Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

LLMCL-GEC: Advancing Grammatical Error Correction with LLM-Driven Curriculum Learning

T0 review · 4 major / 6 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 Training grammatical error correction models on an LLM-designed easy-to-hard curriculum improves F0.5 over one-shot fine-tuning.

desk verdict A plausible, well-executed first application of LLM-scored curriculum learning to GEC, with a real budget-matchedness confound in the main comparison that needs to be resolved before the central claim is trusted. read the letter →

arxiv 2412.12541 v1 pith:NTST334F submitted 2024-12-17 cs.CL cs.AI

Tao Fang , Derek F. Wong , Lusheng Zhang , Keyan Jin , Qiang Zhang , Tianjiao Li , Jinlong Hou , Lidia S. Chao This is my paper · ORCID
classification cs.CLcs.AI
keywords grammaticalerrorcorrectioncurriculumlearningLLM-baseddifficultyscoringeasy-to-hardtrainingsupervisedfine-tuningcLang8T5LLaMA
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

The paper claims that the order in which a grammatical error correction (GEC) model sees its training data is itself a source of accuracy. It introduces LLM-based curriculum learning: a large language model scores each source sentence's correction difficulty on a 1–10 scale, the data are split into easy (1–3), medium (4–7), and hard (8–10) courses, and the student model is fine-tuned first on easy data, then on easy plus medium data, then on all three. On the CoNLL14 and BEA19 benchmarks this schedule beats both one-shot training on all data and a sentence-length curriculum for T5-xl and LLaMA2-13b, adding roughly one point of $F_{0.5}$ in the best configurations. The paper also reports that training hard-to-easy ends near the one-shot baseline, so the direction of the curriculum, not just the selected data, is what carries the benefit.

What carries the argument

The engine of the method is a difficulty-scoring prompt run by a large LLM on every source sentence in the training set, producing an integer rating from 1 to 10 that partitions the data into easy, medium, and hard courses. The training schedule is the second half of the mechanism: at each stage the model maximizes log-likelihood on the union of all previously seen courses plus the newly introduced one, so easier data are repeatedly rehearsed while harder data are gradually added. This rehearsal design is what distinguishes the staged curriculum from a one-shot mixture of the same examples.

What would settle it

Train the same student model on the same data in hard-to-easy order with the same total gradient updates as the easy-to-hard schedule; if the hard-to-easy run matches or beats the easy-to-hard run in $F_{0.5}$ on CoNLL14 and BEA19, the ordering claim fails.

Watch

Extended reading notes

Core claim

The central discovery is that an easy-to-hard curriculum built from LLM-assigned difficulty scores makes GEC fine-tuning more effective than training on the same data all at once. The scoring stage uses a large model as a human-expert stand-in: it rates how hard it is to correct each source sentence, and fixed thresholds turn those ratings into three courses. The training stage then optimizes the student model sequentially on easy, easy+medium, and easy+medium+hard data, always retaining earlier data so that adding harder examples does not erase what was learned. With this schedule, T5-xl reaches $F_{0.5}$ of 69.3 on CoNLL14 and 75.6 on BEA19-test, up from 68.1 and 74.5 for the one-shot baseline; LLaMA2-13b reaches 69.6 and 74.3, up from 68.4 and 73.3. The same-data ablation, which mixes three passes over easy data with two over medium and one over hard, stays below the staged schedule, and the hard-to-easy version ends near baseline, which is why the authors attribute the gain to easy-to-hard ordering itself.

Load-bearing premise

The method assumes the large model's scores really rank sentences from easier to harder for the smaller model being trained, and that the staged schedule's advantage is not merely extra training on easy sentences.

Editorial extensions

If this is right

  • GEC fine-tuning can be improved without new training data or architecture changes, purely by reordering existing examples according to LLM-scored difficulty.
  • A general-purpose large language model can stand in for human expert curriculum design, since its difficulty ratings agreed with human ratings on the 200-sentence sample the paper inspected.
  • Sentence length is an incomplete difficulty signal: length-based curricula improve over the baseline but lag the LLM-scored curriculum under the same staged schedule.
  • Curriculum direction matters: hard-to-easy training finishes near the one-shot baseline, so future GEC training recipes should start from easier corrections and add harder ones.

Reading between the lines

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

  • If the ordering effect transfers, other correction and rewriting tasks, such as text simplification, style transfer, and sentence-level paraphrase, could adopt the same LLM-scored easy-to-hard schedule without any new task-specific data.
  • The fixed cutoffs at scores 3 and 7 and the 'until convergence' stage lengths are untested choices; sweeping them could either enlarge the reported gains or show that only a brief easy-data warm-up does the work.
  • The expensive 70b-model scoring step could likely be distilled into a small classifier, making the curriculum pipeline practical at scale if the difficulty scores prove reusable across student models.
  • A stricter test of the ordering claim would match total gradient updates between staged and one-shot training exactly, since the paper's stage lengths are convergence-based.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes LLMCL-GEC, a curriculum-learning method for grammatical error correction. A frozen LLaMA2-70b model scores the correction difficulty of each source sentence in the cLang8 training set on a 1-10 scale; the scored dataset is split into easy (1-3), medium (4-7), and hard (8-10) buckets; and T5-large/xl and LLaMA2-7b/13b are fine-tuned sequentially on easy, then easy+medium, then easy+medium+hard. The authors evaluate on CoNLL14 test, BEA19 test/dev using F0.5 and compare against standard fine-tuning, length-based curriculum learning, a 3E+2M+1H rebalanced baseline, and a hard-to-easy sequence. They report consistent F0.5 gains for the easy-to-hard LLM-based curriculum and conclude that ordering examples from easy to hard is what drives the improvement.

Significance. The paper attacks a practical and timely question: how to order GEC training data for fine-tuned models. Its strengths are the breadth of the empirical matrix (two model families, four model sizes, three benchmarks), the explicit attempt to control for data repetition through the 3E+2M+1H baseline in Table 3, the hard-to-easy ablation in Table 4, and the human-agreement sanity check in Figure 1. If the ordering effect is real, the method is simple and transferable: a strong LLM scores the data once, and smaller models are trained on the resulting curriculum. However, the manuscript provides no code or checkpoints, no multi-seed error bars, and no training-budget accounting, so the central claim that ordering rather than extra updates or arbitrary thresholds drives the gains is not yet fully established. The contribution is currently a promising empirical result conditional on those controls.

major comments (4)
  1. [§4.1, Table 3; §2.2; Appendix A.1 Table 5] The comparison between LLM-based CL (E->EM->EMH) and the 3E+2M+1H baseline is meant to show that curriculum ordering, not repeated exposure, produces the F0.5 gain, but the paper never reports the total number of gradient updates for either condition. The CL procedure in §2.2 trains "until the model converges" at each stage, and the only epoch count given (Appendix A.1, Table 5, "Epoches 3") is ambiguous as to whether it applies per stage. If each CL stage runs several epochs, an easy example is seen many more times than in the baseline, and the gap in Table 3 (e.g., 69.3 vs 68.6 for T5-xl on CoNLL14; 75.6 vs 74.8 on BEA19-test) could be partly a training-budget effect. The Introduction also describes the baseline as "summing up the data iteratively trained in curriculum learning," which does not exactly match the 3E+2M+1H description in §4.1. Please report epoch/step counts for every condition and clarify the baseline construction; ideally, train the mixed baseline for the same total number of updates and the same data-exposure counts as the CL schedule.
  2. [§2.1, Figure 1] The LLaMA2-70b difficulty scores are validated against human judgment on only 200 sentences, with no details on the number or background of the human raters and no inter-annotator agreement. More importantly, there is no evidence that the LLM scores predict the student models' learning difficulty, e.g., through correlation with per-example loss or convergence speed. The 1-3/4-7/8-10 cutoffs are fixed without any sensitivity analysis. Because the entire curriculum is defined by these scores and thresholds, the paper should add a threshold sweep (or at least report bucket sizes and stability under adjacent cutoffs) and a direct correlation between LLM difficulty scores and student-model learning dynamics.
  3. [§3.4, Table 2; §3.1] The claim of a "significant performance boost" is supported only by a terse "p<0.01" notation with no number of seeds, standard deviations, test procedure, or clarification that the BEA19 test set was excluded from significance testing, as stated in §3.1. The reported differences are small (0.4-1.2 F0.5), so single-run numbers are insufficient to establish stability. Please provide means and variances over at least three runs, or paired bootstrap confidence intervals, and state exactly which comparisons and datasets the significance tests cover.
  4. [§4.2, Table 4] The conclusion that "learning from hard to easy is almost ineffective" rests on a single model (T5-large), a single run, and no significance test. The hard-to-easy and easy-to-hard schedules are also not matched for total gradient updates. This ablation is the only direct evidence for the ordering claim, so it should be replicated on at least one more model and reported with error bars and a matched training budget.
minor comments (6)
  1. [Abstract, Limitations] There are typos in the abstract ("pretrianed") and in the Limitations section ("aome limitations"); the paper should be proofread.
  2. [Table 2 and throughout] The notation "T5-xx/LLaMA2-xx" is used without defining what "xx" ranges over; please spell out the model sizes or define the shorthand.
  3. [Figure 2, Appendix A.2, Figure 5] The prompt text appears with whitespace stripped ("Asagrammarcorrectionexpert..."), making it hard to read; additionally, the model output in Figure 5 includes a corrected sentence and prose explanation, and the post-editing procedure used to extract the numeric score is not described.
  4. [Table 1] The column labeled "SAME" (0.89M) is not explained in the text, and the sizes of easy/medium/hard sum to 1.31M rather than 2.2M; this presumably reflects exclusion of correct-correct pairs, but it should be stated explicitly.
  5. [§3.3] The phrase "This is the first time we have implemented these as comparable baselines" is unclear; it should say that the length-based CL baselines are newly implemented in this paper.
  6. [Figure 3] The y-axis is not labeled in the caption (it is F0.5), and the error-type names on the x-axis are too compressed to read; a larger figure or rotated labels would help.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the LLM difficulty scores come from a fixed external LLaMA2-70b prompt, the curriculum thresholds are fixed, and the central comparisons use controlled baselines; the training-budget confound is a correctness risk, not circularity.

full rationale

The paper's derivation chain is empirical rather than self-referential. The difficulty scores are produced by a fixed LLaMA2-70b prompt (Eq. 2 and Figure 5) with fixed thresholds (1-3 easy, 4-7 medium, 8-10 hard), independent of the student models, test labels, and benchmark scores. The training objectives in Eqs. (3)-(5) are standard supervised fine-tuning losses over curriculum tiers; nothing in these equations is defined in terms of the reported F0.5 results. The 3E+2M+1H baseline in Table 3 is an ablation that reuses the same data with the same proportions, intended to control for repeated exposure to easy data, and it is a legitimate comparison rather than a fitted prediction: its values are not derived from the CL result. The hard-to-easy ablation in Table 4 controls for ordering. The only substantive concern is that the CL stages are trained 'until the model converges' while the same-data baseline's epoch count is not reported, so the CL run may receive more total gradient updates than the baseline. That is a possible experimental confound affecting the strength of the ordering claim, but it is not a circularity: the claimed improvement is not equivalent by construction to the inputs, and no fitted parameter is renamed as a prediction. Self-citations to Fang et al. (2023b, 2023c) are used for implementation details and baseline results, not as load-bearing justification for the curriculum claim. The human-consistency check on 200 sentences is an external sanity check, not a fitted target. Therefore no circular step is present.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The method introduces no new physical or mathematical entities. The main assumptions are the reliability of LLM difficulty scores, the arbitrary tier thresholds, and the equivalence of the staged schedule to the same-data baseline in total training compute. None of these are independently verified with released artifacts or formal proofs.

free parameters (4)
  • Difficulty bucket thresholds = 1-3, 4-7, 8-10
    The boundaries that split the scored data into easy, medium, and hard are chosen ad hoc in Section 2.1. No sweep or validation-based selection is reported, and the results depend on this partition.
  • Number of curriculum stages = 3
    The paper always uses exactly three tiers. Alternative numbers of stages are not tested, so the method's sensitivity to this choice is unknown.
  • LLM scoring prompt and score range = 1-10 prompt with post-editing
    The prompt in Appendix A.2 and the post-editing rule for converting model outputs to numeric scores are hand-designed. No calibration of the scores against actual student learning difficulty is provided.
  • Stopping criterion per CL stage = 'until convergence' / epochs=3
    The paper describes each stage as training until convergence (Section 2.2) but also lists Epoches=3 in Appendix A.1. This ambiguity affects the total number of gradient updates per tier and the fairness of the same-data baseline.
assumptions (5)
  • domain assumption LLM-assigned difficulty scores are a valid proxy for the learning difficulty of GEC samples for the student models.
    Invoked in Section 2.1; only 200 sentences were compared with human ratings in Figure 1, and no direct link to student learning curves is shown.
  • domain assumption The three-stage easy-to-hard schedule is the operative cause of improvement, not the extra gradient updates on repeated easy data.
    Table 3 attempts to control this with the 3E+2M+1H baseline, but convergence-based stage lengths may not match that baseline's total updates.
  • domain assumption Standard fine-tuning on deduplicated cLang8 with the same settings as prior T5 GEC work is a valid baseline.
    Section 3.2 follows Fang et al. (2023b) settings, but the T5 hyperparameters themselves are not given in this paper.
  • standard math M2 and ERRANT metrics provide accepted and comparable evaluation of GEC systems.
    Used in Section 3.1; these are standard in the field, though different scorers for CoNLL14 and BEA19 complicate cross-benchmark comparisons.
  • standard math The reported t-test significance at p<0.01 is valid.
    Section 3.3 claims statistical significance but does not report variance, number of runs, or the exact test procedure, so the assumption cannot be audited.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LLMCL-GEC: Advancing Grammatical Error Correction with LLM-Driven Curriculum Learning." pith.science (2026). https://pith.science/paper/NTST334F

@misc{pith2026241212541,
  author       = {Pith},
  title        = {Pith review of: LLMCL-GEC: Advancing Grammatical Error Correction with LLM-Driven Curriculum Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NTST334F}},
  note         = {Machine review of arXiv:2412.12541}
}
read the original abstract

While large-scale language models (LLMs) have demonstrated remarkable capabilities in specific natural language processing (NLP) tasks, they may still lack proficiency compared to specialized models in certain domains, such as grammatical error correction (GEC). Drawing inspiration from the concept of curriculum learning, we have delved into refining LLMs into proficient GEC experts by devising effective curriculum learning (CL) strategies. In this paper, we introduce a novel approach, termed LLM-based curriculum learning, which capitalizes on the robust semantic comprehension and discriminative prowess inherent in LLMs to gauge the complexity of GEC training data. Unlike traditional curriculum learning techniques, our method closely mirrors human expert-designed curriculums. Leveraging the proposed LLM-based CL method, we sequentially select varying levels of curriculums ranging from easy to hard, and iteratively train and refine using the pretrianed T5 and LLaMA series models. Through rigorous testing and analysis across diverse benchmark assessments in English GEC, including the CoNLL14 test, BEA19 test, and BEA19 development sets, our approach showcases a significant performance boost over baseline models and conventional curriculum learning methodologies.

Figures

Figures reproduced from arXiv: 2412.12541 by the authors.

Figure 1
Figure 1. Different curriculum learning methods exhibit [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall framework of our proposed LLM-based CL method consists of two parts: one involves using [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The F0.5 scores reflect the performance on a selection of fine-grained error types within the CoNLL14 test set. The percentages provided in brackets represent the distribution of each error type. The findings indicate that our LLM-based CL method can significantly improve performance in correcting a majority of these fine-grained error types, surpassing the efficiency of both the Len-based method and other baseline … view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Progression of Recall, Precision, and F0.5 from Easy to Hard stages on CoNLL14 test set using Len-based CL and LLM-based CL methods on T5-xl and LLaMA2-13b. the two models exhibited contrasting performances with their two methods in terms of precision and recall scores…
Figure 5
Figure 5. Figure 5: The designed scoring prompts for the LLaMA2-70b model and evaluated the difficulty of correcting [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. From Silent Signals to Natural Language: A Dual-Stage Transformer-LLM Approach

    cs.CL 2025-09 conditional novelty 3.0 of 10

    A transformer ASR with GPT-2 post-correction reduces word error rate for EMG-based silent speech recognition from 36% to 30% on the Digital Voicing test set.

Reference graph

Works this paper leans on

57 extracted references · 12 canonical work pages · cited by 1 Pith paper

  1. [1]

    Alon Albalak, Yanai Elazar, Sang Michael Xie, Shayne Longpre, Nathan Lambert, Xinyi Wang, Niklas Muennighoff, Bairu Hou, Liangming Pan, Haewon Jeong, Colin Raffel, Shiyu Chang, Tatsunori Hashimoto, and William Yang Wang. 2024. https://arxiv.org/abs/2402.16827 A survey on data selection for language models . arXiv preprint arXiv:2402.16827

    arXiv 2024

  2. [2]

    Yoshua Bengio, J\' e r\^ o me Louradour, Ronan Collobert, and Jason Weston. 2009. https://doi.org/10.1145/1553374.1553380 Curriculum learning . In Proceedings of the 26th Annual International Conference on Machine Learning, ICML '09, page 41–48, New York, NY, USA. Association for Computing Machinery

    arXiv 2009

  3. [3]

    Andrey Bout, Alexander Podolskiy, Sergey Nikolenko, and Irina Piontkovskaya. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.355 Efficient grammatical error correction via multi-task training and optimized training schedule . In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 5800--5816, Singapore. Associati...

    doi:10.18653/v1/2023.emnlp-main.355 2023

  4. [4]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gr...

    2020

  5. [5]

    Andersen, and Ted Briscoe

    Christopher Bryant, Mariano Felice, istein E. Andersen, and Ted Briscoe. 2019. https://doi.org/10.18653/v1/W19-4406 The BEA -2019 shared task on grammatical error correction . In Proceedings of the Fourteenth Workshop on Innovative Use of NLP for Building Educational Applications, pages 52--75, Florence, Italy. Association for Computational Linguistics

    doi:10.18653/v1/w19-4406 2019

  6. [6]

    Christopher Bryant, Mariano Felice, and Ted Briscoe. 2017. https://doi.org/10.18653/v1/P17-1074 Automatic annotation and evaluation of error types for grammatical error correction . In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 793--805, Vancouver, Canada. Association for Computat...

    doi:10.18653/v1/p17-1074 2017

  7. [7]

    Yo Joong Choe, Jiyeon Ham, Kyubyong Park, and Yeoil Yoon. 2019. https://doi.org/10.18653/v1/W19-4423 A neural grammatical error correction system built on better pre-training and sequential transfer learning . In Proceedings of the Fourteenth Workshop on Innovative Use of NLP for Building Educational Applications, pages 213--227, Florence, Italy. Associat...

    work page doi:10.18653/v1/w19-4423 2019

  8. [8]

    Hovy, and Louis-Philippe Morency

    Volkan Cirik, Eduard H. Hovy, and Louis-Philippe Morency. 2016. https://api.semanticscholar.org/CorpusID:17896684 Visualizing and understanding curriculum learning for long short-term memory networks . ArXiv, abs/1611.06204

    arXiv 2016

Show all 57 references
  1. [9]

    Daniel Dahlmeier and Hwee Tou Ng. 2012. https://aclanthology.org/N12-1067 Better evaluation for grammatical error correction . In Proceedings of the 2012 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technologies , p...

    2012

  2. [10]

    Wong, Xiang Wan, Lidia S

    Tao Fang, Jinpeng Hu, Derek F. Wong, Xiang Wan, Lidia S. Chao, and Tsung-Hui Chang. 2023 a . https://doi.org/10.18653/v1/2023.findings-acl.594 Improving grammatical error correction with multimodal feature integration . In Findings of the Association for Computational Linguist...

    2023 doi

  3. [11]

    Wong, Runzhe Zhan, Liang Ding, Lidia S

    Tao Fang, Xuebo Liu, Derek F. Wong, Runzhe Zhan, Liang Ding, Lidia S. Chao, Dacheng Tao, and Min Zhang. 2023 b . https://doi.org/10.18653/v1/2023.findings-acl.223 T rans GEC : Improving grammatical error correction with translationese . In Findings of the Association for Compu...

    2023 doi

  4. [12]

    Wong, Jinpeng Hu, Lidia S

    Tao Fang, Shu Yang, Kaixin Lan, Derek F. Wong, Jinpeng Hu, Lidia S. Chao, and Yue Zhang. 2023 c . https://arxiv.org/abs/2304.01746 Is chatgpt a highly fluent grammatical error correction system? a comprehensive evaluation . arXiv preprint arXiv:2304.01746

    2023 arXiv

  5. [13]

    Amr Hendy, Mohamed Abdelrehim, Amr Sharaf, Vikas Raunak, Mohamed Gabr, Hitokazu Matsushita, Young Jin Kim, Mohamed Afify, and Hany Hassan Awadalla. 2023. https://arxiv.org/pdf/2302.09210.pdf How good are gpt models at machine translation? a comprehensive evaluation . arXiv pre...

    2023 arXiv

  6. [14]

    Ziyang Hui, Chong Feng, and Tianfu Zhang. 2022. Review-based curriculum learning for neural machine translation. In Machine Translation, pages 24--36, Singapore. Springer Nature Singapore

    2022

  7. [15]

    Wenxiang Jiao, Wenxuan Wang, Jen-tse Huang, Xing Wang, and Zhaopeng Tu. 2023. https://www.researchgate.net/publication/367359399_Is_ChatGPT_A_Good_Translator_A_Preliminary_Study Is chatgpt a good translator? a preliminary study . arXiv preprint arXiv:2301.08745, 1(10)

    2023

  8. [16]

    Masahiro Kaneko, Masato Mita, Shun Kiyono, Jun Suzuki, and Kentaro Inui. 2020. https://doi.org/10.18653/v1/2020.acl-main.391 Encoder-decoder models can benefit from pre-trained masked language models in grammatical error correction . In Proceedings of the 58th Annual Meeting o...

    2020 doi

  9. [17]

    Satoru Katsumata and Mamoru Komachi. 2020. https://aclanthology.org/2020.aacl-main.83 Stronger baselines for grammatical error correction using a pretrained encoder-decoder model . In Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computat...

    2020

  10. [18]

    Tom Kocmi and Ond r ej Bojar. 2017. https://doi.org/10.26615/978-954-452-049-6_050 Curriculum learning and minibatch bucketing in neural machine translation . In Proceedings of the International Conference Recent Advances in Natural Language Processing, RANLP 2017 , pages 379-...

    2017 doi

  11. [19]

    Gaurav Kumar, George Foster, Colin Cherry, and Maxim Krikun. 2019. https://doi.org/10.18653/v1/N19-1208 Reinforcement learning based curriculum optimization for neural machine translation . In Proceedings of the 2019 Conference of the North A merican Chapter of the Association...

    2019 doi

  12. [20]

    Shaopeng Lai, Qingyu Zhou, Jiali Zeng, Zhongli Li, Chao Li, Yunbo Cao, and Jinsong Su. 2022. https://doi.org/10.18653/v1/2022.findings-acl.254 Type-driven multi-turn corrections for grammatical error correction . In Findings of the Association for Computational Linguistics: AC...

    2022 doi

  13. [21]

    Wong, Yang Gao, Heyan Huang, and Min Zhang

    Yinghao Li, Xuebo Liu, Shuo Wang, Peiyuan Gong, Derek F. Wong, Yang Gao, Heyan Huang, and Min Zhang. 2023. https://doi.org/10.18653/v1/2023.acl-long.380 T emplate GEC : Improving grammatical error correction with detection template . In Proceedings of the 61st Annual Meeting o...

    2023 doi

  14. [22]

    Wong, and Lidia S

    Xuebo Liu, Houtim Lai, Derek F. Wong, and Lidia S. Chao. 2020. https://doi.org/10.18653/v1/2020.acl-main.41 Norm-based curriculum learning for neural machine translation . In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 427--43...

    2020 doi

  15. [23]

    Mengsay Loem, Masahiro Kaneko, Sho Takase, and Naoaki Okazaki. 2023. https://doi.org/10.18653/v1/2023.bea-1.18 Exploring effectiveness of GPT -3 in grammatical error correction: A study on performance and controllability in prompt-based methods . In Proceedings of the 18th Wor...

    2023 doi

  16. [24]

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

    2024 arXiv

  17. [25]

    Hwee Tou Ng, Siew Mei Wu, Ted Briscoe, Christian Hadiwinoto, Raymond Hendy Susanto, and Christopher Bryant. 2014. https://doi.org/10.3115/v1/W14-1701 The C o NLL -2014 shared task on grammatical error correction . In Proceedings of the Eighteenth Conference on Computational Na...

    2014 doi

  18. [26]

    Hwee Tou Ng, Siew Mei Wu, Yuanbin Wu, Christian Hadiwinoto, and Joel Tetreault. 2013. https://aclanthology.org/W13-3601 The C o NLL -2013 shared task on grammatical error correction . In Proceedings of the Seventeenth Conference on Computational Natural Language Learning: Shar...

    2013

  19. [27]

    Kostiantyn Omelianchuk, Vitaliy Atrasevych, Artem Chernodub, and Oleksandr Skurzhanskyi. 2020. https://doi.org/10.18653/v1/2020.bea-1.16 GECT o R -- grammatical error correction: Tag, not rewrite . In Proceedings of the Fifteenth Workshop on Innovative Use of NLP for Building ...

    2020 doi

  20. [28]

    OpenAI. 2024. https://arxiv.org/abs/2303.08774 Gpt-4 technical report . arXiv preprint arXiv:2303.08774

    2024 arXiv

  21. [29]

    Wenbo Pan, Qiguang Chen, Xiao Xu, Wanxiang Che, and Libo Qin. 2023. https://arxiv.org/abs/2304.04256 A preliminary evaluation of chatgpt for zero-shot dialogue understanding . arXiv preprint arXiv:2304.04256

    2023 arXiv

  22. [30]

    Emmanouil Antonios Platanios, Otilia Stretcu, Graham Neubig, Barnabas Poczos, and Tom Mitchell. 2019. https://doi.org/10.18653/v1/N19-1119 Competence-based curriculum learning for neural machine translation . In Proceedings of the 2019 Conference of the North A merican Chapter...

    2019 doi

  23. [31]

    Muhammad Reza Qorib, Seung-Hoon Na, and Hwee Tou Ng. 2022. https://doi.org/10.18653/v1/2022.naacl-main.143 Frustratingly easy system combination for grammatical error correction . In Proceedings of the 2022 Conference of the North American Chapter of the Association for Comput...

    2022 doi

  24. [32]

    Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, Peter J Liu, et al. 2020. https://jmlr.org/papers/volume21/20-074/20-074.pdf Exploring the limits of transfer learning with a unified text-to-text transformer. Journal o...

    2020

  25. [33]

    Mengye Ren, Wenyuan Zeng, Bin Yang, and Raquel Urtasun. 2019. https://arxiv.org/abs/1803.09050 Learning to reweight examples for robust deep learning . arXiv.1803.09050

    2019 arXiv

  26. [34]

    Sascha Rothe, Jonathan Mallinson, Eric Malmi, Sebastian Krause, and Aliaksei Severyn. 2021. https://doi.org/10.18653/v1/2021.acl-short.89 A simple recipe for multilingual grammatical error correction . In Proceedings of the 59th Annual Meeting of the Association for Computatio...

    2021 doi

  27. [35]

    Shreyas Saxena, Oncel Tuzel, and Dennis DeCoste. 2019. https://papers.nips.cc/paper/9289-data-parameters-a-new-family-of-parameters-for-learning-a-differentiable-curriculum.pdf Data parameters: A new family of parameters for learning a differentiable curriculum . In NeurIPS

    2019

  28. [36]

    Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper, and Bryan Catanzaro. 2020. https://arxiv.org/abs/1909.08053 Megatron-lm: Training multi-billion parameter language models using model parallelism . Preprint, arXiv:1909.08053

    2020 arXiv

  29. [37]

    Spitkovsky, Hiyan Alshawi, and Daniel Jurafsky

    Valentin I. Spitkovsky, Hiyan Alshawi, and Daniel Jurafsky. 2010. https://aclanthology.org/N10-1116 From baby steps to leapfrog: How `` less is more '' in unsupervised dependency parsing . In Human Language Technologies: The 2010 Annual Conference of the North A merican Chapte...

    2010

  30. [38]

    Felix Stahlberg and Shankar Kumar. 2021. https://aclanthology.org/2021.bea-1.4 Synthetic data generation for grammatical error correction with tagged corruption models . In Proceedings of the 16th Workshop on Innovative Use of NLP for Building Educational Applications, pages 3...

    2021

  31. [39]

    Xin Sun, Tao Ge, Furu Wei, and Houfeng Wang. 2021. https://doi.org/10.18653/v1/2021.acl-long.462 Instantaneous grammatical error correction with shallow aggressive decoding . In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11t...

    2021 doi

  32. [40]

    Maksym Tarnavskyi, Artem Chernodub, and Kostiantyn Omelianchuk. 2022. https://doi.org/10.18653/v1/2022.acl-long.266 Ensembling and knowledge distilling of large sequence taggers for grammatical error correction . In Proceedings of the 60th Annual Meeting of the Association for...

    2022 doi

  33. [41]

    Phan, Xingdi Yuan, Jinfeng Rao, Siu Cheung Hui, and Aston Zhang

    Yi Tay, Shuohang Wang, Anh Tuan Luu, Jie Fu, Minh C. Phan, Xingdi Yuan, Jinfeng Rao, Siu Cheung Hui, and Aston Zhang. 2019. https://doi.org/10.18653/v1/P19-1486 Simple and effective curriculum pointer-generator networks for reading comprehension over long narratives . In Proce...

    2019 doi

  34. [42]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timoth \'e e Lacroix, Baptiste Rozi \`e re, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023 a . https://arxiv.org/abs/2302.13971 Llama: Open and efficient foundation language models . arXiv...

    2023 arXiv

  35. [43]

    Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, W...

    2023 arXiv

  36. [44]

    Zhaohong Wan, Xiaojun Wan, and Wenguang Wang. 2020. https://doi.org/10.18653/v1/2020.coling-main.200 Improving grammatical error correction with data augmentation by editing latent representation . In Proceedings of the 28th International Conference on Computational Linguistic...

    2020 doi

  37. [45]

    Deze Wang, Zhouyang Jia, Shanshan Li, Yue Yu, Yun Xiong, Wei Dong, and Xiangke Liao. 2022. https://doi.org/10.1145/3510003.3510062 Bridging pre-trained models and downstream tasks for source code understanding . In Proceedings of the 44th International Conference on Software E...

    2022

  38. [46]

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

    2024 arXiv

  39. [47]

    Helen Yannakoudakis, istein E Andersen, Ardeshir Geranpayeh, Ted Briscoe, and Diane Nicholls. 2018. https://www.tandfonline.com/doi/citedby/10.1080/08957347.2018.1464447?scroll=top&needAccess=true&role=tab Developing an automated writing placement system for esl learners . App...

    2018

  40. [48]

    Helen Yannakoudakis, Ted Briscoe, and Ben Medlock. 2011. https://aclanthology.org/P11-1019 A new dataset and method for automatically grading ESOL texts . In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, p...

    2011

  41. [49]

    Zheng Yuan and Ted Briscoe. 2016. https://doi.org/10.18653/v1/N16-1042 Grammatical error correction using neural machine translation . In Proceedings of the 2016 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technolo...

    2016 doi

  42. [50]

    Martindale, Paul McNamee, Kevin Duh, and Marine Carpuat

    Xuan Zhang, Manish Kumar, Huda Khayrallah, Kenton Murray, Jeremy Gwinnup, Marianna J. Martindale, Paul McNamee, Kevin Duh, and Marine Carpuat. 2018. https://api.semanticscholar.org/CorpusID:53295888 An empirical exploration of curriculum learning for neural machine translation...

    2018 arXiv

  43. [51]

    Yue Zhang, Leyang Cui, Deng Cai, Xinting Huang, Tao Fang, and Wei Bi. 2023. https://arxiv.org/abs/2305.13225 Multi-task instruction tuning of llama for specific scenarios: A preliminary study on writing assistance . arXiv preprint arXiv:2305.13225

    2023 arXiv

  44. [52]

    Yue Zhang, Zhenghua Li, Zuyi Bao, Jiacheng Li, Bo Zhang, Chen Li, Fei Huang, and Min Zhang. 2022. https://doi.org/10.18653/v1/2022.naacl-main.227 M u CGEC : a multi-reference multi-source evaluation dataset for C hinese grammatical error correction . In Proceedings of the 2022...

    2022 doi

  45. [53]

    Xueliang Zhao, Wei Wu, Can Xu, Chongyang Tao, Dongyan Zhao, and Rui Yan. 2020. https://doi.org/10.18653/v1/2020.emnlp-main.272 Knowledge-grounded dialogue generation with pre-trained language models . In Proceedings of the 2020 Conference on Empirical Methods in Natural Langua...

    2020 doi

  46. [54]

    Wangchunshu Zhou, Tao Ge, Chang Mu, Ke Xu, Furu Wei, and Ming Zhou. 2020 a . https://doi.org/10.18653/v1/2020.findings-emnlp.30 Improving grammatical error correction with machine translation pairs . In Findings of the Association for Computational Linguistics: EMNLP 2020, pag...

    2020 doi

  47. [55]

    Wong, Yu Wan, and Lidia S

    Yikai Zhou, Baosong Yang, Derek F. Wong, Yu Wan, and Lidia S. Chao. 2020 b . https://doi.org/10.18653/v1/2020.acl-main.620 Uncertainty-aware curriculum learning for neural machine translation . In Proceedings of the 58th Annual Meeting of the Association for Computational Ling...

    2020 doi

  48. [56]

    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...

  49. [57]

    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 gl...

Pith tools

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