pith. sign in

arxiv: 2605.31272 · v1 · pith:BXLJTXXWnew · submitted 2026-05-29 · 💻 cs.LG

Algorithmic Recourse of In-Context Learning for Tabular Data

Wenshuo Dong , Jiaming Zhang , Shaopneg Fu , Hongbin Lin , Di Wang , Lijie Hu This is my paper

Pith reviewed 2026-06-28 23:23 UTC · model grok-4.3

classification 💻 cs.LG
keywords algorithmic recoursein-context learningtabular datablack-box optimizationzeroth-order methodsconvergence analysismulti-class classification
0
0 comments X

The pith

Recourse for in-context learning on tabular data remains well-defined, bounded, and converges to classical solutions as context size grows.

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

The paper studies how to generate actionable input changes that flip predictions from in-context learning models applied to tabular data without retraining. It provides a theoretical analysis establishing that recourse stays well-defined and bounded under ICL, with explicit characterization of convergence toward the recourse obtained from classically trained models as the number of in-context examples increases. To make this practical, the authors introduce a zeroth-order method called ASR-ICL that finds sparse, actionable recourse by searching in adaptive subspaces while querying the black-box model only a limited number of times. Experiments on real datasets confirm that the generated recourse matches the quality of prior methods yet requires fewer queries and that the predicted convergence occurs.

Core claim

The paper establishes that algorithmic recourse remains well-defined and bounded for in-context learning models on tabular data, and characterizes the convergence of this recourse toward classical solutions as context size increases; it further supplies the ASR-ICL zeroth-order framework that produces actionable and sparse recourse for black-box ICL predictors, with natural extension to multi-class tasks.

What carries the argument

Adaptive Subspace Recourse for In-Context Learning (ASR-ICL), a zeroth-order black-box optimization procedure that restricts search to adaptive low-dimensional subspaces to produce sparse recourse actions.

If this is right

  • Recourse can be computed for black-box ICL predictors without access to internal gradients or model weights.
  • Larger context sizes produce recourse that more closely matches what would be recommended by a trained model on the same data.
  • The same framework applies to multi-class tabular prediction tasks without modification.
  • Fewer model queries suffice to reach recourse quality comparable to gradient-based or white-box methods.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the convergence result holds, then ICL could serve as a drop-in replacement for trained models in recourse pipelines once context is sufficiently large.
  • The subspace adaptation technique might lower query budgets for other black-box optimization tasks on tabular inputs beyond recourse.
  • Checking whether boundedness persists when ICL is applied to non-tabular modalities would test the scope of the theoretical claim.

Load-bearing premise

That standard definitions of recourse and zeroth-order optimization can be applied directly to ICL models on tabular data without additional assumptions specific to how the context is formed or how the language model processes it.

What would settle it

An experiment in which the recourse actions obtained from an ICL model with increasing context size fail to approach the recourse actions obtained from a classically trained model on the identical tabular dataset and task.

Figures

Figures reproduced from arXiv: 2605.31272 by Di Wang, Hongbin Lin, Jiaming Zhang, Lijie Hu, Shaopneg Fu, Wenshuo Dong.

Figure 1
Figure 1. Figure 1: Effect of the number of context instances on recourse quality under ICL. Increasing the number of instances improves stability and reduces recourse cost. validity. In contrast, ASR-ICL achieves consistently strong validity, including both tabular ICL models and general LLMs, while maintaining a low recourse cost. For exam￾ple, on Australian Credit, ASR-ICL attains perfect validity with TabPFN at an average… view at source ↗
Figure 2
Figure 2. Figure 2: Stability of ASR-ICL under context resampling. We repeat experiments with independently sampled 32-shot context instances while fixing test instances. Bars show mean validity and average recourse cost, with error bars indicating standard deviation. number of in-context examples increases. Our analysis is motivated by Lemma 4.4 and Theorem 4.6, which sug￾gest that as the context size grows, model behave inc… view at source ↗
Figure 3
Figure 3. Figure 3: Validity versus query budget across different tasks (Aus￾tralian Credit and Corporate Rating). Each point corresponds to a model–method pair, and vertical dashed lines indicate the mean query budget required by each method. ASR-ICL consis￾tently achieves comparable validity with substantially fewer model queries. See Appendix F.1 for additional results. 6.5. Ablation and Sensitive Analysis We provide addit… view at source ↗
Figure 4
Figure 4. Figure 4: Prompt template used for model. Each query consists of an instruction prompt, m labeled demonstrations, and a query instance. output exactly one class label. F. Additional Experimental Results F.1. Budget and Query Efficiency [PITH_FULL_IMAGE:figures/full_fig_p020_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Final feature concentration on Australian Credit (32 shots). ASR-ICL yields lower effective feature counts, indicating stronger focus on a small subset of important feature compared to Full ZO [PITH_FULL_IMAGE:figures/full_fig_p022_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Final feature concentration on Corporate Rating (48 shots). ASR-ICL remains consistently more focused than full-space baselines, demonstrating that adaptive subspace search extends naturally beyond binary settings. 0.001 0.01 0.05 0.1 0.2 0.3 0.5 0.0 0.2 0.4 0.6 0.8 1.0 Validity Validity vs 0.001 0.01 0.05 0.1 0.2 0.3 0.5 3.0 3.5 4.0 4.5 Average Cost Cost vs LLaMA-3.1-8B LLaMA-3.2-3B Qwen3-4B-Instruct Qwen… view at source ↗
Figure 7
Figure 7. Figure 7: Sensitivity of ASR-ICL to the trade-off parameter λ on Australian Credit across different models. 1 2 3 4 5 6 7 Subspace size k 0.0 0.2 0.4 0.6 0.8 1.0 Validity Validity Average Cost 2 3 4 5 Average Cost Sensitivity to Subspace Size k [PITH_FULL_IMAGE:figures/full_fig_p023_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Sensitivity of ASR-ICL to the subspace size k on a representative model. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_8.png] view at source ↗
read the original abstract

As predictive models are increasingly deployed in high-stakes settings such as credit approval, there is a growing need for post-hoc methods that provide recourse to affected individuals. Many such models operate on tabular data, where features correspond to real-world attributes. Recently, in-context learning (ICL) has enabled large language models to perform tabular prediction by conditioning on labeled examples at inference time, without explicit training. However, algorithmic recourse for tabular decision-making under ICL remains largely unexplored. In this work, we present the first study of algorithmic recourse for tabular data under ICL. We carry out a theoretical analysis, showing that recourse remains well-defined and bounded, and we characterize how recourse converges toward classical solutions as the context size increases. In practice, we propose a novel zeroth-order recourse framework, Adaptive Subspace Recourse for In-Context Learning (ASR-ICL), that efficiently generates actionable and sparse recourse for black-box ICL models. The proposed framework naturally extends to multi-class tabular tasks. Experiments across multiple real-world datasets and models demonstrate that ASR-ICL achieves recourse quality comparable to existing methods with fewer queries and empirically confirm the predicted convergence behavior, supporting our theoretical analysis.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 2 minor

Summary. The paper presents the first study of algorithmic recourse for tabular data under in-context learning (ICL). It carries out a theoretical analysis showing that recourse remains well-defined and bounded for ICL predictors (modeled as functions of context examples) and characterizes convergence toward classical recourse solutions as context size increases. It proposes the Adaptive Subspace Recourse for In-Context Learning (ASR-ICL) zeroth-order framework to generate actionable and sparse recourse for black-box ICL models, with a natural extension to multi-class tasks. Experiments across multiple real-world datasets and models demonstrate that ASR-ICL achieves recourse quality comparable to existing methods with fewer queries while empirically confirming the predicted convergence behavior.

Significance. If the theoretical results and empirical findings hold, this work is significant as the first exploration of recourse under ICL for tabular data in high-stakes settings. The theoretical contributions establishing well-definedness, boundedness, and convergence to classical solutions provide a foundation for the area. The ASR-ICL framework offers a practical, query-efficient zeroth-order approach for black-box models. Credit is due for the convergence characterization, the modeling choices that directly support treating ICL recourse within the proposed framework, and the supporting experiments that validate both the theory and practical performance.

minor comments (2)
  1. The description of how the adaptive subspace is constructed and updated in ASR-ICL could be expanded with pseudocode or a clearer algorithmic outline to improve reproducibility.
  2. Figure captions and axis labels in the experimental section would benefit from explicit mention of the number of queries used by each baseline for direct comparison.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive review, recognition of the theoretical contributions on well-definedness, boundedness, and convergence, and the recommendation for minor revision. We appreciate the acknowledgment of the practical value of the ASR-ICL framework.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The abstract and skeptic summary describe a theoretical analysis establishing that recourse is well-defined and bounded for ICL predictors (modeled as functions of context), plus a convergence result to classical recourse as context size grows, and an independent zeroth-order ASR-ICL framework validated by experiments. No equations, fitting procedures, or self-citation chains are visible that reduce any load-bearing claim to its own inputs by construction. The modeling choices directly support treating ICL recourse as amenable to the proposed analysis and method without self-referential definitions or renamed fits. This is the most common honest finding when no explicit reduction is exhibited.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Insufficient information in the abstract to identify any free parameters, axioms, or invented entities.

pith-pipeline@v0.9.1-grok · 5755 in / 1322 out tokens · 31712 ms · 2026-06-28T23:23:15.010041+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

51 extracted references · 18 canonical work pages · 6 internal anchors

  1. [1]

    Transformers learn to implement preconditioned gradient descent for in-context learning

    Ahn, K., Cheng, X., Daneshmand, H., and Sra, S. Transformers learn to implement preconditioned gradient descent for in-context learning. Advances in Neural Information Processing Systems, 36: 0 45614--45650, 2023

    2023

  2. [2]

    Diabetes dataset

    Akturk, M. Diabetes dataset. https://www.kaggle.com/datasets/mathchi/diabetes-data-set, 2020. Kaggle repository, accessed 2026-01-27

    2020

  3. [3]

    Machine bias

    Angwin, J., Larson, J., Mattu, S., and Kirchner, L. Machine bias. In Ethics of data and analytics, pp.\ 254--264. Auerbach Publications, 2022

    2022

  4. [4]

    and Selbst, A

    Barocas, S. and Selbst, A. D. Big data's disparate impact. Calif. L. Rev., 104: 0 671, 2016

    2016

  5. [5]

    d., Bae, S., Cha, S., and Yun, S.-Y

    Breejen, F. d., Bae, S., Cha, S., and Yun, S.-Y. Fine-tuned in-context learning transformers are excellent tabular data classifiers. arXiv preprint arXiv:2405.13396, 2024

    work page arXiv 2024

  6. [6]

    D., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al

    Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. Language models are few-shot learners. Advances in neural information processing systems, 33: 0 1877--1901, 2020

    1901

  7. [7]

    N., Bhaila, K., Edemacu, K., and Wu, X

    Carey, A. N., Bhaila, K., Edemacu, K., and Wu, X. Dp-tabicl: In-context learning with differentially private tabular data. In 2024 IEEE International Conference on Big Data (BigData), pp.\ 1552--1557. IEEE, 2024

    2024

  8. [8]

    Framing algorithmic recourse for anomaly detection

    Datta, D., Chen, F., and Ramakrishnan, N. Framing algorithmic recourse for anomaly detection. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp.\ 283--293, 2022

    2022

  9. [9]

    Time can invalidate algorithmic recourse

    De Toni, G., Teso, S., Lepri, B., and Passerini, A. Time can invalidate algorithmic recourse. In Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency, pp.\ 89--107, 2025

    2025

  10. [10]

    Robustness implies fairness in causal algorithmic recourse

    Ehyaei, A.-R., Karimi, A.-H., Sch \"o lkopf, B., and Maghsudi, S. Robustness implies fairness in causal algorithmic recourse. In Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency, pp.\ 984--1001, 2023

    2023

  11. [11]

    P., Zhang, W., and Yu, P

    Fang, L., Liu, A., Zhang, H., Zou, H. P., Zhang, W., and Yu, P. S. TABGEN - ICL : Residual-aware in-context example selection for tabular data generation. In Che, W., Nabende, J., Shutova, E., and Pilehvar, M. T. (eds.), Findings of the Association for Computational Linguistics: ACL 2025, pp.\ 20027--20041, Vienna, Austria, July 2025. Association for Comp...

    work page doi:10.18653/v1/2025.findings-acl.1027 2025

  12. [12]

    Understanding and Improving Continuous Adversarial Training for LLMs via In-context Learning Theory

    Fu, S. and Wang, D. Understanding and improving continuous adversarial training for llms via in-context learning theory. arXiv preprint arXiv:2604.12817, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  13. [13]

    Short-length adversarial training helps llms defend long-length jailbreak attacks: Theoretical and empirical evidence

    Fu, S., Ding, L., Zhang, J., and Wang, D. Short-length adversarial training helps llms defend long-length jailbreak attacks: Theoretical and empirical evidence. arXiv preprint arXiv:2502.04204, 2025

    work page arXiv 2025

  14. [14]

    and Lakkaraju, H

    Gao, R. and Lakkaraju, H. On the impact of algorithmic recourse on social segregation. In Krause, A., Brunskill, E., Cho, K., Engelhardt, B., Sabato, S., and Scarlett, J. (eds.), Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pp.\ 10727--10743. PMLR, 23--29 Jul 2023. URL https:...

    2023

  15. [15]

    C., and Schmidt, L

    Gardner, J., Perdomo, J. C., and Schmidt, L. Large scale transfer learning for tabular data via language modeling, 2024. URL https://arxiv. org/abs/2406.12031

    work page arXiv 2024

  16. [16]

    Corporate credit rating dataset

    Gewerc, A. Corporate credit rating dataset. https://www.kaggle.com/datasets/agewerc/corporate-credit-rating, 2019. Kaggle repository, accessed 2026-01-27

    2019

  17. [17]

    The Llama 3 Herd of Models

    Grattafiori, A., Dubey, A., Jauhri, A., Pandey, A., Kadian, A., Al-Dahle, A., Letman, A., Mathur, A., Schelten, A., Vaughan, A., et al. The llama 3 herd of models. arXiv preprint arXiv:2407.21783, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  18. [18]

    TabPFN-2.5: Advancing the State of the Art in Tabular Foundation Models

    Grinsztajn, L., Fl \"o ge, K., Key, O., Birkel, F., Jund, P., Roof, B., J \"a ger, B., Safaric, D., Alessi, S., Hayler, A., et al. Tabpfn-2.5: Advancing the state of the art in tabular foundation models. arXiv preprint arXiv:2511.08667, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  19. [19]

    Tabpfn: A transformer that solves small tabular classification problems in a second

    Hollmann, N., M \"u ller, S., Eggensperger, K., and Hutter, F. Tabpfn: A transformer that solves small tabular classification problems in a second. In International Conference on Learning Representations 2023, 2023

    2023

  20. [20]

    u ller, S., Purucker, L., Krishnakumar, A., K \

    Hollmann, N., M \"u ller, S., Purucker, L., Krishnakumar, A., K \"o rfer, M., Hoo, S. B., Schirrmeister, R. T., and Hutter, F. Accurate predictions on small data with a tabular foundation model. Nature, 01 2025. doi:10.1038/s41586-024-08328-6. URL https://www.nature.com/articles/s41586-024-08328-6

    work page doi:10.1038/s41586-024-08328-6 2025

  21. [21]

    GPT-4o System Card

    Hurst, A., Lerer, A., Goucher, A. P., Perelman, A., Ramesh, A., Clark, A., Ostrow, A., Welihinda, A., Hayes, A., Radford, A., et al. Gpt-4o system card. arXiv preprint arXiv:2410.21276, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  22. [22]

    Learning decision trees and forests with algorithmic recourse

    Kanamori, K., Takagi, T., Kobayashi, K., and Ike, Y. Learning decision trees and forests with algorithmic recourse. arXiv preprint arXiv:2406.01098, 2024

    work page arXiv 2024

  23. [23]

    Learning gradient boosted decision trees with algorithmic recourse

    Kanamori, K., Kobayashi, K., and Takagi, T. Learning gradient boosted decision trees with algorithmic recourse. Advances in Neural Information Processing Systems, 38: 0 76509--76542, 2026

    2026

  24. [24]

    u gelgen, J., Sch \

    Karimi, A.-H., Von K \"u gelgen, J., Sch \"o lkopf, B., and Valera, I. Algorithmic recourse under imperfect causal knowledge: a probabilistic approach. Advances in neural information processing systems, 33: 0 265--277, 2020

    2020

  25. [25]

    Algorithmic recourse: from counterfactual explanations to interventions

    Karimi, A.-H., Sch \"o lkopf, B., and Valera, I. Algorithmic recourse: from counterfactual explanations to interventions. In Proceedings of the 2021 ACM conference on fairness, accountability, and transparency, pp.\ 353--362, 2021

    2021

  26. [26]

    Kharoua, R. E. Students performance dataset. https://www.kaggle.com/datasets/rabieelkharoua/students-performance-dataset, 2022. Kaggle repository, accessed 2026-01-27

    2022

  27. [27]

    Human decisions and machine predictions

    Kleinberg, J., Lakkaraju, H., Leskovec, J., Ludwig, J., and Mullainathan, S. Human decisions and machine predictions. The quarterly journal of economics, 133 0 (1): 0 237--293, 2018

    2018

  28. [28]

    Early stopping tabular in-context learning

    K \"u ken, J., Purucker, L., and Hutter, F. Early stopping tabular in-context learning. arXiv preprint arXiv:2506.21387, 2025

    work page arXiv 2025

  29. [29]

    E., Papailiopoulos, D., and Oymak, S

    Li, Y., Ildiz, M. E., Papailiopoulos, D., and Oymak, S. Transformers as algorithms: Generalization and stability in in-context learning. In International conference on machine learning, pp.\ 19565--19594. PMLR, 2023

    2023

  30. [30]

    Zoopt: Toolbox for derivative-free optimization

    Liu, Y.-R., Hu, Y.-Q., Qian, H., Qian, C., and Yu, Y. Zoopt: Toolbox for derivative-free optimization. arXiv preprint arXiv:1801.00329, 2017

    work page arXiv 2017

  31. [31]

    K., Sharma, A., and Tan, C

    Mothilal, R. K., Sharma, A., and Tan, C. Explaining machine learning classifiers through diverse counterfactual explanations. In Proceedings of the 2020 conference on fairness, accountability, and transparency, pp.\ 607--617, 2020

    2020

  32. [32]

    Learning model-agnostic counterfactual explanations for tabular data

    Pawelczyk, M., Broelemann, K., and Kasneci, G. Learning model-agnostic counterfactual explanations for tabular data. In Proceedings of the web conference 2020, pp.\ 3126--3132, 2020

    2020

  33. [33]

    Face: feasible and actionable counterfactual explanations

    Poyiadzi, R., Sokol, K., Santos-Rodriguez, R., De Bie, T., and Flach, P. Face: feasible and actionable counterfactual explanations. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, pp.\ 344--350, 2020

    2020

  34. [34]

    Qu, J., Holzm \"u ller, D., Varoquaux, G., and Morvan, M. L. Tabicl: A tabular foundation model for in-context learning on large data. arXiv preprint arXiv:2502.05564, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  35. [35]

    Statlog (Australian Credit Approval)

    Quinlan, R. Statlog (Australian Credit Approval) . UCI Machine Learning Repository, 1987. DOI : https://doi.org/10.24432/C59012

    work page doi:10.24432/c59012 1987

  36. [36]

    Learning models for actionable recourse

    Ross, A., Lakkaraju, H., and Bastani, O. Learning models for actionable recourse. Advances in Neural Information Processing Systems, 34: 0 18734--18746, 2021

    2021

  37. [37]

    Retrieval & fine-tuning for in-context tabular models

    Thomas, V., Ma, J., Hosseinzadeh, R., Golestan, K., Yu, G., Volkovs, M., and Caterini, A. Retrieval & fine-tuning for in-context tabular models. Advances in Neural Information Processing Systems, 37: 0 108439--108467, 2024

    2024

  38. [38]

    Tropp, J. A. et al. An introduction to matrix concentration inequalities. Foundations and Trends in Machine Learning , 8 0 (1-2): 0 1--230, 2015

    2015

  39. [39]

    Ellice: Efficient and provably robust algorithmic recourse via the rashomon sets

    Turbal, B., Voitsitska, I., and Semenova, L. Ellice: Efficient and provably robust algorithmic recourse via the rashomon sets. In The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025

    2025

  40. [40]

    Towards robust and reliable algorithmic recourse

    Upadhyay, S., Joshi, S., and Lakkaraju, H. Towards robust and reliable algorithmic recourse. Advances in Neural Information Processing Systems, 34: 0 16926--16937, 2021

    2021

  41. [41]

    Actionable recourse in linear classification

    Ustun, B., Spangher, A., and Liu, Y. Actionable recourse in linear classification. In Proceedings of the conference on fairness, accountability, and transparency, pp.\ 10--19, 2019

    2019

  42. [42]

    and Von dem Bussche, A

    Voigt, P. and Von dem Bussche, A. The eu general data protection regulation (gdpr). A practical guide, 1st ed., Cham: Springer International Publishing, 10 0 (3152676): 0 10--5555, 2017

    2017

  43. [43]

    Transformers learn in-context by gradient descent

    Von Oswald, J., Niklasson, E., Randazzo, E., Sacramento, J., Mordvintsev, A., Zhmoginov, A., and Vladymyrov, M. Transformers learn in-context by gradient descent. In International Conference on Machine Learning, pp.\ 35151--35174. PMLR, 2023

    2023

  44. [44]

    Counterfactual explanations without opening the black box: Automated decisions and the gdpr

    Wachter, S., Mittelstadt, B., and Russell, C. Counterfactual explanations without opening the black box: Automated decisions and the gdpr. Harv. JL & Tech., 31: 0 841, 2017

    2017

  45. [45]

    E., and Wang, D

    Wang, L., Ren, J., Xu, H., Wang, J., Xie, H., Keyes, D. E., and Wang, D. Zo2: Scalable zeroth-order fine-tuning for extremely large language models with limited gpu memory. arXiv preprint arXiv:2503.12668, 2025 a

    work page arXiv 2025

  46. [46]

    Distzo2: High-throughput and memory-efficient zeroth-order fine-tuning llms with distributed parallel computing

    Wang, L., Xie, H., and Wang, D. Distzo2: High-throughput and memory-efficient zeroth-order fine-tuning llms with distributed parallel computing. arXiv preprint arXiv:2507.03211, 2025 b

    work page arXiv 2025

  47. [47]

    Scalable in-context learning on tabular data via retrieval-augmented large language models

    Wen, X., Zheng, S., Xu, Z., Sun, Y., and Bian, J. Scalable in-context learning on tabular data via retrieval-augmented large language models. arXiv preprint arXiv:2502.03147, 2025

    work page arXiv 2025

  48. [48]

    Qwen3 Technical Report

    Yang, A., Li, A., Yang, B., Zhang, B., Hui, B., Zheng, B., Yu, B., Gao, C., Huang, C., Lv, C., et al. Qwen3 technical report. arXiv preprint arXiv:2505.09388, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  49. [49]

    Derivative-free optimization via classification

    Yu, Y., Qian, H., and Hu, Y.-Q. Derivative-free optimization via classification. In Proceedings of the AAAI conference on artificial intelligence, volume 30, 2016

    2016

  50. [50]

    Zhang, R., Frei, S., and Bartlett, P. L. Trained transformers learn linear models in-context. Journal of Machine Learning Research, 25 0 (49): 0 1--55, 2024

    2024

  51. [51]

    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 format.date year duplicate empty "emp...