Pith. sign in

REVIEW 4 major objections 5 minor 42 references

Trigger$^3$: Refining Query Correction via Adaptive Model Selector

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read Trigger3 claims that a three-trigger cascade—small model first, LLM only when needed, original query as fallback—outperforms query-correction baselines while cutting LLM calls.

desk verdict A sensible three-trigger router for query correction with broad experiments, but the synthetic 97.8%-error datasets leave external validity unproven. read the letter →

arxiv 2412.12701 v1 pith:7X3V6FXU submitted 2024-12-17 cs.CL

classification cs.CL
keywords querycorrectionmodelselectionlarge-smallcollaborationLLMcascadetriggerclassifierssearchengineChineseefficiency
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 search-query correction can be improved by treating it as an adaptive model-selection problem: a small correction model handles routine typos, a large language model is called only when the small model's rewrite looks inadequate, and a final check can fall back to the original query rather than accept a bad rewrite. This three-trigger cascade, named Trigger3, is trained with binary BERT classifiers and is model-agnostic; the paper tests it with three small models and two LLMs on Chinese search datasets. The reported result is higher character-level F0.5 than all correction baselines while the fraction of queries sent to the LLM stays low. A sympathetic reader would take away that the expensive LLM should be a targeted second opinion, not a default corrector, and that returning the original query is a legitimate correction outcome.

What carries the argument

The load-bearing object is the adaptive model selector made of three binary classifiers, each a BERT model using the [CLS] representation: CT scores P(Incorrect|x), LT scores P(Invoke LLM | x, ysmall) with [SEP] separating the two texts, and FT scores P(Return x | x, yc). Training labels for LT and FT are derived from character-edit indicators (TP, FP, FN) comparing the small model's edits and the LLM's edits against the gold query: LT positive when the small model misses or errs but the LLM does not, FT positive when neither model produces a correct edit. These trigger labels turn the model-selection question into three binary classification problems, which is what makes the framework trainable and model-agnostic.

What would settle it

Run Trigger3 on a held-out sample of naturally occurring user queries with human-labeled corrections, including correct queries and domain-specific terms, and compare Char-F0.5 and LLM coverage against Margin Sampling; if the F0.5 advantage shrinks substantially or LLM coverage rises above the synthetic-test values, the triggers are tuned to the corruption pipeline rather than to real errors.

Watch

Extended reading notes

Core claim

The central claim is that Trigger3, by chaining a Correction Trigger (CT), an LLM Trigger (LT), and a Fallback Trigger (FT), outperforms correction baselines while maintaining efficiency. Given an original query x, CT decides whether correction is needed; if so, the small model produces ysmall; LT, looking at (x, ysmall), decides whether the LLM is needed; and FT, looking at (x, yc), decides whether to return the original query. In the main tables, the best Trigger3 configuration reaches Char-F0.5 74.60 on the Commercial dataset with GECToR, compared to 67.48 for Margin Sampling, and does so with LLM coverage 32.09 versus 45.63. The paper interprets this as evidence that a cascade with correct-query filtering, targeted LLM invocation, and fallback to the original query is a better large-small model collaboration than routing or pure cascading.

Load-bearing premise

The evaluation datasets—a commercial set that is 97.8% erroneous and half synthetic, plus a public set corrupted the same way—must resemble the queries a real search engine sees; if real user errors are rarer or differently distributed, the measured F0.5 gains and trigger thresholds will not transfer.

Editorial extensions

If this is right

  • Production search engines can add Trigger3 as a drop-in layer on top of an existing small correction model, sending only a minority of queries to an LLM.
  • The LT threshold gives operators a direct control over LLM coverage, letting them trade a small amount of F0.5 for lower serving cost.
  • Formalizing 'return the original query' as a decision reduces the damage from over-correction, which the paper observes in both small models and LLMs.
  • Because the trigger labels are defined purely by edit-level TP/FP/FN, the same recipe can be re-trained for new small-model/LLM pairs without redesigning the framework.

Reading between the lines

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

  • The paper does not report a sweep of CT/LT/FT thresholds; tuning them per query class or per traffic mix is a natural next step and could shift the efficiency-quality tradeoff further.
  • The trigger labels require gold corrections for training; in a live system without gold labels, the same edit-indicator scheme could be bootstrapped from high-confidence online correction logs.
  • The framework's logic is generic to any rewrite task where a cheap model and an expensive model disagree, such as style transfer or query expansion; whether it transfers is an open question.
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 / 5 minor

Summary. The paper proposes Trigger3, a three-stage adaptive selector for Chinese query correction. A Correction Trigger (CT) decides whether an input query should be corrected; if so, a small model rewrites it; an LLM Trigger (LT) decides whether the LLM should be invoked using the original query and the small-model rewrite; and a Fallback Trigger (FT) decides whether to return the original query when neither model produces a usable correction. The three triggers are BERT-based binary classifiers trained with cross-entropy on labels derived from edit-level agreement among the small model, the LLM, and ground-truth corrections. Experiments compare against six routing/cascading baselines on two datasets, three small models, and two LLMs, reporting character- and word-level F0.5 as well as LLM coverage. The central claim is that Trigger3 achieves the best correction quality while making fewer LLM calls than the baselines.

Significance. The framework is simple, model-agnostic, and the evaluation is more extensive than is typical for this task: three small models, two LLMs, two datasets, significance tests, and an explicit efficiency metric (LLM coverage). If the results transfer to real query traffic, the contribution is a practical model-selection layer that could reduce LLM cost while improving correction quality. However, transfer is not established because the test distribution is dominated by synthetic corruptions and an unnaturally high error rate. In addition, one stated conclusion (no recall decrease) is contradicted by the paper's own Table 2, and several reproducibility details need work. Overall the internal comparisons are credible, but the external validity of the central claim remains to be demonstrated.

major comments (4)
  1. [4.1, Table 1; 3.3] The efficiency claim depends on CT correctly identifying correct queries, yet the evaluation gives CT almost no correct queries to classify: the Commercial test set is 97.8% erroneous, so only about 324 of its 14,737 queries are correct, and half of all Commercial errors are synthetic character-level corruptions. The CT threshold is therefore calibrated to a distribution in which almost every query is wrong; its false-positive rate on naturally correct queries, which dominate real traffic, is effectively unmeasured. Please add a natural-error holdout, or at least a correct-query-only evaluation, and report CT specificity/fall-out. Without such evidence, the reported F0.5 and LLM-coverage numbers do not support the production-efficiency claim.
  2. [4.2, Table 2] The first bullet of Section 4.2 states that Trigger3 'ensur[es] no decrease in recall rate.' Table 2 directly contradicts this for GECToR on Commercial: character-level recall drops from 76.30 (small model) to 74.33 (Trigger3). The claim should be corrected or qualified (for example, 'no systematic decrease' with a list of exceptions), and the discussion should address whether this recall loss is acceptable given the F0.5 gains.
  3. [3.3-3.5, Algorithm 1] The operation of each trigger is specified only as a probability exceeding 'a certain threshold' in Eqs. (4)-(6), and Algorithm 1 writes pCT = 1 as the branch condition. The paper never states the thresholds used at inference, how they were chosen, or whether they were tuned on a validation split. Because F0.5 and LLM coverage are directly controlled by these thresholds, the reported efficiency/performance trade-off is not reproducible and could be optimistic if thresholds were selected on the test set. Please report the threshold values, the selection procedure, and the resulting operating point.
  4. [A.1, Table 2] Margin Sampling is a key baseline, but Appendix A.1 defines it as deciding based on 'the uncertainty of the first token output by the small model.' GECToR is a Seq2Edit model that outputs edit tags, not a generated first token; the paper does not explain how a first-token margin is obtained for GECToR. If the baseline was adapted in an ad hoc way, the comparison on GECToR is not meaningful. Please specify the exact uncertainty measure used for each small model.
minor comments (5)
  1. [Title and Abstract] The title and abstract use the notation Trigger$^3$, while the body and tables consistently use Trigger3; please unify the notation.
  2. [Figure 1] Figure 1 appears to show the same three example panels under labels (a), (b), and (c), which is likely a layout error; the captions describe different correction types, but the actual examples are identical.
  3. [Table 2] The row label 'Cascade (LLM)' for mT5 differs from 'Cascading (LLM)' used elsewhere in the same table and in Table 5; the terminology is inconsistent.
  4. [Appendix C] The prompt templates in Appendix C have spaces removed (for example, 'Pleasedetectandcorrecterrors'), which makes the prompts hard to read; please reformat them.
  5. [Section 4.1] Section 4.1 says 'The training of the LLMs and the small models can be found in Section 4.1,' but the actual training details are in Appendix A.2; the cross-reference should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the trigger classifiers are supervised routing modules trained on separate labels, and the reported gains are measured on held-out test sets.

full rationale

The paper's claimed derivation—that chaining Correction Trigger, LLM Trigger, and Fallback Trigger improves correction F0.5 while reducing LLM coverage—rests on supervised classifiers trained with binary cross-entropy (Eq. 7) on labels constructed in Section 3.6 from edit-level comparisons (TP/FP/FN) between the small model, LLM, and ground truth. These labels are computed on the training split; evaluation in Table 2 uses held-out Commercial and QQ test sets with the ChERRANT scorer. The trigger thresholds are the only free parameters, and the reported LLM coverage is measured from the executed pipeline rather than derived from the routing labels. The paper's citations to prior work by co-author Xiao Zhang (Zhang, Liao, and Liao 2019; Zhang and Liao 2020) appear only in the Related Work discussion of model selection and are not load-bearing for the proposed method or its evaluation. The strongest concern—synthetic corruptions and high error rates in dataset construction—affects external validity, not circularity, because the test labels are not used to construct the trigger labels and the final metric is not a restatement of any fitted quantity.

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

The central claim is empirical and rests on supervised routing; no new physical or mathematical entities are introduced. The main unstated inputs are the dataset construction choices and the efficiency proxy, plus the domain assumption that returning the original query is safer than a wrong rewrite.

free parameters (3)
  • Trigger decision thresholds for CT, LT, FT = unreported (Algorithm 1 implies p=1 branch, likely 0.5 default)
    The thresholds determine the precision/recall tradeoff of each trigger and are not reported or swept in the paper.
  • Synthetic corruption rates and operation mix = unreported
    Dataset construction uses homophone substitution, near-sound replacement, transposition, and random character edits with unspecified probabilities; this determines the error distribution and therefore the measured F0.5.
  • Negative sampling ratio for LT and FT training = equal to positive count, exact selection not specified
    The negative sampling procedure is described only as excluding positives and matching the positive count; the sampled subset affects trigger training.
assumptions (4)
  • domain assumption Smaller models are more inference-efficient than larger models, so LLM coverage is a valid efficiency proxy.
    Invoked in Section 4.4 to equate efficiency with the fraction of queries handled by the LLM; no wall-clock latency or cost measurement is reported.
  • domain assumption Synthetic corruptions and high-confidence online correction logs produce a representative sample of real query errors.
    Section 4.1 constructs both datasets this way; if the corruption distribution differs from real errors, the trigger thresholds and F0.5 rankings may not transfer.
  • domain assumption Returning the original query is preferable when both models' corrections are wrong.
    Section 3.5 justifies FT by appealing to user experience and LLM refusal research, but the paper does not measure search quality or user satisfaction directly.
  • domain assumption Edit-level TP/FP/FN indicators from the ChERRANT-style scorer correctly characterize whether a model's correction is beneficial.
    Section 3.6 defines LT and FT training labels from these indicators; errors in edit alignment would propagate into trigger supervision.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Trigger$^3$: Refining Query Correction via Adaptive Model Selector." pith.science (2026). https://pith.science/paper/7X3V6FXU

@misc{pith2026241212701,
  author       = {Pith},
  title        = {Pith review of: Trigger$^3$: Refining Query Correction via Adaptive Model Selector},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7X3V6FXU}},
  note         = {Machine review of arXiv:2412.12701}
}
abstract

In search scenarios, user experience can be hindered by erroneous queries due to typos, voice errors, or knowledge gaps. Therefore, query correction is crucial for search engines. Current correction models, usually small models trained on specific data, often struggle with queries beyond their training scope or those requiring contextual understanding. While the advent of Large Language Models (LLMs) offers a potential solution, they are still limited by their pre-training data and inference cost, particularly for complex queries, making them not always effective for query correction. To tackle these, we propose Trigger$^3$, a large-small model collaboration framework that integrates the traditional correction model and LLM for query correction, capable of adaptively choosing the appropriate correction method based on the query and the correction results from the traditional correction model and LLM. Trigger$^3$ first employs a correction trigger to filter out correct queries. Incorrect queries are then corrected by the traditional correction model. If this fails, an LLM trigger is activated to call the LLM for correction. Finally, for queries that no model can correct, a fallback trigger decides to return the original query. Extensive experiments demonstrate Trigger$^3$ outperforms correction baselines while maintaining efficiency.

Figures

Figures reproduced from arXiv: 2412.12701 by the authors.

Figure 1
Figure 1. Examples of query correction, where the red char [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The architecture of the proposed framework [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Average LLM Coverage of Trigger3 and the three frameworks when the LLM is Qwen1.5-7B-Chat. The lower the bar, the better. flects to some extent that the errors Seq2Seq and LLM can solve may be more alike. However, as the errors that the LLM and Seq2Seq small model can correct are different, this can also enhance the base model’s correction performance. We perform the experiment with similar conclusions when LLM is B… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: LLM Templates of Zero-shot, Few-shot and Few [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 19 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  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]

    Awasthi, A.; Sarawagi, S.; Goyal, R.; Ghosh, S.; and Piratla, V. 2019. Parallel Iterative Edit Models for Local Sequence Transduction. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), 4260--4270

  4. [4]

    Bai, J.; Bai, S.; Chu, Y.; Cui, Z.; Dang, K.; Deng, X.; Fan, Y.; Ge, W.; Han, Y.; Huang, F.; Hui, B.; Ji, L.; Li, M.; Lin, J.; Lin, R.; Liu, D.; Liu, G.; Lu, C.; Lu, K.; Ma, J.; Men, R.; Ren, X.; Ren, X.; Tan, C.; Tan, S.; Tu, J.; Wang, P.; Wang, S.; Wang, W.; Wu, S.; Xu, B.; Xu, J.; Yang, A.; Yang, H.; Yang, J.; Yang, S.; Yao, Y.; Yu, B.; Yuan, H.; Yuan,...

  5. [5]

    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. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33: 1877--1901

  6. [6]

    Chen, J.; Lin, H.; Han, X.; and Sun, L. 2024. Benchmarking large language models in retrieval-augmented generation. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 17754--17762

  7. [7]

    Coyne, S.; and Sakaguchi, K. 2023. An Analysis of GPT-3's Performance in Grammatical Error Correction. arXiv preprint arXiv:2303.14342

  8. [8]

    Davis, C.; Caines, A.; Andersen, .; Taslimipoor, S.; Yannakoudakis, H.; Yuan, Z.; Bryant, C.; Rei, M.; and Buttery, P. 2024. Prompting open-source and commercial language models for grammatical error correction of English learner text. arXiv preprint arXiv:2401.07702

Show all 42 references
  1. [9]

    Devlin, J.; Chang, M.-W.; Lee, K.; and Toutanova, K. 2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Tec...

  2. [10]

    Ding, D.; Mallick, A.; Wang, C.; Sim, R.; Mukherjee, S.; R \"u hle, V.; Lakshmanan, L. V. S.; and Awadallah, A. H. 2024. Hybrid LLM : Cost-Efficient and Quality-Aware Query Routing. In The Twelfth International Conference on Learning Representations

  3. [11]

    Ding, J.; Tarokh, V.; and Yang, Y. 2018. Model selection techniques: An overview. IEEE Signal Processing Magazine, 35(6): 16--34

  4. [12]

    Fan, Y.; Jiang, F.; Li, P.; and Li, H. 2023. Grammargpt: Exploring open-source llms for native chinese grammatical error correction with supervised fine-tuning. In CCF International Conference on Natural Language Processing and Chinese Computing, 69--80. Springer

  5. [13]

    F.; Hu, J.; Chao, L

    Fang, T.; Yang, S.; Lan, K.; Wong, D. F.; Hu, J.; Chao, L. S.; and Zhang, Y. 2023. Is chatgpt a highly fluent grammatical error correction system? a comprehensive evaluation. arXiv preprint arXiv:2304.01746

  6. [14]

    J.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; Chen, W.; et al

    Hu, E. J.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; Chen, W.; et al. 2021. LoRA: Low-Rank Adaptation of Large Language Models. In International Conference on Learning Representations

  7. [15]

    P.; and Ba, J

    Kingma, D. P.; and Ba, J. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

  8. [16]

    Li, Y.; Huang, H.; Ma, S.; Jiang, Y.; Li, Y.; Zhou, F.; Zheng, H.-T.; and Zhou, Q. 2023. On the (in) effectiveness of large language models for chinese text correction. arXiv preprint arXiv:2307.09007

  9. [17]

    Liang, D.; Zheng, C.; Guo, L.; Cui, X.; Xiong, X.; Rong, H.; and Dong, J. 2020. BERT enhanced neural machine translation and sequence tagging model for Chinese grammatical error diagnosis. In Proceedings of the 6th Workshop on Natural Language Processing Techniques for Educati...

  10. [19]

    Lu, K.; Yuan, H.; Lin, R.; Lin, J.; Yuan, Z.; Zhou, C.; and Zhou, J. 2023. Routing to the expert: Efficient reward-guided ensemble of large language models. arXiv preprint arXiv:2311.08692

  11. [20]

    P.; Mishra, S.; Zhou, P.; Gupta, A.; Rajagopal, D.; Kappaganthu, K.; Yang, Y.; et al

    Madaan, A.; Aggarwal, P.; Anand, A.; Potharaju, S. P.; Mishra, S.; Zhou, P.; Gupta, A.; Rajagopal, D.; Kappaganthu, K.; Yang, Y.; et al. 2023. Automix: Automatically mixing language models. arXiv preprint arXiv:2310.12963

  12. [21]

    Omelianchuk, K.; Atrasevych, V.; Chernodub, A.; and Skurzhanskyi, O. 2020. GECToR--Grammatical Error Correction: Tag, Not Rewrite. In Proceedings of the Fifteenth Workshop on Innovative Use of NLP for Building Educational Applications, 163--170

  13. [22]

    Ouyang, L.; Wu, J.; Jiang, X.; Almeida, D.; Wainwright, C.; Mishkin, P.; Zhang, C.; Agarwal, S.; Slama, K.; Ray, A.; et al. 2022. Training language models to follow instructions with human feedback. Advances in neural information processing systems, 35: 27730--27744

  14. [23]

    Pande, M.; Kakkar, V.; Bansal, M.; Kumar, S.; Sharma, C.; Malhotra, H.; and Mehta, P. 2022. Learning-to-Spell: Weak Supervision based Query Correction in E-Commerce Search with Small Strong Labels. In Proceedings of the 31st ACM International Conference on Information & Knowle...

  15. [24]

    Raffel, C.; Shazeer, N.; Roberts, A.; Lee, K.; Narang, S.; Matena, M.; Zhou, Y.; Li, W.; and Liu, P. J. 2020. Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of machine learning research, 21(140): 1--67

  16. [25]

    Ram \' rez, G.; Birch, A.; and Titov, I. 2024. Optimising Calls to Large Language Models with Uncertainty-Based Two-Tier Selection. arXiv preprint arXiv:2405.02134

  17. [26]

    S akota, M.; Peyrard, M.; and West, R. 2024. Fly-swat or cannon? cost-effective language model choice via meta-modeling. In Proceedings of the 17th ACM International Conference on Web Search and Data Mining, 606--615

  18. [27]

    Shao, Y.; Geng, Z.; Liu, Y.; Dai, J.; Yan, H.; Yang, F.; Li, Z.; Bao, H.; and Qiu, X. 2024. Cpt: A pre-trained unbalanced transformer for both chinese language understanding and generation. Science China Information Sciences, 67(5): 1--13

  19. [28]

    Shen, C.; Zhang, X.; Shi, T.; Zhang, C.; Xie, G.; and Xu, J. 2024. A survey of controllable learning: M ethods and applications in information retrieval. arXiv preprint arXiv:2308.05374

  20. [29]

    Sun, H.; Zhang, Z.; Deng, J.; Cheng, J.; and Huang, M. 2023. Safety assessment of chinese large language models. arXiv preprint arXiv:2304.10436

  21. [30]

    Wang, W.; Bi, B.; Yan, M.; Wu, C.; Xia, J.; Bao, Z.; Peng, L.; and Si, L. 2019. StructBERT: Incorporating Language Structures into Pre-training for Deep Language Understanding. In International Conference on Learning Representations

  22. [31]

    Wolf, T.; Debut, L.; Sanh, V.; Chaumond, J.; Delangue, C.; Moi, A.; Cistac, P.; Rault, T.; Louf, R.; Funtowicz, M.; et al. 2020. Transformers: State-of-the-art natural language processing. In Proceedings of the 2020 conference on empirical methods in natural language processin...

  23. [32]

    Xu, L.; Wu, J.; Peng, J.; Fu, J.; and Cai, M. 2022. FCGEC: Fine-Grained Corpus for Chinese Grammatical Error Correction. In Findings of the Association for Computational Linguistics: EMNLP 2022, 1900--1918

  24. [33]

    Xue, L.; Constant, N.; Roberts, A.; Kale, M.; Al-Rfou, R.; Siddhant, A.; Barua, A.; and Raffel, C. 2021. mT5: A Massively Multilingual Pre-trained Text-to-Text Transformer. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational...

  25. [34]

    Yang, A.; Xiao, B.; Wang, B.; Zhang, B.; Bian, C.; Yin, C.; Lv, C.; Pan, D.; Wang, D.; Yan, D.; et al. 2023. Baichuan 2: Open large-scale language models. arXiv preprint arXiv:2309.10305

  26. [35]

    Ye, D.; Tian, B.; Fan, J.; Liu, J.; Zhou, T.; Chen, X.; Li, M.; and Ma, J. 2023. Improving Query Correction Using Pre-train Language Model In Search Engines. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, 2999--3008

  27. [36]

    Yue, M.; Zhao, J.; Zhang, M.; Du, L.; and Yao, Z. 2023. Large language model cascades with mixture of thoughts representations for cost-efficient reasoning. arXiv preprint arXiv:2310.03094

  28. [37]

    H.; and Wang, C

    Zhang, J.; Krishna, R.; Awadallah, A. H.; and Wang, C. 2023. Ecoassistant: Using llm assistant more affordably and accurately. arXiv preprint arXiv:2310.03046

  29. [38]

    Zhang, X.; and Liao, S. 2020. Hypothesis sketching for online kernel selection in continuous kernel space. In Proceedings of the 29th International Joint Conference on Artificial Intelligence, 2498--2504

  30. [39]

    Zhang, X.; Liao, Y.; and Liao, S. 2019. A survey on online kernel selection for online kernel learning. WIREs Data Mining and Knowledge Discovery, 9(2): e1295

  31. [40]

    Zhang, Y.; Li, Z.; Bao, Z.; Li, J.; Zhang, B.; Li, C.; Huang, F.; and Zhang, M. 2022. MuCGEC: a Multi-Reference Multi-Source Evaluation Dataset for Chinese Grammatical Error Correction. In Proceedings of the 2022 Conference of the North American Chapter of the Association for ...

  32. [41]

    Zhang, Z.; Zhang, H.; Chen, K.; Guo, Y.; Hua, J.; Wang, Y.; and Zhou, M. 2021. Mengzi: Towards lightweight yet ingenious pre-trained models for chinese. arXiv preprint arXiv:2110.06696

  33. [42]

    Zhao, Z.; and Wang, H. 2020. Maskgec: Improving neural grammatical error correction via dynamic masking. In Proceedings of the AAAI conference on artificial intelligence, volume 34, 1226--1233

  34. [43]

    Zheng, C.; Yin, F.; Zhou, H.; Meng, F.; Zhou, J.; Chang, K.-W.; Huang, M.; and Peng, N. 2024. On prompt-driven safeguarding for large language models. In Forty-first International Conference on Machine Learning

Pith tools

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