the paper never directly checks whether the generated explanations are factually correct, coherent, or faithful to the reasoning that produced the answer.

Editorial extensions

If this is right

  • A medical LLM can be built that does not lose accuracy when clinicians ask for a rationale, removing a practical barrier to clinical adoption.
  • RPO specifically, rather than CPT or DPO alone, is the component that stabilizes reasoning; CPT+DPO still dropped from 0.868 to 0.848 under explanation prompts.
  • The CPT+RPO combination generalizes zero-shot to other Japanese medical QA tasks (MedQA, MedMCQA, PubMedQA, MMMLU-med), raising the average score from 0.710 to 0.716.
  • The stabilization is achievable on a 72B model with quantized low-rank adaptation, suggesting the recipe can scale without full fine-tuning.
  • Releasing the model weights lets others reproduce and build on the stabilized reasoning behavior in the Japanese medical domain.

Reading between the lines

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

  • The paper measures stability only through answer accuracy under explanation prompts; a natural next test is whether the explanations themselves are factually correct and coherent, since stable answer selection with misleading rationales would not constitute trustworthy reasoning.
  • GPT-4o actually improves its score to 0.881 under explanation prompts, so the method's distinctive contribution is to open-weight, domain-adapted models rather than to beating all generalist proprietary models on explanation robustness.
  • The stabilization may come from RPO's NLL term anchoring the explanation distribution to curated ground-truth explanations; a testable hypothesis is that the effect grows with the NLL weight, which the paper sets to 10.
  • The recipe is language- and domain-specific but could transfer to other non-English medical settings or other high-stakes fields such as law or finance, as long as a comparable corpus and a preference hierarchy over explanations can be constructed.
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

3 major / 6 minor

Summary. The paper proposes a two-stage fine-tuning recipe for a Japanese medical LLM: continued pretraining (CPT) on Japanese medical texts, followed by Reasoning Preference Optimization (RPO), a DPO variant with an additional NLL term. The resulting model, Preferred-MedLLM-Qwen-72B, is evaluated on the IgakuQA benchmark under standard 3-shot prompting and under a 3-shot prompt that additionally requests explanations. The headline result is that the model achieves 0.868 accuracy in both conditions, whereas the CPT-only model drops from 0.867 to 0.834 and the base model drops from 0.802 to 0.710. The authors interpret this as evidence that RPO stabilizes reasoning generation while preserving accuracy, and they also report results on other Japanese medical QA benchmarks. The model weights are released.

Significance. If the central claim is supported, this is a useful empirical contribution: it shows that a preference-optimization stage can prevent the accuracy drop that many models exhibit when asked to produce explanations, and it does so in a non-English, high-stakes domain with a publicly released 72B model. The ablation design is informative, separating CPT, RPO, DPO, and the combined pipeline, and the evaluation is on an external public benchmark rather than on a bespoke metric. The main significance, however, depends on whether the stability measure actually captures reasoning quality rather than just answer accuracy under a different prompt format; the manuscript currently does not supply that evidence.

major comments (3)
  1. [Section 4.1-4.2, Tables 1-2] The central claim that RPO 'stabilizes reasoning generation' is supported only by answer accuracy under a prompt that requests explanations. The manuscript does not report whether the models in the '3-shot w/ explanation' condition actually produce step-by-step explanations, nor does it assess the length, coherence, prompt-compliance, or factual correctness of the generated explanations. The 0.868/0.868 result is therefore compatible with the RPO model suppressing or ignoring the explanation request, and the observation that GPT-4o improves from 0.866 to 0.881 in the explanation condition makes prompt non-compliance a concrete alternative explanation. This is load-bearing because the paper's stated contribution is trustworthy reasoning, not merely stable answer accuracy. I ask for output-level analyses (e.g., compliance rate, explanation length, expert or LLM-judged quality, and error analysis) or, failing that, for the claim to be narrowed to 'stable answer accuracy under explanation-requesting prompts.'
  2. [Tables 1-2, Section 4.1] The paper reports no confidence intervals, significance tests, or repeated evaluation runs. The SOTA claim over GPT-4o rests on a difference of 0.002 (0.868 vs. 0.866), which is within the range of sampling noise for a five-exam benchmark, and GPT-4o's score in the explanation condition (0.881) is actually higher than the proposed model's 0.868. Additionally, the ablation shows CPT+RPO and CPT+DPO both at 0.868 in the standard setting, so the distinguishing advantage rests entirely on the explanation-condition numbers. Please report per-year scores, bootstrap confidence intervals, or a significance test to establish that the observed differences are not artifacts of a small number of questions.
  3. [Section 4.2, ablation study] The comparison between CPT+RPO and CPT+DPO is described only as 'a comparison using standard DPO instead of RPO,' with no details on whether the same preference pairs, the same preference hierarchy, the same number of epochs, and the same hyperparameters were used. Since the paper attributes the stabilization benefit to RPO's additional NLL term, the DPO control must be matched on all other factors; otherwise the difference could be due to preference-data construction or tuning rather than to the loss function. Please specify the exact DPO configuration and, if possible, include ablations over the RPO alpha value.
minor comments (6)
  1. [Sections 3.1-3.2] Please report the size and composition of the CPT corpus and the RPO preference dataset; the current description is qualitative, which makes it difficult to assess data diversity and to reproduce the pipeline.
  2. [Table 3] The MMMLU(med, jp) column is an average over five subsets; please provide the per-subset scores, since the paper claims the model achieves 'the highest scores on the Japanese medical related tasks of MMMLU.'
  3. [Table 1, footnotes] For the proprietary models, please state the API version and evaluation date; this is important for fair comparison with a fixed public checkpoint like Qwen2.5-72B.
  4. [Abstract and Section 4.1] The percentage drops of 11.5% and 3.8% are relative to the base-condition scores; please state this explicitly to avoid confusion with absolute percentage-point changes.
  5. [Section 3.2] RPO is introduced as 'Iterative Reasoning Preference Optimization' by Pang et al., but the paper applies it for one epoch and does not describe any iterative loop; please clarify whether this is the original iterative algorithm or a non-iterative adaptation, since that affects how the method should be cited and understood.
  6. [Appendix A] Please include the exact prompt templates for both the 3-shot and 3-shot-with-explanation conditions; the single example in Table A.1 is useful but does not show how the prompt is assembled with the three examples.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the reported accuracy and stability results are direct measurements on an external benchmark, not derived from fitted inputs or self-citations.

full rationale

The paper's central claims are empirical measurements on the held-out IgakuQA benchmark (2018-2022), against models trained only on JMLE material up to 2017. No equation, fitted parameter, or prior result of the authors is used to define the reported 0.868 scores; the RPO objective is cited from Pang et al. (2024) and applied with a standard TRL implementation, and the evaluation metric is answer accuracy, not a quantity constructed from the training data. The preference dataset and CPT corpus are drawn from the same exam family but a different year range, which is domain adaptation rather than circular derivation. The paper's own Section 5 limitations acknowledge the MCQ-only scope and the absence of human feedback in preference data; the skeptic's worry that answer accuracy under the explanation condition may not fully measure reasoning quality is a construct-validity or correctness concern, not circularity, because no reduction by definition or by self-citation is present. No circular steps were identified.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central result rests on a handful of domain assumptions about the benchmark and the preference data, plus one hand-chosen loss weight. No new entities are introduced. The assumptions are reasonable for this type of empirical paper, but the preference-quality assumption is untested.

free parameters (3)
  • RPO alpha (weight for NLL loss) = 10
    Chosen by hand; weights the negative log-likelihood term relative to the DPO loss in the RPO loss (Section 3.2). Not swept or motivated by data, and it directly affects the training objective.
  • Number of CPT epochs = 2
    Set by the authors for stage 1; no sweep or justification is provided (Section 3.1).
  • Number of RPO epochs = 1
    Set by the authors for stage 2; no sweep or justification is provided (Section 3.2).
assumptions (4)
  • domain assumption IgakuQA is a valid and representative benchmark for Japanese medical reasoning ability.
    The paper uses IgakuQA (2018-2022) as the primary yardstick for both accuracy and stability claims (Section 4.1). If this benchmark is flawed or unrepresentative, the central result weakens.
  • domain assumption The official JMLE explanations used as ground-truth preferences are higher-quality reasoning than the CPT model's generated explanations.
    The RPO preference hierarchy ranks ground-truth responses above model-generated chosen responses (Section 3.2); the paper does not validate this assumption with human ratings.
  • domain assumption Training on JMLE materials up to 2017 does not leak into the IgakuQA benchmark covering 2018-2022.
    The paper asserts the year cutoff prevents leakage (Section 3.1), but does not check for systematic similarity between the 2017 and 2018 exams that could give an unfair advantage.
  • domain assumption Qwen2.5-72B is a strong multilingual base model suitable for Japanese medical adaptation.
    The choice is justified by general benchmarks (Section 3.1), not by an analysis of Japanese medical competence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Stabilizing Reasoning in Medical LLMs with Continued Pretraining and Reasoning Preference Optimization." pith.science (2026). https://pith.science/paper/5J3NGKAF

@misc{pith2026250418080,
  author       = {Pith},
  title        = {Pith review of: Stabilizing Reasoning in Medical LLMs with Continued Pretraining and Reasoning Preference Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5J3NGKAF}},
  note         = {Machine review of arXiv:2504.18080}
}
read the original abstract

Large Language Models (LLMs) show potential in medicine, yet clinical adoption is hindered by concerns over factual accuracy, language-specific limitations (e.g., Japanese), and critically, their reliability when required to generate reasoning explanations -- a prerequisite for trust. This paper introduces Preferred-MedLLM-Qwen-72B, a 72B-parameter model optimized for the Japanese medical domain to achieve both high accuracy and stable reasoning. We employ a two-stage fine-tuning process on the Qwen2.5-72B base model: first, Continued Pretraining (CPT) on a comprehensive Japanese medical corpus instills deep domain knowledge. Second, Reasoning Preference Optimization (RPO), a preference-based method, enhances the generation of reliable reasoning pathways while preserving high answer accuracy. Evaluations on the Japanese Medical Licensing Exam benchmark (IgakuQA) show Preferred-MedLLM-Qwen-72B achieves state-of-the-art performance (0.868 accuracy), surpassing strong proprietary models like GPT-4o (0.866). Crucially, unlike baseline or CPT-only models which exhibit significant accuracy degradation (up to 11.5\% and 3.8\% respectively on IgakuQA) when prompted for explanations, our model maintains its high accuracy (0.868) under such conditions. This highlights RPO's effectiveness in stabilizing reasoning generation. This work underscores the importance of optimizing for reliable explanations alongside accuracy. We release the Preferred-MedLLM-Qwen-72B model weights to foster research into trustworthy LLMs for specialized, high-stakes applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 6 canonical work pages

  1. [1]

    Brodeur, Thomas A

    Peter G. Brodeur, Thomas A. Buckley, Zahir Kanjee, Ethan Goh, Evelyn Bin Ling, Priyank Jain, Stephanie Cabral, Raja-Elie Abdulnour, Adrian Haimovich, Jason A. Freed, Andrew Olson, Daniel J. Morgan, Jason Hom, Robert Gallo, Eric Horvitz, Jonathan Chen, Arjun K. Manrai, and Adam Rodman. Superhuman performance of a large language model on the reasoning tasks...

    arXiv 2024

  2. [2]

    Huatuogpt-o1, towards medical complex reasoning with llms

    Junying Chen, Zhenyang Cai, Ke Ji, Xidong Wang, Wanlong Liu, Rongsheng Wang, Jianye Hou, and Benyou Wang. Huatuogpt-o1, towards medical complex reasoning with llms. arXiv preprint arXiv:2412.18925, 2024. URL https://arxiv.org/abs/2412.18925

    arXiv 2024

  3. [3]

    Meditron-70b: Scaling medical pretraining for large language models

    Zeming Chen, Alejandro Hernández Cano, Angelika Romanou, Antoine Bonnet, Kyle Matoba, Francesco Salvi, Matteo Pagliardini, Simin Fan, Andreas Köpf, Amirkeivan Mohtashami, Alexandre Sallinen, Alireza Sakhaeirad, Vinitra Swamy, Igor Krawczuk, Deniz Bayazit, Axel Marmet, Syrielle Montariol, Mary-Anne Hartley, Martin Jaggi, and Antoine Bosselut. Meditron-70b:...

    arXiv 2023

  4. [4]

    Beyond fine-tuning: Unleashing the potential of continuous pretraining for clinical LLM s

    Clement Christophe, Tathagata Raha, Svetlana Maslenkova, Muhammad Umar Salman, Praveenkumar Kanithi, Marco AF Pimentel, and Shadab Khan. Beyond fine-tuning: Unleashing the potential of continuous pretraining for clinical LLM s. In Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen, editors, Findings of the Association for Computational Linguistics: EMNLP 2...

    work page doi:10.18653/v1/2024.findings-emnlp.618 2024

  5. [5]

    Furlotte , Zhun Yang , Chace Lee , Erik Schenck , Yojan Patel , Jian Cui , Logan Douglas Schneider , Robby Bryant , Ryan G

    Justin Cosentino , Anastasiya Belyaeva , Xin Liu , Nicholas A. Furlotte , Zhun Yang , Chace Lee , Erik Schenck , Yojan Patel , Jian Cui , Logan Douglas Schneider , Robby Bryant , Ryan G. Gomes , Allen Jiang , Roy Lee , Yun Liu , Javier Perez , Jameson K. Rogers , Cathy Speed , Shyam Tailor , Megan Walker , Jeffrey Yu , Tim Althoff , Conor Heneghan , John ...

    arXiv doi:10.48550/arxiv.2406.06474 2024

  6. [6]

    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...

    arXiv doi:10.48550/arxiv.2501.12948 2025

  7. [7]

    The case for 4-bit precision: k-bit Inference Scaling Laws

    Tim Dettmers and Luke Zettlemoyer . The case for 4-bit precision: k-bit Inference Scaling Laws . arXiv e-prints, art. arXiv:2212.09720, December 2022. doi:10.48550/arXiv.2212.09720

    arXiv doi:10.48550/arxiv.2212.09720 2022

  8. [8]

    Qlora: Efficient finetuning of quantized llms

    Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. Qlora: Efficient finetuning of quantized llms. Advances in neural information processing systems, 36: 0 10088--10115, 2023

    2023

Show all 41 references
  1. [9]

    Does fine-tuning llms on new knowledge encourage hallucinations? arXiv preprint arXiv:2405.05904, 2024

    Zorik Gekhman, Gal Yona, Roee Aharoni, Matan Eyal, Amir Feder, Roi Reichart, and Jonathan Herzig. Does fine-tuning llms on new knowledge encourage hallucinations? arXiv preprint arXiv:2405.05904, 2024

    2024 arXiv

  2. [10]

    The Llama 3 Herd of Models

    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 , A...

    2024 doi arXiv

  3. [11]

    Suchin Gururangan, Ana Marasovi \'c , Swabha Swayamdipta, Kyle Lo, Iz Beltagy, Doug Downey, and Noah A. Smith. Don`t stop pretraining: Adapt language models to domains and tasks. In Dan Jurafsky, Joyce Chai, Natalie Schluter, and Joel Tetreault, editors, Proceedings of the 58t...

    2020 doi

  4. [12]

    Measuring Massive Multitask Language Understanding

    Dan Hendrycks , Collin Burns , Steven Basart , Andy Zou , Mantas Mazeika , Dawn Song , and Jacob Steinhardt . Measuring Massive Multitask Language Understanding . arXiv e-prints, art. arXiv:2009.03300, September 2020. doi:10.48550/arXiv.2009.03300

    2009 doi arXiv

  5. [13]

    Lora: Low-rank adaptation of large language models

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models. arxiv 2021. arXiv preprint arXiv:2106.09685, 2021

    2021 arXiv

  6. [14]

    JMedBench: A Benchmark for Evaluating Japanese Biomedical Large Language Models

    Junfeng Jiang , Jiahao Huang , and Akiko Aizawa . JMedBench: A Benchmark for Evaluating Japanese Biomedical Large Language Models . arXiv e-prints, art. arXiv:2409.13317, September 2024. doi:10.48550/arXiv.2409.13317

    2024 doi arXiv

  7. [15]

    What Disease does this Patient Have? A Large-scale Open Domain Question Answering Dataset from Medical Exams

    Di Jin , Eileen Pan , Nassim Oufattole , Wei-Hung Weng , Hanyi Fang , and Peter Szolovits . What Disease does this Patient Have? A Large-scale Open Domain Question Answering Dataset from Medical Exams . arXiv e-prints, art. arXiv:2009.13081, September 2020. doi:10.48550/arXiv....

    2009 doi arXiv

  8. [16]

    P ub M ed QA : A dataset for biomedical research question answering

    Qiao Jin, Bhuwan Dhingra, Zhengping Liu, William Cohen, and Xinghua Lu. P ub M ed QA : A dataset for biomedical research question answering. In Kentaro Inui, Jing Jiang, Vincent Ng, and Xiaojun Wan, editors, Proceedings of the 2019 Conference on Empirical Methods in Natural La...

    2019 doi

  9. [17]

    An evaluation framework for clinical use of large language models in patient interaction tasks

    Shreya Johri, Jaehwan Jeong, Benjamin A Tran, Daniel I Schlessinger, Shannon Wongvibulsin, Leandra A Barnes, Hong-Yu Zhou, Zhuo Ran Cai, Eliezer M Van Allen, David Kim, et al. An evaluation framework for clinical use of large language models in patient interaction tasks. Natur...

    2025

  10. [18]

    MEDIC: Towards a Comprehensive Framework for Evaluating LLMs in Clinical Applications

    Praveen K Kanithi , Cl \'e ment Christophe , Marco AF Pimentel , Tathagata Raha , Nada Saadi , Hamza Javed , Svetlana Maslenkova , Nasir Hayat , Ronnie Rajan , and Shadab Khan . MEDIC: Towards a Comprehensive Framework for Evaluating LLMs in Clinical Applications . arXiv e-pri...

    2024 doi arXiv

  11. [19]

    Evaluating gpt-4 and chatgpt on japanese medical licensing examinations, 2023

    Jungo Kasai, Yuhei Kasai, Keisuke Sakaguchi, Yutaro Yamada, and Dragomir Radev. Evaluating gpt-4 and chatgpt on japanese medical licensing examinations, 2023. URL https://arxiv.org/abs/2303.18027

    2023 arXiv

  12. [20]

    Medical Hallucinations in Foundation Models and Their Impact on Healthcare

    Yubin Kim , Hyewon Jeong , Shan Chen , Shuyue Stella Li , Mingyu Lu , Kumail Alhamoud , Jimin Mun , Cristina Grau , Minseok Jung , Rodrigo Gameiro , Lizhou Fan , Eugene Park , Tristan Lin , Joonsik Yoon , Wonjin Yoon , Maarten Sap , Yulia Tsvetkov , Paul Liang , Xuhai Xu , Xin...

    2025 doi

  13. [21]

    ReFT: Reasoning with Reinforced Fine-Tuning

    Trung Quoc Luong , Xinbo Zhang , Zhanming Jie , Peng Sun , Xiaoran Jin , and Hang Li . ReFT: Reasoning with Reinforced Fine-Tuning . arXiv e-prints, art. arXiv:2401.08967, January 2024. doi:10.48550/arXiv.2401.08967

    2024 doi arXiv

  14. [22]

    Radlink: Linking clinical entities from radiology reports

    Yongli Mou, Hanbin Chen, Gwendolyn Isabella Lode, Daniel Truhn, Sulayman Sowe, and Stefan Decker. Radlink: Linking clinical entities from radiology reports. In 2024 2nd International Conference on Foundation and Large Language Models (FLLM), pages 443--449. IEEE, 2024

    2024

  15. [23]

    Can generalist foundation models outcompete special-purpose tuning? case study in medicine

    Harsha Nori, Yin Tat Lee, Sheng Zhang, Dean Carignan, Richard Edgar, Nicolo Fusi, Nicholas King, Jonathan Larson, Yuanzhi Li, Weishung Liu, Renqian Luo, Scott Mayer McKinney, Robert Osazuwa Ness, Hoifung Poon, Tao Qin, Naoto Usuyama, Chris White, and Eric Horvitz. Can generali...

    2023 arXiv

  16. [24]

    Can Generalist Foundation Models Outcompete Special-Purpose Tuning? Case Study in Medicine

    Harsha Nori , Yin Tat Lee , Sheng Zhang , Dean Carignan , Richard Edgar , Nicolo Fusi , Nicholas King , Jonathan Larson , Yuanzhi Li , Weishung Liu , Renqian Luo , Scott Mayer McKinney , Robert Osazuwa Ness , Hoifung Poon , Tao Qin , Naoto Usuyama , Chris White , and Eric Horv...

    2023 doi arXiv

  17. [25]

    From medprompt to o1: Exploration of run-time strategies for medical challenge problems and beyond

    Harsha Nori, Naoto Usuyama, Nicholas King, Scott Mayer McKinney, Xavier Fernandes, Sheng Zhang, and Eric Horvitz. From medprompt to o1: Exploration of run-time strategies for medical challenge problems and beyond. arXiv preprint arXiv:2411.03590, 2024. URL https://arxiv.org/ab...

    2024 arXiv

  18. [26]

    openai/MMMLU , 2024

    OpenAI. openai/MMMLU , 2024. https://huggingface.co/datasets/openai/MMMLU

    2024

  19. [27]

    OpenAI , : , Aaron Hurst , Adam Lerer , Adam P. Goucher , Adam Perelman , Aditya Ramesh , Aidan Clark , AJ Ostrow , Akila Welihinda , Alan Hayes , Alec Radford , Aleksander Madry , Alex Baker-Whitcomb , Alex Beutel , Alex Borzunov , Alex Carney , Alex Chow , Alex Kirillov , Al...

    doi arXiv

  20. [28]

    Training language models to follow instructions with human feedback

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. Advances in neural information processing systems, 35: 0 2...

    2022

  21. [29]

    MedMCQA : A Large-scale Multi-Subject Multi-Choice Dataset for Medical domain Question Answering

    Ankit Pal , Logesh Kumar Umapathi , and Malaikannan Sankarasubbu . MedMCQA : A Large-scale Multi-Subject Multi-Choice Dataset for Medical domain Question Answering . arXiv e-prints, art. arXiv:2203.14371, March 2022. doi:10.48550/arXiv.2203.14371

    2022 doi arXiv

  22. [30]

    Iterative Reasoning Preference Optimization

    Richard Yuanzhe Pang , Weizhe Yuan , Kyunghyun Cho , He He , Sainbayar Sukhbaatar , and Jason Weston . Iterative Reasoning Preference Optimization . arXiv e-prints, art. arXiv:2404.19733, April 2024. doi:10.48550/arXiv.2404.19733

    2024 doi arXiv

  23. [31]

    PLaMo-100B: A Ground-Up Language Model Designed for Japanese Proficiency

    Preferred Elements , : , Kenshin Abe , Kaizaburo Chubachi , Yasuhiro Fujita , Yuta Hirokawa , Kentaro Imajo , Toshiki Kataoka , Hiroyoshi Komatsu , Hiroaki Mikami , Tsuguo Mogami , Shogo Murai , Kosuke Nakago , Daisuke Nishino , Toru Ogawa , Daisuke Okanohara , Yoshihiko Ozaki...

    doi arXiv

  24. [32]

    Qwen2.5 Technical Report

    Qwen , : , An Yang , Baosong Yang , Beichen Zhang , Binyuan Hui , Bo Zheng , Bowen Yu , Chengyuan Li , Dayiheng Liu , Fei Huang , Haoran Wei , Huan Lin , Jian Yang , Jianhong Tu , Jianwei Zhang , Jianxin Yang , Jiaxi Yang , Jingren Zhou , Junyang Lin , Kai Dang , Keming Lu , K...

    2024 doi arXiv

  25. [33]

    Manning , and Chelsea Finn

    Rafael Rafailov , Archit Sharma , Eric Mitchell , Stefano Ermon , Christopher D. Manning , and Chelsea Finn . Direct Preference Optimization: Your Language Model is Secretly a Reward Model . arXiv e-prints, art. arXiv:2305.18290, May 2023. doi:10.48550/arXiv.2305.18290

    2023 doi arXiv

  26. [34]

    Khaled Saab, Tao Tu, Wei-Hung Weng, Ryutaro Tanno, David Stutz, Ellery Wulczyn, Fan Zhang, Tim Strother, Chunjong Park, Elahe Vedadi, Juanma Zambrano Chaves, Szu-Yeu Hu, Mike Schaekermann, Aishwarya Kamath, Yong Cheng, David G. T. Barrett, Cathy Cheung, Basil Mustafa, Anil Pal...

    2024 arXiv

  27. [35]

    Toward expert-level medical question answering with large language models

    Karan Singhal, Tao Tu, Juraj Gottweis, Rory Sayres, Ellery Wulczyn, Mohamed Amin, Le Hou, Kevin Clark, Stephen R Pfohl, Heather Cole-Lewis, et al. Toward expert-level medical question answering with large language models. Nature Medicine, pages 1--8, 2025

    2025

  28. [36]

    Development and bilingual evaluation of japanese medical large language model within reasonably low computational resources

    Issey Sukeda. Development and bilingual evaluation of japanese medical large language model within reasonably low computational resources. arXiv preprint arXiv:2409.11783, 2024. URL https://arxiv.org/abs/2409.11783

    2024 arXiv

  29. [37]

    70B-parameter large language models in Japanese medical question-answering

    Issey Sukeda , Risa Kishikawa , and Satoshi Kodera . 70B-parameter large language models in Japanese medical question-answering . arXiv e-prints, art. arXiv:2406.14882, June 2024. doi:10.48550/arXiv.2406.14882

    2024 doi arXiv

  30. [38]

    Towards conversational diagnostic artificial intelligence

    Tao Tu, Mike Schaekermann, Anil Palepu, Khaled Saab, Jan Freyberg, Ryutaro Tanno, Amy Wang, Brenna Li, Mohamed Amin, Yong Cheng, et al. Towards conversational diagnostic artificial intelligence. Nature, pages 1--9, 2025

    2025

  31. [39]

    Adapted large language models can outperform medical experts in clinical text summarization

    Dave Van Veen, Cara Van Uden, Louis Blankemeier, Jean-Benoit Delbrouck, Asad Aali, Christian Bluethgen, Anuj Pareek, Malgorzata Polacin, Eduardo Pontes Reis, Anna Seehofnerov \'a , et al. Adapted large language models can outperform medical experts in clinical text summarizati...

    2024

  32. [40]

    A preliminary study of o1 in medicine: Are we closer to an ai doctor? arXiv preprint arXiv:2409.15277, 2024

    Yunfei Xie, Juncheng Wu, Haoqin Tu, Siwei Yang, Bingchen Zhao, Yongshuo Zong, Qiao Jin, Cihang Xie, and Yuyin Zhou. A preliminary study of o1 in medicine: Are we closer to an ai doctor? arXiv preprint arXiv:2409.15277, 2024. URL https://arxiv.org/abs/2409.15277

    2024 arXiv

  33. [41]

    Huatuogpt, towards taming language model to be a doctor

    Hongbo Zhang, Junying Chen, Feng Jiang, Fei Yu, Zhihong Chen, Jianquan Li, Guiming Chen, Xiangbo Wu, Zhiyi Zhang, Qingying Xiao, Xiang Wan, Benyou Wang, and Haizhou Li. Huatuogpt, towards taming language model to be a doctor. arXiv preprint arXiv:2305.15075, 2023. URL https://...

    2023 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.