RobustFT: Robust Supervised Fine-tuning for Large Language Models under Noisy Response

Jingyang Yuan; Junyu Luo; Kaize Ding; Ming Zhang; Xiao Luo; Zhiping Xiao

REVIEW 5 major objections 5 minor 3 cited by

RobustFT: Robust Supervised Fine-tuning for Large Language Models under Noisy Response

T0 review · 5 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read RobustFT claims that a multi-model consistency check plus context-based relabeling can detect and fix noisy responses in supervised fine-tuning data, keeping LLM downstream performance high even at 70% noise.

desk verdict A plausible and broadly effective noisy-SFT pipeline, but the load-bearing noise-detection step is never evaluated against the known noise mask, and the numbers need cleaning up. read the letter →

arxiv 2412.14922 v1 pith:ECHN6S6W submitted 2024-12-19 cs.CL cs.AI

Junyu Luo , Xiao Luo , Kaize Ding , Jingyang Yuan , Zhiping Xiao , Ming Zhang This is my paper · ORCID

classification cs.CLcs.AI

keywords noisylabelssupervisedfine-tuninglargelanguagemodelsnoisedetectiondatarelabelingmulti-expertconsistencyentropy-basedselectioncontext-enhancedretrieval

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

Supervised fine-tuning (SFT) of large language models is only as good as its training responses, and collected data inevitably contains wrong answers. This paper proposes RobustFT, a framework that tries to make SFT noise-robust by detecting and relabeling noisy responses before training. It claims that a consensus check among the base LLM, a reasoning-refined LLM, and the original label separates clean from noisy samples, and that relabeling with retrieved clean context plus review restores most of the lost accuracy. Across five benchmarks and three open LLMs, RobustFT reports higher downstream accuracy than vanilla inference or standard SFT at 30%, 50%, and 70% synthetic noise. If the claim holds, practitioners can fine-tune on imperfect real-world collections without sacrificing reliability.

What carries the argument

The load-bearing object is the Checker consistency rule of Eq. (3), which maps the triple of original label $y_i$, base prediction $\hat{y}_i$, and reasoning-enhanced prediction $\hat{y}^{\mathrm{reas}}_i$ to a binary reliability flag $r_i \in \{0,1\}$; it is what partitions the dataset into $D_{\mathrm{clean}}$ and $D_{\mathrm{noise}}$. A second mechanism is the entropy-based selection of Eq. (8), which computes the per-token negative log-likelihood of context-enhanced responses and keeps only the top-$\beta$ most confident repaired samples. Supporting machinery includes the reasoning-enhanced LLM that alternates reasoning and reflection, and the Review Agent that synthesizes context-enhanced and reasoning-enhanced candidate responses into a repaired label. Together these components define the pipeline that converts a noisy task dataset into the filtered fine-tuning set $D_{\mathrm{ft}}$.

What would settle it

Inject noise with known ground truth into the five datasets, run only the Checker, and compare its clean/noisy flags with the true injected noise bit; if the Checker misses most noisy samples or flags many clean ones, the reported fine-tuning gains cannot be attributed to accurate noise detection.

Watch

Extended reading notes

Core claim

The paper's central claim is that noisy response data, the kind that inevitably accumulates in real-world SFT collections, can be corrected through a fully self-contained detection-and-denoising loop, without external clean data. The detector is a consistency check: a sample is labeled clean only if the original answer, the base LLM's direct prediction, and the prediction from an iterative reasoning-reflection LLM all agree; any disagreement routes the sample to a denoising stage. Denoising first retrieves the most similar clean samples as context, generates a context-enhanced candidate, and then has a Review Agent synthesize a repaired label from that candidate and the reasoning-enhanced prediction. Finally, response entropy ranks the repaired samples and only the most confident fraction is kept. The reported result is that this pipeline outperforms vanilla models, standard SFT, and several noisy-label baselines across MMLU, ARC, PubMedQA, Drop, and FPB under 30–70% injected noise on Llama-3.2-3B, Llama-3.1-8B, and Gemma2-9B.

Load-bearing premise

The framework assumes that agreement among the original label, the base model's direct answer, and its reasoning-refined answer reliably separates clean from noisy samples, but the paper never reports how often that consensus judgment is actually correct.

Editorial extensions

If this is right

Fine-tuning can tolerate noisy collection pipelines: at 70% injected noise, RobustFT-trained Llama-3.1-8B reaches 67.6 on MMLU and 84.1 on ARC, above the vanilla model's 65.3 and 82.7.
No external clean dataset is needed; the framework is self-contained, so it applies wherever only noisy downstream data is available.
Smaller models gain the most from denoising: Llama-3.2-3B moves from 38.3 to 57.9 on MMLU at 70% noise, closing much of the gap to larger models.
Domain-specific fine-tuning becomes viable on noisy data, including finance (FPB) and biomedical QA (PubMedQA), where base models often lack strong priors.
The ablations imply all three stages—consistency detection, context-enhanced relabeling, and entropy selection—are needed for the full gain; removing any one lowers accuracy.

Reading between the lines

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

(Editorial) The Checker's agreement rule is never validated against ground-truth noise, so a natural test is to inject known noise, run only the detection stage, and measure how often the clean/noisy split matches the injected labels; that would separate detection quality from relabeling quality.
(Editorial) The entropy gate keeps only the most confident half of repaired samples, indicating a deliberate trade-off of data quantity for confidence; the same gate could be reused as a generic quality filter for self-training or preference-tuning datasets.
(Editorial) Because the final model is trained partly on the base model's own predictions, some of the gain over vanilla could come from self-training regularization rather than from correcting the specific noisy answers; a variant that relabels all samples without the Checker partition would expose how much of the gain is detection-driven.
(Editorial) The cross-model results suggest data repair matters more than scale for noisy SFT, a hypothesis that could be stress-tested by running RobustFT on a broader family of base models and on natural, rather than synthetically injected, noise.

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.

Desk Editor's Note

read the letter

Colleague,

Here's the takeaway: RobustFT is a practical, well-tested pipeline for fine-tuning LLMs on noisy response data, and the gains it reports are consistent across three models and five datasets. But the central claim—that it detects and corrects noisy labels—rests on a noise-detection step that is never directly evaluated, and there's an inconsistency in the reported numbers that needs resolving.

What's actually new: the combination of multi-expert consensus (base prediction, reasoning-enhanced prediction, original label) to flag noisy samples, followed by retrieval-augmented relabeling and entropy-based selection, applied to SFT. The components are individually known, but the integrated framework is new. The ablations show each piece contributes, and the improvements over vanilla SFT at high noise rates are large (e.g., 81% relative on MMLU at 70% noise). That's worth taking seriously.

What's soft: First, the Checker in Eq. 3 partitions the data into clean and noisy, and everything downstream depends on that partition. Noise is synthetically injected, so the ground-truth mask is known. The paper never reports precision, recall, or a confusion matrix for this binary split. That's a load-bearing omission. If the Checker is biased—say, it marks noisy labels that agree with the model as clean, or sends correct labels that disagree with the model for relabeling—the end-task gains might come from filtering or prior reinforcement, not from actual noise correction. A direct evaluation against the injected mask would settle it.

Second, Table 1 gives FPB at 70% noise as 76.2 for Llama-3.1-8B, but Table 2 gives 73.2 for the same model and setting. One of these is wrong, and that makes me want to verify the other numbers. Third, there are no error bars or significance tests on the main tables. The stability analysis in Fig. 6 is about instruction rephrasing, not the core results. Fourth, Eq. 7 is labeled entropy but is actually per-token negative log-likelihood (log perplexity). Minor, but it should be fixed.

The circularity concern is not a real problem here—this is a self-training-style pipeline, and that's standard. The missing Checker evaluation is the real issue.

Who's this for? Anyone fine-tuning LLMs on imperfect data, and researchers working on data cleaning for SFT. It's a useful paper with a clear contribution. It deserves a serious referee, but the revision should be conditional on evaluating the Checker against the known noise mask, fixing the table inconsistency, and adding some measure of variance.

My recommendation: send it to review, but insist on those additions.

Referee Report

5 major / 5 minor

Summary. The paper proposes RobustFT, a two-stage framework for robust supervised fine-tuning of LLMs under noisy response data. In the noise-detection stage, the base LLM generates a prediction for each training sample, a reasoning-enhanced LLM generates a second prediction via iterative reasoning and reflection, and a 'Checker' partitions the data into clean and noisy sets based on agreement among the original label and the two model predictions. In the denoising stage, noisy samples are relabeled using retrieved context from the clean set plus a Review Agent, and an entropy-based selection step retains the most confident relabeled samples. The final fine-tuning set is the union of the clean set and the selected relabeled set. The method is evaluated on five benchmarks (MMLU, ARC, PubMedQA, DROP, FPB) at 30%, 50%, and 70% injected noise, using three base LLMs, with ablations and sensitivity analyses.

Significance. The problem addressed is practically important: noisy training responses are common in real-world data collection, and the paper demonstrates large end-task degradations from noise and consistent improvements of RobustFT over vanilla SFT across datasets, noise levels, and model families. Strengths include the breadth of the experimental study (three models, five datasets, three noise rates), the ablation study showing that each framework component contributes, the public release of code and data, and the clear framing of a self-contained pipeline without external models. If the detection-and-denosing mechanism is actually identifying and correcting the injected noise, this is a useful contribution to noise-robust LLM adaptation. The main caveat, discussed below, is that the central Checker component is both under-specified and never directly evaluated against the known noise mask, leaving the claimed mechanism unverified.

major comments (5)

[§3.2, Eq. (3)] The Checker is never concretely defined. Equation (3) states ri = Checker(yi, ŷi, ŷ_i^reas) ∈ {0,1}, but the text gives no operational definition of the consistency metric, no description of the prompt or rule used to combine the three inputs, and no formal meaning of the threshold θ that appears in §4.1.3. Without this specification the method is not reproducible, and the claim that this 'Checker mechanism' performs noise detection cannot be tested.
[§3.3, Eq. (4)] The Encoder used for query retrieval is not specified. Equation (4) introduces hi = Encoder(qi) ∈ R^d, but the paper does not state what model or embedding is used, how d is chosen, or whether the encoder is trained or frozen. Since the quality of the retrieved context in Eq. (5) directly depends on the encoder's semantic similarity, this missing detail is load-bearing for the denoising mechanism.
[§4, 'Main Result' and §3.2] The central noise-detection mechanism is never evaluated against the known synthetic noise mask. The experiments inject noise into training responses, so the ground-truth clean/noisy status of every sample is known; yet the paper reports no precision, recall, or confusion matrix for the Checker's partition into D_clean and D_noise. This omission is load-bearing because if the Checker has low precision, clean samples are unnecessarily relabeled or discarded, and if it has low recall, noisy labels pass through to fine-tuning. End-task improvements could therefore arise from aggressive filtering or from reinforcing the base model's prior rather than from accurate noise correction.
[Table 1 vs. Table 2] The reported results for Llama3.1-8B on FPB at 70% noise are inconsistent: Table 1 lists the RobustFT result as 76.2, while Table 2 lists the same condition as 73.2. Since these tables are both used to support the central claim of consistent improvement, the discrepancy must be reconciled (or explained, e.g., by different evaluation splits or seeds).
[§4.2, Tables 1 and 2] The abstract and conclusion claim 'significant improvements,' but the main tables report only point estimates with no error bars, significance tests, or multiple-seed training runs. The stability analysis in §4.3.5 and Figure 6 covers only MMLU and ARC and measures inference variability under rephrased instructions, not variance of the fine-tuning procedure. Please provide variance estimates for at least the main comparisons, or soften the statistical claim.

minor comments (5)

[§4.1.3] The sentence 'We set then = 4and θ = 50%' contains a typo; it should read 'We set n = 4 and θ = 50%.'
[§4.1.1] The DROP dataset is used in the experiments but no reference is provided for it; please add a citation for DROP.
[§3.4, Eq. (7)] The quantity in Eq. (7) is the average negative log-likelihood per token (which becomes perplexity when exponentiated), not the Shannon entropy of the token distribution; please rename or clarify the terminology to avoid confusion.
[§4.3.1] The ablation variants (w/oSelection, w/oChecker, w/oReviewer, w/oCER, w/oREL) are not fully defined in the main text; please specify what each variant removes and how the removed component is replaced.
[§4.1.2] The baselines SelfLabel, SelfSelect, and SelfRAG are only mentioned by name in the main text with implementation details deferred to the appendix; please add one-sentence descriptions in the main text so that a reader can understand the comparisons without consulting the appendix.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: RobustFT's reported gains are empirical outcomes of a detection-denoising pipeline, not identities forced by its equations.

full rationale

RobustFT's chain is empirical rather than definitional. The Checker (Eq. 3) partitions data by agreement among yi, the base prediction, and the reasoning-enhanced prediction; clean samples are kept in Dclean, noisy samples are relabeled via context-enhanced generation (Eq. 5) and a Review Agent (Eq. 6), then filtered by entropy (Eqs. 7-8), and the model is fine-tuned on Dft = Dclean ∪ Dselect (Eq. 9). None of these equations define the reported downstream accuracies: the MMLU/ARC/PubMedQA/Drop/FPB numbers are measured after fine-tuning, so the central claim is not equivalent to its inputs by construction. The only self-citation with author overlap (SemiEvol, Luo et al. 2024) appears in Related Work as context and is not load-bearing. The failure to report precision/recall of the Checker against the synthetic noise mask is a real empirical validation gap, but it is not circularity: the ground-truth mask exists and the split could be evaluated. Similarly, tuning β and k on MMLU (Sec. 4.3.2) is a test-set leakage concern, not a definitional reduction, since the final accuracy is not an algebraic function of those hyperparameters. No uniqueness theorem, ansatz-smuggling citation, or renaming of a known result was found.

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

The framework rests on three domain assumptions about LLM behavior: consensus indicates correctness, confidence indicates quality, and retrieval context improves relabeling. These are heuristics, not derived guarantees. Free parameters beta, k, n, and theta are set by hand or validated on a single dataset.

free parameters (4)

beta (selection ratio) = 0.5 (default)
Controls the fraction of denoised samples kept. Set to 50% based on sensitivity analysis on MMLU (Section 4.3.2) showing peak at 40-50%.
k (context length) = 3 (default)
Number of similar clean samples retrieved for relabeling. Default 3, validated on MMLU where performance plateaus for k=3-5.
n (reasoning-reflection iterations) = 4 (stated as 'then = 4')
Number of reasoning-reflection iterations for the reasoning-enhanced LLM; set without reported sensitivity analysis.
theta (checker threshold) = 0.5 (stated as 'theta = 50%')
Mentioned in implementation details but not defined in the main text; likely a consistency threshold for the Checker.

assumptions (3)

domain assumption Prediction agreement among the original label, base LLM, and reasoning-enhanced LLM indicates label correctness.
The Checker in Eq. 3 partitions data into clean and noisy solely on this consistency heuristic; no ground-truth validation of detected noise is reported.
domain assumption Lower average negative log-likelihood of a context-enhanced response corresponds to higher answer quality.
The data selection step (Eq. 7-8) assumes confidence measured this way correlates with correctness and retains only the top-beta samples.
domain assumption Retrieving similar clean samples and feeding them as context produces more reliable relabeled answers than unaided generation.
The context-enhanced relabeling in Eq. 5 relies on this; the quality of the retrieval encoder is not evaluated.

how reviews work

0 comments

Cite this review

Pith. "Pith review of RobustFT: Robust Supervised Fine-tuning for Large Language Models under Noisy Response." pith.science (2026). https://pith.science/paper/ECHN6S6W

@misc{pith2026241214922,
  author       = {Pith},
  title        = {Pith review of: RobustFT: Robust Supervised Fine-tuning for Large Language Models under Noisy Response},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ECHN6S6W}},
  note         = {Machine review of arXiv:2412.14922}
}

read the original abstract

Supervised fine-tuning (SFT) plays a crucial role in adapting large language models (LLMs) to specific domains or tasks. However, as demonstrated by empirical experiments, the collected data inevitably contains noise in practical applications, which poses significant challenges to model performance on downstream tasks. Therefore, there is an urgent need for a noise-robust SFT framework to enhance model capabilities in downstream tasks. To address this challenge, we introduce a robust SFT framework (RobustFT) that performs noise detection and relabeling on downstream task data. For noise identification, our approach employs a multi-expert collaborative system with inference-enhanced models to achieve superior noise detection. In the denoising phase, we utilize a context-enhanced strategy, which incorporates the most relevant and confident knowledge followed by careful assessment to generate reliable annotations. Additionally, we introduce an effective data selection mechanism based on response entropy, ensuring only high-quality samples are retained for fine-tuning. Extensive experiments conducted on multiple LLMs across five datasets demonstrate RobustFT's exceptional performance in noisy scenarios.

Figures

Figures reproduced from arXiv: 2412.14922 by the authors.

**Figure 1.** Impact of noisy data on LLM performance during SFT. Increasing noise levels deteriorates model performance, highlighting the critical need for noiserobust fine-tuning approaches. 2024), scenario-specific collection (Clark et al., 2019), and model-based self-labeling (Wang et al., 2024), these data sources inherently contain noise stemming from both human annotation errors and model hallucinations (Farquhar et al., … view at source ↗

**Figure 2.** Overview of ROBUSTFT. Our ROBUSTFT enhances model performance through a two-stage noise detection-and-denoising framework, leveraging collaborative learning among expert LLMs for noise detection and context-enhanced reasoning for data denoising, ultimately enabling robust downstream fine-tuning. 3 Methodology 3.1 Overview Adapting and fine-tuning Large Language Models (LLMs) in real-world scenarios presents signific… view at source ↗

**Figure 3.** Sensitivity analysis on MMLU under different β and k with varying noise levels. to domain experts. While Gemma2-9B showed strong general capabilities, it initially performed worse on domain-specific tasks. However, after ROBUSTFT, it effectively adapted to these domains and outperformed Llama3.1-8B, demonstrating the importance of denoising in LLM adaptation. Critical importance of denoising for smaller models. Sm… view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Perplexity analysis of ROBUSTFT on MMLU and ARC with varying noise levels. biology business chemistry economics computer_science engineering health history law math other philosophy physics psychology MMLU Accuracy (Llama-3.1 8B) SFT 30% Ours 30% SFT 50% Ours 50% SFT 7…

**Figure 6.** Figure 6: Stability analysis on MMLU and ARC. range due to the inclusion of excessive noisy samples. For context length, performance improves with increasing k but plateaus, particularly in the range of k = 3−5, suggesting that moderate k provide sufficient reasoning support. …

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

Positive-Unlabeled Preference Optimization For Chest X-ray Report Generation
cs.CV 2026-08 conditional novelty 6.0 of 10

PU-DPO applies positive-unlabeled learning to preference optimization so that report generators learn to mention findings that are present but missing from noisy training reports.
We Should Identify and Mitigate Third-Party Safety Risks in MCP-Powered Agent Systems
cs.LG 2025-06 conditional novelty 5.0 of 10

MCP-powered LLM agents are vulnerable to prompt injection from third-party services, and simple detection or filtering defenses do not reliably stop these attacks.
A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment
cs.CR 2025-04 conditional novelty 5.0 of 10

A large collaborative survey organizes LLM and LLM-agent safety issues into a full-stack lifecycle framework from data preparation to deployment.

Reference graph

Works this paper leans on

51 extracted references · 13 canonical work pages · cited by 3 Pith papers

[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]

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]

Gantavya Bhatt, Yifang Chen, Arnav M Das, Jifan Zhang, Sang T Truong, Stephen Mussmann, Yinglun Zhu, Jeffrey Bilmes, Simon S Du, Kevin Jamieson, et al. 2024. An experimental design framework for label-efficient supervised finetuning of large language models. arXiv preprint arXiv:2401.06692

arXiv 2024
[4]

Federico Bianchi, Mirac Suzgun, Giuseppe Attanasio, Paul R \"o ttger, Dan Jurafsky, Tatsunori Hashimoto, and James Zou. 2023. Safety-tuned llamas: Lessons from improving the safety of large language models that follow instructions. arXiv preprint arXiv:2309.07875

arXiv 2023
[5]

Alexander Bukharin and Tuo Zhao. 2023. Data diversity matters for robust instruction tuning. arXiv preprint arXiv:2311.14736

arXiv 2023
[6]

Hyeong Kyu Choi, Xuefeng Du, and Yixuan Li. 2024. Safety-aware fine-tuning of large language models. arXiv preprint arXiv:2410.10014

arXiv 2024
[7]

Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Yunxuan Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, et al. 2024. Scaling instruction-finetuned language models. Journal of Machine Learning Research, 25(70):1--53

2024
[8]

Christopher Clark, Kenton Lee, Ming-Wei Chang, Tom Kwiatkowski, Michael Collins, and Kristina Toutanova. 2019. Boolq: Exploring the surprising difficulty of natural yes/no questions. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Pa...

2019

Show all 51 references

[9]

Peter Clark, Isaac Cowhey, Oren Etzioni, Tushar Khot, Ashish Sabharwal, Carissa Schoenick, and Oyvind Tafjord. 2018. Think you have solved question answering? try arc, the ai2 reasoning challenge. arXiv preprint arXiv:1803.05457

2018 arXiv
[10]

Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783

2024 arXiv
[11]

Sebastian Farquhar, Jannik Kossen, Lorenz Kuhn, and Yarin Gal. 2024. Detecting hallucinations in large language models using semantic entropy. Nature, 630(8017):625--630

2024
[12]

Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2020. Measuring massive multitask language understanding. arXiv preprint arXiv:2009.03300

2020 arXiv
[13]

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

2021 arXiv
[14]

Tiansheng Huang, Sihao Hu, Fatih Ilhan, Selim Furkan Tekin, and Ling Liu. 2024. Harmful fine-tuning attacks and defenses for large language models: A survey. arXiv preprint arXiv:2409.18169

2024 arXiv
[15]

Daniel P Jeong, Zachary C Lipton, and Pradeep Ravikumar. 2024. Llm-select: Feature selection with large language models. arXiv preprint arXiv:2407.02694

2024 arXiv
[16]

Qiao Jin, Bhuwan Dhingra, Zhengping Liu, William Cohen, and Xinghua Lu. 2019. Pubmedqa: A dataset for biomedical research question answering. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on...

2019
[17]

Yeachan Kim, Junho Kim, and SangKeun Lee. 2024. Towards robust and generalized parameter-efficient fine-tuning for noisy label learning. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

2024
[18]

Nathan Lambert, Jacob Morrison, Valentina Pyatkin, Shengyi Huang, Hamish Ivison, et al. 2024. T " ulu 3: Pushing frontiers in open language model post-training. arXiv preprint arXiv:2411.15124

2024 arXiv
[19]

u ttler, Mike Lewis, Wen-tau Yih, Tim Rockt \

Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich K \"u ttler, Mike Lewis, Wen-tau Yih, Tim Rockt \"a schel, et al. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in Neural Information Proc...

2020
[20]

Jiaxiang Li, Siliang Zeng, Hoi-To Wai, Chenliang Li, Alfredo Garcia, and Mingyi Hong. 2024. Getting more juice out of the sft data: Reward learning from human demonstration improves sft for llm alignment. arXiv preprint arXiv:2405.17888

2024 arXiv
[21]

Xize Liang, Chao Chen, Jie Wang, Yue Wu, Zhihang Fu, Zhihao Shi, Feng Wu, and Jieping Ye. 2024. Robust preference optimization with provable noise tolerance for llms. arXiv preprint arXiv:2404.04102

2024 arXiv
[22]

Shayne Longpre, Le Hou, Tu Vu, Albert Webson, Hyung Won Chung, Yi Tay, Denny Zhou, Quoc V Le, Barret Zoph, Jason Wei, et al. 2023. The flan collection: Designing data and methods for effective instruction tuning. In International Conference on Machine Learning, pages 22631--22...

2023
[23]

Junyu Luo, Xiao Luo, Xiusi Chen, Zhiping Xiao, Wei Ju, and Ming Zhang. 2024. Semievol: Semi-supervised fine-tuning for llm adaptation. arXiv preprint arXiv:2410.14745

2024 arXiv
[24]

Kaifeng Lyu, Haoyu Zhao, Xinran Gu, Dingli Yu, Anirudh Goyal, and Sanjeev Arora. 2024. Keeping llms aligned after fine-tuning: The crucial role of prompt templates. arXiv preprint arXiv:2402.18540

2024 arXiv
[25]

Pekka Malo, Ankur Sinha, Pekka Korhonen, Jyrki Wallenius, and Pyry Takala. 2014. Good debt or bad debt: Detecting semantic orientations in economic texts. Journal of the Association for Information Science and Technology, 65(4):782--796

2014
[26]

Elena Merdjanovska, Ansar Aynetdinov, and Alan Akbik. 2024. N oise B ench: Benchmarking the impact of real label noise on named entity recognition. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing

2024
[27]

Shervin Minaee, Tomas Mikolov, Narjes Nikzad, Meysam Chenaghlu, Richard Socher, Xavier Amatriain, and Jianfeng Gao. 2024. Large language models: A survey. arXiv preprint arXiv:2402.06196

2024 arXiv
[28]

Subhabrata Mukherjee, Arindam Mitra, Ganesh Jawahar, Sahaj Agarwal, Hamid Palangi, and Ahmed Awadallah. 2023. Orca: Progressive learning from complex explanation traces of gpt-4. arXiv preprint arXiv:2306.02707

2023 arXiv
[29]

Jishnu Mukhoti, Yarin Gal, Philip HS Torr, and Puneet K Dokania. 2023. Fine-tuning can cripple your foundation model; preserving features may be the solution. arXiv preprint arXiv:2308.13320

2023 arXiv
[30]

Curtis Northcutt, Lu Jiang, and Isaac Chuang. 2021. Confident learning: Estimating uncertainty in dataset labels. Journal of Artificial Intelligence Research, 70:1373--1411

2021
[31]

Ritik Sachin Parkar, Jaehyung Kim, Jong Inn Park, and Dongyeop Kang. 2024. Selectllm: Can llms select important instructions to annotate? arXiv preprint arXiv:2401.16553

2024 arXiv
[32]

Zhenting Qi, Xiaoyu Tan, Chao Qu, Yinghui Xu, and Yuan Qi. 2023. S a FER : A robust and efficient framework for fine-tuning BERT -based classifier with noisy labels. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 5: Industry Track)

2023
[33]

Dan Qiao, Chenchen Dai, Yuyang Ding, Juntao Li, Qiang Chen, Wenliang Chen, and Min Zhang. 2022. Selfmix: Robust learning against textual label noise with self-mixup training. arXiv preprint arXiv:2210.04525

2022 arXiv
[34]

Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. 2020. Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of machine learning research, 21(140):1--67

2020
[35]

Kihyuk Sohn, David Berthelot, Nicholas Carlini, Zizhao Zhang, Han Zhang, Colin A Raffel, Ekin Dogus Cubuk, Alexey Kurakin, and Chun-Liang Li. 2020. Fixmatch: Simplifying semi-supervised learning with consistency and confidence. Advances in neural information processing systems...

2020
[36]

Qi Sun, Kun Huang, Xiaocui Yang, Pengfei Hong, Kun Zhang, and Soujanya Poria. 2023. Uncertainty guided label denoising for document-level distant relation extraction. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

2023
[37]

Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B Hashimoto. 2023. Stanford alpaca: An instruction-following llama model

2023
[38]

Gemma Team, Morgane Riviere, Shreya Pathak, Pier Giuseppe Sessa, Cassidy Hardin, Surya Bhupatiraju, L \'e onard Hussenot, Thomas Mesnard, Bobak Shahriari, Alexandre Ram \'e , et al. 2024. Gemma 2: Improving open language models at a practical size, 2024. https://arxiv. org/abs...

2024 arXiv
[39]

Ryan Teknium, Jeffrey Quesnelle, and Chen Guang. 2024. https://arxiv.org/abs/2408.11857 Hermes 3 technical report . Preprint, arXiv:2408.11857

2024 arXiv
[40]

Tao Tu, Anil Palepu, Mike Schaekermann, Khaled Saab, Jan Freyberg, Ryutaro Tanno, Amy Wang, Brenna Li, Mohamed Amin, Nenad Tomasev, et al. 2024. Towards conversational diagnostic ai. arXiv preprint arXiv:2401.05654

2024 arXiv
[41]

Song Wang, Zhen Tan, Ruocheng Guo, and Jundong Li. 2023 a . Noise-robust fine-tuning of pretrained language models via external guidance. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 12528--12540

2023
[42]

Tianlu Wang, Ilia Kulikov, Olga Golovneva, Ping Yu, Weizhe Yuan, Jane Dwivedi-Yu, Richard Yuanzhe Pang, Maryam Fazel-Zarandi, Jason Weston, and Xian Li. 2024. Self-taught evaluators. arXiv preprint arXiv:2408.02666

2024 arXiv
[43]

Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A Smith, Daniel Khashabi, and Hannaneh Hajishirzi. 2023 b . Self-instruct: Aligning language models with self-generated instructions. In Proceedings of the 61st Annual Meeting of the Association for Computational Lin...

2023
[44]

Boyi Wei, Kaixuan Huang, Yangsibo Huang, Tinghao Xie, Xiangyu Qi, Mengzhou Xia, Prateek Mittal, Mengdi Wang, and Peter Henderson. 2024. Assessing the brittleness of safety alignment via pruning and low-rank modifications. arXiv preprint arXiv:2402.05162

2024 arXiv
[45]

Mengzhou Xia, Sadhika Malladi, Suchin Gururangan, Sanjeev Arora, and Danqi Chen. 2024. Less: Selecting influential data for targeted instruction tuning. arXiv preprint arXiv:2402.04333

2024 arXiv
[46]

Pengyu Xu, Liping Jing, and Jian Yu. 2024. Enhancing multi-label text classification under label-dependent noise: A label-specific denoising framework. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 5674--5688

2024
[47]

Bo Yuan, Yulin Chen, Yin Zhang, and Wei Jiang. 2024. Hide and seek in noise labels: Noise-robust collaborative active learning with llms-powered assistance. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), page...

2024
[48]

Yivan Zhang, Gang Niu, and Masashi Sugiyama. 2021. Learning noise transition matrix from only noisy labels via total variation regularization. In International Conference on Machine Learning, pages 12501--12512. PMLR

2021
[49]

Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, et al. 2023. A survey of large language models. arXiv preprint arXiv:2303.18223

2023 arXiv
[50]

Yaowei Zheng, Richong Zhang, Junhao Zhang, Yanhan Ye, Zheyan Luo, Zhangchi Feng, and Yongqiang Ma. 2024. http://arxiv.org/abs/2403.13372 Llamafactory: Unified efficient fine-tuning of 100+ language models . In Proceedings of the 62nd Annual Meeting of the Association for Compu...

2024 arXiv
[51]

Yuchen Zhuang, Yue Yu, Lingkai Kong, Xiang Chen, and Chao Zhang. 2023. Dygen: Learning from noisy labels via dynamics-enhanced generative modeling. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 3674--3686

2023

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[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]

Gantavya Bhatt, Yifang Chen, Arnav M Das, Jifan Zhang, Sang T Truong, Stephen Mussmann, Yinglun Zhu, Jeffrey Bilmes, Simon S Du, Kevin Jamieson, et al. 2024. An experimental design framework for label-efficient supervised finetuning of large language models. arXiv preprint arXiv:2401.06692

arXiv 2024

[4] [4]

Federico Bianchi, Mirac Suzgun, Giuseppe Attanasio, Paul R \"o ttger, Dan Jurafsky, Tatsunori Hashimoto, and James Zou. 2023. Safety-tuned llamas: Lessons from improving the safety of large language models that follow instructions. arXiv preprint arXiv:2309.07875

arXiv 2023

[5] [5]

Alexander Bukharin and Tuo Zhao. 2023. Data diversity matters for robust instruction tuning. arXiv preprint arXiv:2311.14736

arXiv 2023

[6] [6]

Hyeong Kyu Choi, Xuefeng Du, and Yixuan Li. 2024. Safety-aware fine-tuning of large language models. arXiv preprint arXiv:2410.10014

arXiv 2024

[7] [7]

Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Yunxuan Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, et al. 2024. Scaling instruction-finetuned language models. Journal of Machine Learning Research, 25(70):1--53

2024

[8] [8]

Christopher Clark, Kenton Lee, Ming-Wei Chang, Tom Kwiatkowski, Michael Collins, and Kristina Toutanova. 2019. Boolq: Exploring the surprising difficulty of natural yes/no questions. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Pa...

2019