Pith. sign in

REVIEW 4 major objections 6 minor 33 references

By treating schema linking as a tunable precision-recall problem and using model internal confidence to prune beam-search candidates, CHS-SQL lifts sub-7B language models to state-of-the-art text-to-SQL accuracy on Spider and BIRD.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-02 10:55 UTC pith:IHIMOXRX

load-bearing objection A plausible DTS-SQL extension, but the confidence signal is uncalibrated and the main gains may be dev-set threshold tuning. the 4 major comments →

arxiv 2607.22624 v1 pith:IHIMOXRX submitted 2026-06-17 cs.AI

CHS-SQL: A Text-to-SQL approach based on Confidence-Guided Heuristic Search Schema Linking process

Minghao Yang , Yanjun Xu This is my paper
classification cs.AI
keywords text-to-SQLschema linkingprecision-recall trade-offbeam searchtrace confidencesmall language modelsLoRA fine-tuningexecution accuracy
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

CHS-SQL claims that the precision-recall trade-off in schema linking is the decisive factor for text-to-SQL accuracy in small language models, and that it can be controlled explicitly. The paper introduces a two-stage framework: a schema linking module that uses beam search to increase recall and a trace-confidence filter to increase precision, followed by a SQL generation module that applies the same confidence-guided selection and an ensemble of models. Evaluated on Spider and BIRD with sub-7B models, it reports state-of-the-art execution accuracy, improving over DTS-SQL by 2 to 9 points. The core message is that balancing, not maximising, schema recall and precision is what drives downstream SQL correctness.

Core claim

The central claim is that high recall of ground-truth tables and columns is a necessary precondition for accurate SQL generation, but that indiscriminate recall introduces noise that hurts performance; the optimal operating point is a controlled trade-off. CHS-SQL implements this with beam search over schema-linking traces (which raises recall) and a trace-confidence filter (which raises precision by discarding high-variance, low-certainty traces). The filtered high-recall table set is then expanded to include all columns of the chosen tables, further increasing column recall. During SQL generation, the same confidence heuristic selects the most stable query from beam-searched candidates pro

What carries the argument

Trace Confidence (TC): the standard deviation of per-token negative log-probabilities over a generated trace, interpreted as a model-certainty score. Low TC indicates high certainty and is used both to discard low-quality schema-linking traces (with a threshold θ) and to pick the final SQL among ensemble candidates (lowest TC). Beam width k and θ are the two hyperparameters that quantitatively set the precision-recall operating point; the paper also uses the 'All Columns Chosen' expansion to recover column recall once tables are selected.

Load-bearing premise

Trace Confidence — the spread of token-level probabilities — is a reliable signal of whether a generated schema link or SQL query is actually correct; if it is not calibrated for a given model or dataset, the filtering mechanism that provides the precision gain breaks down.

What would settle it

A calibration check: on a held-out set of schema-linking predictions, compute TC for correct vs incorrect traces. If the distributions overlap substantially or higher TC does not correspond to higher error rates, then thresholding on TC cannot be what drives the reported precision improvements, and the gains would be attributable to beam-search recall alone. Repeating the comparison on BIRD with a model whose probabilities are known to be miscalibrated (e.g., a quantized model under distribution shift) would test the assumption directly.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Schema linking should be treated as an explicit precision-recall trade-off problem, not just a filtering step, and evaluated with both table and column metrics.
  • Sub-7B models fine-tuned with LoRA can reach execution accuracies on Spider and BIRD that rival much larger models, making private, single-GPU text-to-SQL deployment feasible.
  • The same trace-confidence criterion can be reused at both pipeline stages — schema linking and SQL generation — suggesting a general method for uncertainty-aware generation with SLMs.
  • Adjusting beam width and confidence threshold provides a practical two-knob control for adapting text-to-SQL pipelines to different accuracy requirements or noisy schemas.

Where Pith is reading between the lines

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

  • If trace confidence is a valid correctness signal, it generalises beyond SQL: any structured generation task with beam search could use per-trace variance to prune candidates, e.g., code generation or information extraction.
  • The paper's reported optimum favours recall over precision (table F1 0.793 vs 0.954 for baselines) — a reader might infer that for schema linking, missing a relevant table is costlier than including an irrelevant one, which could inform dataset design and evaluation weighting.
  • The limitation section implies an open problem: achieve 100% recall while raising precision. Combining CHS-SQL with hard constraints (e.g., string match on values/dates, as the 'TEXT/NUM/DATE match' variant does) is a concrete direction to push precision higher without sacrificing recall.
  • The ensemble result (single-model 0.736 vs ensemble 0.773 EX) suggests that diversity across models contributes beyond the schema linking gain; ablating model diversity separately would quantify that residual.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes CHS-SQL, a two-stage text-to-SQL pipeline for small language models. In the schema-linking stage, a LoRA-fine-tuned SLM generates candidate table/column sequences via beam search; the candidates are then filtered using a proposed 'Trace Confidence' metric (the standard deviation of per-token negative log-probabilities) with a threshold θ, and the surviving tables' columns are all retained. In the SQL generation stage, multiple fine-tuned SLMs generate candidate SQL queries via beam search, and the query with the lowest Trace Confidence is selected as the final answer. The authors report gains over DTS-SQL on Spider Dev (e.g., EX 0.766 vs 0.674 for Qwen3-4B), Spider Test, and BIRD Dev, and claim that CHS-SQL achieves an optimal precision–recall trade-off in schema linking and state-of-the-art results for SLMs.

Significance. If the reported results are robust, the paper would make a useful empirical contribution by showing that beam-search-based schema linking with confidence filtering can improve text-to-SQL accuracy on small models. The manuscript also provides code, detailed LoRA hyperparameters, and a clear two-stage architecture, which supports reproducibility. However, the central claims rest on two load-bearing assumptions: (1) that the Trace Confidence metric actually measures model certainty, and (2) that the configuration selected on the dev sets represents an 'optimal' precision–recall trade-off. The current evidence for both is weak, and Table 2 appears to contradict the precision–recall claim. The theoretical flaw in the confidence metric is not a mere presentation issue; it affects the core mechanism, so the significance of the contribution depends on whether the authors can provide calibration or reformulate the selection criterion.

major comments (4)
  1. [§4.2.3, Eqs. (5)–(7); §4.3, Eq. (19)] Trace Confidence is defined as the standard deviation of token-level confidence values T_j. A trace in which the model is uniformly uncertain at every position will have nearly constant T_j, hence TC≈0, and would be treated as maximally confident by Eq. (19). Conversely, a trace with mixed confident and uncertain tokens gets a high TC. Thus TC measures variability, not certainty, and the paper's statement that 'low trace confidence corresponds to greater model certainty' is not supported by the definition. Since this metric is used both to filter schema-linking traces (Eq. 10) and to select the final SQL (Eq. 19), the entire confidence mechanism is built on an unvalidated assumption. No calibration analysis is provided linking low TC to higher accuracy. Please either correct the definition (e.g., use T_avg or entropy) or provide empirical evidence that low TC correlates with correctness
  2. [Table 2, §5.3] The central claim of an 'optimal precision–recall trade-off' is not supported by the reported data. In Table 2, CHS-SQL has the lowest table precision (T-P=0.661) and by far the lowest column precision (C-P=0.192) among the compared methods, and its table/column F1 (0.793/0.322) are also below DTS-SQL (0.954/0.415). The method essentially achieves high recall at the expense of precision, with a small EX gain (0.736 vs 0.731) and a lower EM (0.697 vs 0.720) relative to DTS-SQL on the same CodeLlama-7B backbone. Calling this configuration 'optimal' is not justified by the table. The later statement in §5.3 that 'the combination of Beam Search and Trace Confidence filtration helps CHS-SQL find the perfect precision-recall balance' is directly contradicted by the numbers. Please reframe the claim to 'high-recall, lower-precision configuration' and provide evidence that this trade-off is inde
  3. [Eq. (12), §5.1.3, Tables 1–3] The hyperparameters (k*, θ*) are selected to maximize the sum of table/column precision and recall (Eq. 12), and the final EX/EM are then reported on the same Spider Dev and BIRD Dev sets used for this selection. This constitutes test-set tuning for the main reported numbers. The 'gains' over DTS-SQL may therefore partly reflect overfitting to the dev set rather than a generalizable improvement. The paper reports only one held-out test split (Spider Test, CodeLlama-7B), so the claim of SOTA is not established. Please report results on held-out test sets for all models and datasets, or use a validation split separate from the final evaluation, and show sensitivity of EX/EM to θ and k.
  4. [§5.1.2 and Abstract] The paper repeatedly states that the experiments use 'SLMs with fewer than 1 billion parameters' and 'sub-billion-parameter LLMs', but the actual models are CodeLlama-7B, Qwen3-4B, and DeepSeek-Coder-6.7B-Instruct—all above 1B parameters (Qwen3-4B is 4B). This is inconsistent with the abstract's emphasis on small models and with the claim that the method works for 'SLMs'. Either correct the model sizes or revise the claim to 'models in the 4B–7B range'. This is not merely a typo; it changes the scope of the contribution.
minor comments (6)
  1. [§4.2.4] The 'Max-Gap Midpoint Sampling' method for choosing θ is mentioned but never described. Please provide the algorithm or a reference, as it is essential for reproducibility.
  2. [§5.3, Figure 2] The figure caption uses abbreviations M1–M7, but the legend in the text (M1: FT-SLM, etc.) does not exactly match the order listed in the caption. Please align them.
  3. [§5.4] In the ablation text, 'standard CHS-SQL (the voting version)' is contrasted with 'single model version CHS-SQL', but the table shows CHS-SQL (only with Codellama-7b) as the single-model version. The wording is confusing; clarify which version is used in each ablation row.
  4. [§5.2] The text in the final paragraph of §5.2 says 'our scores reached 75.4% (EX) and 75.2% (EM)', but Table 1 lists the ensemble as 0.773 EX and 0.739 EM. These numbers do not match; please correct.
  5. [§3.2, Eq. (2)] The notation P_M∫ uses an integral sign that appears to be a typo (likely P_M_s). The same issue appears in Eq. (3) with L_M∗. Please use consistent subscript formatting.
  6. [References] Reference [17] is cited for Beam Search but points to 'Efficient Estimation of Word Representations in Vector Space' (Mikolov et al.). Please cite a proper beam search reference.

Circularity Check

1 steps flagged

No equation-level circularity, but the schema-linking 'optimal precision-recall trade-off' is self-confirming because the threshold is tuned on the same dev set whose downstream SQL accuracy is then reported.

specific steps
  1. fitted input called prediction [Sec. 4.2.4, Eq. (12); Sec. 5.3, Table 2 and discussion]
    "The beam width k and confidence threshold θ are chosen to maximize table/column recall and precision: (k∗, θ∗) = arg max k∈K, θ∈Θ (Rtab +Ptab +Rcol +Pcol). ... Standard CHS-SQL has the highest recall for both tables and columns compared to the other three strategies... Despite this, it achieved the highest EX and EM scores."

    The hyperparameters θ and k are selected on the Spider Dev set by directly maximizing the schema-linking metrics (table/column recall + precision). Table 2 then reports that this same procedure yields the highest recall and the 'perfect precision-recall balance' on that same set, and Table 1 reports the downstream SQL EX/EM on the same Spider Dev set. Thus the high recall and optimal trade-off are by construction on Spider Dev, not an out-of-sample finding, and the associated SQL improvement is partly self-confirming because no held-out tuning evaluation of θ/k is reported. The claim retains some independent support from the Spider Test and BIRD results, but the central 'optimal trade-off' conclusion is not a true prediction.

full rationale

The paper's derivation chain is mostly empirical rather than equation-level: SQL accuracies are measured on standard benchmarks, not derived from the confidence formulas. There is no self-citation chain, and the Trace Confidence idea is attributed to external prior work ([6], [15]). The main circular aspect is the dev-set self-confirmation: Eq. (12) chooses k and θ to maximize schema-linking recall+precision on Spider Dev, and the same dev set is then used both to demonstrate the 'optimal precision-recall trade-off' in Table 2 and to report the headline SQL gains in Table 1. This does not make the SQL accuracy itself a tautology, but it weakens the causal claim that the trade-off was discovered rather than tuned. The paper also provides some independent support: Spider Test EX improves 3.4% over DTS-SQL, and the ablation study isolates components. However, the Trace Confidence validity assumption (low std-dev equals high correctness) is never independently calibrated, so the score is moderate. The limitation section honestly notes that 100%-recall baselines come close to CHS-SQL, further tempering the claim. Overall, this is partial circularity from dev-set hyperparameter fitting, not full derivation-by-construction.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 0 invented entities

No new physical or conceptual entities are introduced; 'Trace Confidence' is a composite metric over existing model probabilities, not an independent entity. The paper's contribution rests mostly on four fitted hyperparameters and a domain assumption that confidence (as defined) tracks correctness.

free parameters (4)
  • Beam width k (schema linking and SQL generation) = not reported
    Eq. (4)/(8)/(12): k controls the number of candidate sequences; chosen to maximize precision+recall on dev; no value given in paper.
  • Confidence threshold θ = not reported
    Eq. (10): candidates with TC <= θ are kept; chosen by 'Max-Gap Midpoint Sampling' on dev; central to precision-recall trade-off but not specified.
  • Top-k token count used in token confidence = not reported
    Eq. (5): token confidence averages over top k token probabilities; the paper never sets this k.
  • LoRA hyperparameters = rank=64, alpha=32, lr=5e-5, batch=2 (CodeLlama-7b)
    Sec. 5.1.2: shared across models; fine-tuning setup affects all downstream numbers.
axioms (5)
  • domain assumption Both precision and recall of Schema Linking directly affect final SQL accuracy.
    Abstract and Sec. 1; the paper's motivation is inferred from their own experiments, not an independently established benchmark.
  • domain assumption Low trace confidence corresponds to greater model certainty and more accurate predictions.
    Sec. 4.2.3; this is the load-bearing premise for the filtering mechanism; only supported by citing [15] and by their own results.
  • domain assumption Beam search explores a hypothesis space broad enough to raise recall of relevant schema elements.
    Sec. 4.2.2; standard heuristic, not proven for schema-linking sequences.
  • domain assumption Fine-tuned SLMs with LoRA can perform both schema linking and SQL generation.
    Sec. 4.2.1/4.3; relies on prior PEFT literature and is not independently validated.
  • ad hoc to paper Max-Gap Midpoint Sampling yields a valid threshold θ.
    Sec. 4.2.4; method is introduced without definition or citation, so the threshold selection is effectively a black box.

pith-pipeline@v1.3.0-alltime-deepseek · 15098 in / 14068 out tokens · 123773 ms · 2026-08-02T10:55:14.863497+00:00 · methodology

0 comments
read the original abstract

Recently, there have been several works in the Text-to-SQL domain that utilize Small Language Models (SLMs) for training. These approaches achieve performance close to that of large models in generating SQL, using only the computational power of a single NVIDIA RTX 4090 GPU, while also ensuring data security. Most existing methods filter out redundant tables and columns during Schema Linking to improve Text-to-SQL accuracy. However, they do not consider the precision-recall trade-off when selecting the candidate schema subset. Our research found that both the precision and recall of Schema Linking directly affect the final SQL accuracy. Therefore, we propose a novel framework for efficiently fine-tuning SLMs on Text-to-SQL tasks, CHS-SQL, that not only balances precision and recall but also improves overall performance on Text-to-SQL tasks. Its main innovation lies in the Schema Linking phase, where a heuristic search combined with model internal confidence is employed to achieve an optimal precision-recall trade-off. This elaborated mechanism maximizes the precision of relevant schema candidates for the generated SQL queries while suppressing irrelevant noise. The same strategy is further applied during SQL generation to refine candidate queries while helping the SLM to avoid trapping in a local optimum. Our method achieves state-of-the-art (SOTA) results on Text-to-SQL tasks via SLMs.

Figures

Figures reproduced from arXiv: 2607.22624 by Minghao Yang, Yanjun Xu.

Figure 1
Figure 1. Figure 1: CHS-SQL Structure. During the Schema Linking phase, CHS-SQL [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The relationship between Schema Linking performance (Ta [PITH_FULL_IMAGE:figures/full_fig_p018_2.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

33 extracted references · 11 linked inside Pith

  1. [1]

    Sadga: Structure- aware dual graph aggregation network for text-to-sql.Advances in Neural Information Processing Systems, 34:7664–7676, 2021

    Ruichu Cai, Jinjie Yuan, Boyan Xu, and Zhifeng Hao. Sadga: Structure- aware dual graph aggregation network for text-to-sql.Advances in Neural Information Processing Systems, 34:7664–7676, 2021

    2021

  2. [2]

    Lgesql: Line graph enhanced text-to-sql model with mixed local and non- local relations

    Ruisheng Cao, Lu Chen, Zhi Chen, Yanbin Zhao, Su Zhu, and Kai Yu. Lgesql: Line graph enhanced text-to-sql model with mixed local and non- local relations. InProceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 2541– ...

    2021

  3. [3]

    Ryansql: Recursively applying sketch-based slot fillings for com- plex text-to-sql in cross-domain databases.Computational Linguistics, 47(2):309–332, 2021

    DongHyun Choi, Myeong Cheol Shin, EungGyun Kim, and Dong Ryeol Shin. Ryansql: Recursively applying sketch-based slot fillings for com- plex text-to-sql in cross-domain databases.Computational Linguistics, 47(2):309–332, 2021

    2021

  4. [4]

    Bert: Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. InProceedings of the 2019 conference of the North American chapter of 22 the association for computational linguistics: human language technologies, volume 1 (long and short papers), pages 4171–4186, 2019

    2019

  5. [5]

    C3: Zero-shot text-to-sql with chatgpt.arXiv preprint arXiv:2307.07306, 2023

    Xuemei Dong, Chao Zhang, Yuhang Ge, Yuren Mao, Yunjun Gao, Jinshu Lin, Dongfang Lou, et al. C3: Zero-shot text-to-sql with chatgpt.arXiv preprint arXiv:2307.07306, 2023

    Pith/arXiv arXiv 2023

  6. [6]

    Deep think with confidence.arXiv preprint arXiv:2508.15260, 2025

    Yichao Fu, Xuewei Wang, Yuandong Tian, and Jiawei Zhao. Deep think with confidence.arXiv preprint arXiv:2508.15260, 2025

    Pith/arXiv arXiv 2025

  7. [7]

    Natural sql: Making sql easier to infer from natural language specifications

    Yujian Gan, Xinyun Chen, Jinxia Xie, Matthew Purver, John R Woodward, John Drake, and Qiaofu Zhang. Natural sql: Making sql easier to infer from natural language specifications. InFindings of the Association for Computational Linguistics: EMNLP 2021, pages 2030–2042, 2021

    2021

  8. [8]

    Text-to-sql empowered by large language models: A benchmark evaluation.arXiv preprint arXiv:2308.15363, 2023

    Dawei Gao, Haibin Wang, Yaliang Li, Xiuyu Sun, Yichen Qian, Bolin Ding, and Jingren Zhou. Text-to-sql empowered by large language models: A benchmark evaluation.arXiv preprint arXiv:2308.15363, 2023

    Pith/arXiv arXiv 2023

  9. [9]

    Long short-term memory.Supervised sequence labelling with recurrent neural networks, pages 37–45, 2012

    Alex Graves. Long short-term memory.Supervised sequence labelling with recurrent neural networks, pages 37–45, 2012

    2012

  10. [10]

    Towards complex text-to-sql in cross-domain database with intermediate representation

    Jiaqi Guo, Zecheng Zhan, Yan Gao, Yan Xiao, Jian-Guang Lou, Ting Liu, and Dongmei Zhang. Towards complex text-to-sql in cross-domain database with intermediate representation. InProceedings of the 57th annual meeting of the association for computational linguistics, pages 4524–4535, 2019

    2019

  11. [11]

    Lora: Low-rank adaptation of large language models.Iclr, 1(2):3, 2022

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Liang Wang, Weizhu Chen, et al. Lora: Low-rank adaptation of large language models.Iclr, 1(2):3, 2022

    2022

  12. [12]

    Relation aware semi-autoregressive semantic parsing for nl2sql.arXiv preprint arXiv:2108.00804, 2021

    Junyang Huang, Yongbo Wang, Yongliang Wang, Yang Dong, and Yanghua Xiao. Relation aware semi-autoregressive semantic parsing for nl2sql.arXiv preprint arXiv:2108.00804, 2021

    Pith/arXiv arXiv 2021

  13. [13]

    Improving text-to-sql with schema dependency learning

    Binyuan Hui, Xiang Shi, Ruiying Geng, Binhua Li, Yongbin Li, Jian Sun, and Xiaodan Zhu. Improving text-to-sql with schema dependency learning. arXiv preprint arXiv:2103.04399, 2021

    Pith/arXiv arXiv 2021

  14. [14]

    A comprehensive exploration on wikisql with table-aware word contextual- ization.arXiv preprint arXiv:1902.01069, 2019

    Wonseok Hwang, Jinyeong Yim, Seunghyun Park, and Minjoon Seo. A comprehensive exploration on wikisql with table-aware word contextual- ization.arXiv preprint arXiv:1902.01069, 2019

    Pith/arXiv arXiv 1902

  15. [15]

    Scalable best-of-n selection for large language models via self-certainty.Advances in neural information processing systems, 38:19720–19745, 2026

    Zhewei Kang, Xuandong Zhao, and Dawn Song. Scalable best-of-n selection for large language models via self-certainty.Advances in neural information processing systems, 38:19720–19745, 2026

    2026

  16. [16]

    Resdsql: Decoupling schema linking and skeleton parsing for text-to-sql

    Haoyang Li, Jing Zhang, Cuiping Li, and Hong Chen. Resdsql: Decoupling schema linking and skeleton parsing for text-to-sql. InProceedings of the AAAI Conference on Artificial Intelligence, volume 37, pages 13067–13075, 2023. 23

    2023

  17. [17]

    Effi- cient estimation of word representations in vector space.arXiv preprint arXiv:1301.3781, 2013

    Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. Effi- cient estimation of word representations in vector space.arXiv preprint arXiv:1301.3781, 2013

    Pith/arXiv arXiv 2013

  18. [18]

    Enhancing text-to-sql capa- bilities of large language models: A study on prompt design strategies

    Linyong Nan, Yilun Zhao, Weijin Zou, Narutatsu Ri, Jaesung Tae, Ellen Zhang, Arman Cohan, and Dragomir Radev. Enhancing text-to-sql capa- bilities of large language models: A study on prompt design strategies. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 14935–14956, 2023

    2023

  19. [19]

    Din-sql: Decomposed in- context learning of text-to-sql with self-correction.Advances in neural in- formation processing systems, 36:36339–36348, 2023

    Mohammadreza Pourreza and Davood Rafiei. Din-sql: Decomposed in- context learning of text-to-sql with self-correction.Advances in neural in- formation processing systems, 36:36339–36348, 2023

    2023

  20. [20]

    Dts-sql: Decomposed text- to-sql with small large language models

    Mohammadreza Pourreza and Davood Rafiei. Dts-sql: Decomposed text- to-sql with small large language models. InFindings of the Association for Computational Linguistics: EMNLP 2024, pages 8212–8220, 2024

    2024

  21. [21]

    Rasat: Integrat- ing relational structures into pretrained seq2seq model for text-to-sql

    Jiexing Qi, Jingyao Tang, Ziwei He, Xiangpeng Wan, Yu Cheng, Chenghu Zhou, Xinbing Wang, Quanshi Zhang, and Zhouhan Lin. Rasat: Integrat- ing relational structures into pretrained seq2seq model for text-to-sql. In Proceedings of the 2022 conference on empirical methods in natural lan- guage processing, pages 3215–3229, 2022

    2022

  22. [22]

    A survey on text- to-sql parsing: Concepts, methods, and future directions.arXiv preprint arXiv:2208.13629, 2022

    Bowen Qin, Binyuan Hui, Lihan Wang, Min Yang, Jinyang Li, Binhua Li, Ruiying Geng, Rongyu Cao, Jian Sun, Luo Si, et al. A survey on text- to-sql parsing: Concepts, methods, and future directions.arXiv preprint arXiv:2208.13629, 2022

    Pith/arXiv arXiv 2022

  23. [23]

    Picard: Pars- ing incrementally for constrained auto-regressive decoding from language models

    Torsten Scholak, Nathan Schucher, and Dzmitry Bahdanau. Picard: Pars- ing incrementally for constrained auto-regressive decoding from language models. InProceedings of the 2021 conference on empirical methods in natural language processing, pages 9895–9901, 2021

    2021

  24. [24]

    Sequence to sequence learn- ing with neural networks.Advances in neural information processing sys- tems, 27, 2014

    Ilya Sutskever, Oriol Vinyals, and Quoc V Le. Sequence to sequence learn- ing with neural networks.Advances in neural information processing sys- tems, 27, 2014

    2014

  25. [25]

    An investigation between schema linking and text-to-sql performance

    Yasufumi Taniguchi, Hiroki Nakayama, Kubo Takahiro, and Jun Suzuki. An investigation between schema linking and text-to-sql performance. arXiv preprint arXiv:2102.01847, 2021

    Pith/arXiv arXiv 2021

  26. [26]

    Efficient large language models: A survey.arXiv preprint arXiv:2312.03863, 2023

    Zhongwei Wan, Xin Wang, Che Liu, Samiul Alam, Yu Zheng, Jiachen Liu, Zhongnan Qu, Shen Yan, Yi Zhu, Quanlu Zhang, et al. Efficient large language models: A survey.arXiv preprint arXiv:2312.03863, 2023

    Pith/arXiv arXiv 2023

  27. [27]

    Rat-sql: Relation-aware schema encoding and link- ing for text-to-sql parsers

    Bailin Wang, Richard Shin, Xiaodong Liu, Oleksandr Polozov, and Matthew Richardson. Rat-sql: Relation-aware schema encoding and link- ing for text-to-sql parsers. InProceedings of the 58th annual meeting of the association for computational linguistics, pages 7567–7578, 2020. 24

    2020

  28. [28]

    Mac-sql: A multi-agent collaborative framework for text-to-sql

    Bing Wang, Changyu Ren, Jian Yang, Xinnian Liang, Jiaqi Bai, Linzheng Chai, Zhao Yan, Qian-Wen Zhang, Di Yin, Xing Sun, et al. Mac-sql: A multi-agent collaborative framework for text-to-sql. InProceedings of the 31st International Conference on Computational Linguistics, pages 540– 557, 2025

    2025

  29. [29]

    Fali Wang, Zhiwei Zhang, Xianren Zhang, Zongyu Wu, Tzuhao Mo, Qi- uhao Lu, Wanjing Wang, Rui Li, Junjie Xu, Xianfeng Tang, et al. A comprehensive survey of small language models in the era of large lan- guage models: Techniques, enhancements, applications, collaboration with llms, and trustworthiness.ACM Transactions on Intelligent Systems and Technology,...

    2025

  30. [30]

    Chain-of-thought prompting elic- its reasoning in large language models.Advances in neural information processing systems, 35:24824–24837, 2022

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. Chain-of-thought prompting elic- its reasoning in large language models.Advances in neural information processing systems, 35:24824–24837, 2022

    2022

  31. [31]

    Sqlnet: Generating struc- tured queries from natural language without reinforcement learning.arXiv preprint arXiv:1711.04436, 2017

    Xiaojun Xu, Chang Liu, and Dawn Song. Sqlnet: Generating struc- tured queries from natural language without reinforcement learning.arXiv preprint arXiv:1711.04436, 2017

    Pith/arXiv arXiv 2017

  32. [32]

    Syntaxsqlnet: Syntax tree networks for complex and cross-domain text-to-sql task

    Tao Yu, Michihiro Yasunaga, Kai Yang, Rui Zhang, Dongxu Wang, Zifan Li, and Dragomir Radev. Syntaxsqlnet: Syntax tree networks for complex and cross-domain text-to-sql task. InProceedings of the 2018 conference on empirical methods in natural language processing, pages 1653–1663, 2018

    2018

  33. [33]

    Benchmarking the text-to-sql capability of large language models: A comprehensive evalua- tion.arXiv preprint arXiv:2403.02951, 2024

    Bin Zhang, Yuxiao Ye, Guoqing Du, Xiaoru Hu, Zhishuai Li, Sun Yang, Chi Harold Liu, Rui Zhao, Ziyue Li, and Hangyu Mao. Benchmarking the text-to-sql capability of large language models: A comprehensive evalua- tion.arXiv preprint arXiv:2403.02951, 2024. 25

    arXiv 2024