Pith. sign in

REVIEW 2 major objections 6 minor 53 references

Filling Memory Gaps: Enhancing Continual Semantic Parsing via SQL Syntax Variance-Guided LLMs without Real Data Replay

T0 review · 2 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 Continual text-to-SQL can beat replay without any historical data by using SQL syntax gaps to guide an LLM's memory reconstruction.

desk verdict Smart, replay-free continual text-to-SQL with a skeleton-guided LLM memory, but the 'beyond upper bound' FWT claim is a resource-comparison artifact. read the letter →

arxiv 2412.07246 v1 pith:2ISWMLC2 submitted 2024-12-10 cs.CL cs.DB

Ruiheng Liu , Jinyu Zhang , Yanqi Song , Yu Zhang , Bailong Yang This is my paper · ORCID
classification cs.CLcs.DB
keywords continualsemanticparsingtext-to-SQLcatastrophicforgettingSQLskeletonLLMdistillationmemoryreconstructionforwardtransferdata-freelearning
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

Continual semantic parsing models forget old SQL knowledge and overfit when each task in a stream has only a few annotated examples, and the usual remedies replay historical data or assume the test task's identity is known in advance. This paper tries to establish that neither is necessary. It claims that the difference between SQL syntax used by earlier tasks and the syntax used by the current task, represented as a set of domain-erased SQL skeletons, tells an open-source LLM exactly which memories to reconstruct, and that calibrated pseudo-samples plus dual-teacher distillation transfer that knowledge into a small parser. If the claim holds, continual text-to-SQL can run without storing past data or requiring task labels at test time, and it can generalize to unseen tasks better than an oracle that trains on all tasks' data at once.

What carries the argument

The load-bearing objects are the SQL skeleton and the component bias built from it. A SQL skeleton is a domain-erased query shape such as SELECT [COL] FROM [TAB] WHERE [COL] = [VAL] GROUP BY [COL], obtained by masking schema entities; the component bias $\Delta A^{(t)}$ selects skeletons present in the union of previous tasks' cluster-center sets but absent from the current task's set. The bias serves as the instruction to the LLM for what pseudo-sample memory to generate on the current schema, the calibration stage keeps only samples whose SQL executes and whose de-erased form is close to the target skeleton by edit distance, and the dual-teacher distillation propagates the resulting memory into the student while preserving agreement with the previous student's outputs.

What would settle it

Construct a task stream in which a past task's distinguishing SQL pattern (for example, COUNT(DISTINCT ...) with GROUP BY) appears in the training set but never lands close to any of the K cluster centers; if the student then forgets that pattern as badly as sequential fine-tuning, the claim that $\Delta A^{(t)}$ preserves the syntactically complete memory gap is falsified.

Watch

Extended reading notes

Core claim

The paper's central discovery is that SQL syntax variance alone can serve as the memory signal in continual semantic parsing. For each task, LECSP erases database-specific entities from question-SQL pairs, clusters the resulting domain-agnostic skeletons with K-means on CodeT5 representations, and keeps K cluster-center skeletons as the task's component feature set $A^{(t)}$. The component bias $\Delta A^{(t)} = (A^{(1)} \cup A^{(2)} \cup \dots \cup A^{(t-1)}) - A^{(t)}$ consists of skeletons seen in earlier tasks but missing from the current task; prompting an LLM with $\Delta A^{(t)}$ and the current database schema produces pseudo question-SQL pairs that fill exactly those gaps. A calibration stage of iterative execution-checked self-correction and edit-distance sampling filters out hallucinated samples, and a task-aware dual-teacher distillation trains the student on original data, pseudo-memory, and KL agreement with the previous student. On Spider-stream-semi with T5-large, the method reaches 58.6 ACCa-EM without replay, versus 52.2 for SFNET with data replay, and its forward-transfer score 44.3 FWT-EX exceeds the ORACLE bound of 36.0.

Load-bearing premise

The method assumes that a limited set of typical SQL query shapes extracted from each task's training examples captures every piece of SQL knowledge the current task lacks, so prompting an LLM with those shapes can reconstruct the missing memory.

Editorial extensions

If this is right

  • A continual semantic parser can retain historical task accuracy with zero stored real data, which matters when database contents are private or memory is constrained.
  • Knowledge can transfer forward to unseen tasks without task identifiers at test time, because the memory is reconstructed from syntax rather than from per-task modules.
  • The method is robust to cold-start task orders where the first task has few annotations, while prompt-based PET methods degrade sharply in that setting.
  • Pseudo-memory quality, as measured by SQL execution success, is higher than self-training based pseudo-labels, so the calibration stage is essential to the gains.
  • Only compact skeleton sets need to be saved across tasks, so the storage cost of remembering past tasks is small and does not grow with data size.

Reading between the lines

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

  • If syntax-variance-driven memory reconstruction is the active ingredient, the same recipe should transfer to other structured generation streams, such as code generation, SPARQL, or JSON transformation, where task differences can be summarized by skeleton shapes.
  • The forward-transfer scores above ORACLE suggest that the LLM teacher contributes knowledge beyond what the task annotations contain, so the true ceiling for a small parser is not the oracle trained only on task data; explicitly measuring teacher-only knowledge would make this explicit.
  • A testable extension is to replace the cluster-center skeletons with an exhaustive syntax inventory derived from a grammar; if performance does not improve, the K-center approximation is already sufficient, and if it does, the paper's bias signal is currently under-sampling rare syntax.
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

2 major / 6 minor

Summary. The paper proposes LECSP, a continual semantic parsing method that avoids replaying real historical data and does not rely on the ideal task-identity setting used by prompt-based continual learners. The method reconstructs pseudo-memories by (i) removing domain-specific information, (ii) clustering SQL skeletons per task via CodeT5, (iii) computing a syntax-level component bias between the current task and previous tasks, and (iv) prompting an open-source LLM (Mixtral-8x7B) to generate pseudo question-SQL pairs, which are then filtered by execution checks, LLM self-correction, and skeleton-based edit-distance sampling. A dual-teacher distillation framework trains a T5 student using the reconstructed memories, the previous student as a second teacher, and the current task's annotated data. Experiments on Spider-stream-semi and Combined-stream report that LECSP outperforms rehearsal and PET baselines in both warm-start and cold-start protocols, and the paper claims FWT performance beyond the ORACLE upper bound.

Significance. If the empirical results are robust, the paper makes a useful contribution: it demonstrates an effective no-replay, no-ideal-setting recipe for continual text-to-SQL by shifting the burden of memory reconstruction to an offline open-source LLM, and it provides a decomposition (inter-task skeleton completion, intra-task CFG augmentation, calibration, and distillation) with ablations showing that each component matters. The cold-start experiments and the memory-size comparison are valuable additions to the CSP evaluation landscape. However, the headline claim of surpassing an upper bound is overstated, and the absence of variance reporting makes the word 'significantly' unsupported as stated. The core comparison against replay-based baselines remains credible and is the main strength of the paper.

major comments (2)
  1. [Section 2] The claim that ORACLE represents 'the performance upper bound for CSP' is used in Section 2 and in the discussion. As noted above, this is not a theorem; it is an empirical baseline. This should be stated and the 'beyond upper bound' language removed or carefully qualified.
  2. [Experiments, 'Quality and Quantity of Pseudo Samples'; Figure 5(a)] The quality comparison in Figure 5(a) appears to compare LECSP pseudo-samples (evaluated by manual inspection) with SFNET and Mixtral samples (evaluated using their original labels). This is an asymmetric evaluation: manual judgment may use different criteria than the automatic SQL execution used for the baselines. Please state clearly whether the same execution-based metric was applied to all three methods, or provide a human-evaluation protocol applied identically to all methods.
minor comments (6)
  1. [Abstract and throughout] The name is introduced as 'Continual Semantic Parsing' but the abstract first uses 'Continuous Semantic Parsing'; please ensure consistent terminology.
  2. [Methodology, 'Task-Aware Dual-Teacher Distillation learning Framework'] The heading has inconsistent capitalization ('learning' is lowercase); please polish the formatting.
  3. [Experiments, 'Comparison with ORACLE and LLMs'] The phrase 'exceeds most performance upper bound on FWT' is grammatically awkward and should be reworded, especially after the upper-bound concern is addressed.
  4. [Appendix B.1] The cold-start task orders are given as lists of indices [3, 5, 7, 0, 2, 6, 8, 4, 9, 1] and [3, 2, 4, 0, 6, 5, 1], but the original task names are not listed in the appendix; please provide a mapping to the original task domains for reproducibility.
  5. [Table 7] The hyperparameter ablation lists K=60, 70, 90 but not K=80, the value used in the main experiments; including K=80 would make the sensitivity curve complete.
  6. [Figure 1(c)] The caption says 'additional data required refers to extra historical data or unsupervised data', but the figure is not explicitly described in the text; please add a sentence in the Introduction explaining what the axes and symbols represent.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the pipeline is empirically benchmarked rather than defined into its outputs; the ORACLE 'upper bound' claim is a labeling/interpretation issue, not a circular reduction.

full rationale

LECSP's derivation chain is empirical and self-contained in the relevant sense. The component sets A(t) are produced by clustering CodeT5 representations of current-task samples (Eqs. 6-7); the component bias ΔA(t) is a set difference over SQL skeletons (Eq. 1); pseudo-samples are generated by prompting an external LLM with ΔA(t) and the current schema (Eq. 2); calibration filters by execution checks and edit distance; and the student is trained with the losses in Eqs. 3-5. None of these equations sets a predicted quantity equal to its own input by construction. The FWT values (e.g., LECSP 41.4/44.3 vs ORACLE 34.3/36.0 in Table 1) are measured on held-out test sets against independently implemented baselines. The 'beyond the upper limit' claim is not a circular derivation: ORACLE is called an upper bound in the Baselines section, but ORACLE's FWT is not a formal upper bound for a method with an external LLM teacher, and training on more prior-task data can even reduce forward transfer. This is a correctness/interpretation concern rather than a definitional reduction. The only self-citation found is Li et al. (2024) in Related Work, used for the incidental claim that closed-source LLMs hinder continual learning due to their black-box nature; it is not load-bearing for any central premise. No step meets the standard of a quoted equation-to-equation self-reduction.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical or mathematical entities. Its free parameters are standard hyperparameters tuned on the benchmarks, and its key assumptions concern the representational power of SQL skeleton clustering and the reliability of LLM-based generation and verification. The heavy dependence on the memory calibration module is the most fragile assumption, as shown by the ablation.

free parameters (6)
  • K (number of cluster centers) = 80
    Number of SQL skeletons extracted per task via K-means; chosen by hand and ablated in Appendix C.5 (values 60, 70, 90 degrade or slightly change performance).
  • Nske (pseudo-samples per skeleton) = 10
    Number of LLM-generated inter-task memory samples per SQL skeleton; set in Implementation Details without sensitivity analysis.
  • Ncfg (CFG-synthesized instances per annotated sample) = 3
    Number of intra-task memory samples per original example; set in Implementation Details.
  • lambda (distillation weight) = 0.1
    Weight of KL divergence loss in Eq. 5; selected by grid search over {0.03, 0.05, 0.1, 0.2, 0.3} on the benchmark.
  • M (max self-correction iterations) = 3
    Upper limit for LLM revision in Algorithm 1; set in A.3 without ablation.
  • R (top samples retained per skeleton) = 3
    Number of closest-to-skeleton pseudo-samples kept in Algorithm 2; set in A.3 without ablation.
assumptions (4)
  • domain assumption Tasks have disjoint database schemas: for i != j, S(Di) ∩ S(Dj) = ∅.
    Preliminaries; this is the CSP task definition inherited from prior work.
  • domain assumption CodeT5 embeddings of (Qde, Z) provide a meaningful similarity space for K-means clustering of SQL syntax.
    Section A.1; the entire component bias analysis relies on this. No independent validation of the embedding quality is provided.
  • domain assumption The LLM (e.g., Mixtral-8x7B) can generate faithful question/SQL pairs from simplified skeletons and a database schema, and can verify and correct its own outputs reliably.
    Section A.2 and Algorithm 1; the pseudo-sample quality depends entirely on LLM capability and self-verification accuracy.
  • ad hoc to paper Representing each cluster by the SQL skeleton of the sample closest to its center (Eq. 7) is sufficient to capture that cluster's syntactic information.
    Section A.1; this is a heuristic choice with no theoretical justification and no ablation on alternative representatives.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Filling Memory Gaps: Enhancing Continual Semantic Parsing via SQL Syntax Variance-Guided LLMs without Real Data Replay." pith.science (2026). https://pith.science/paper/2ISWMLC2

@misc{pith2026241207246,
  author       = {Pith},
  title        = {Pith review of: Filling Memory Gaps: Enhancing Continual Semantic Parsing via SQL Syntax Variance-Guided LLMs without Real Data Replay},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2ISWMLC2}},
  note         = {Machine review of arXiv:2412.07246}
}
read the original abstract

Continual Semantic Parsing (CSP) aims to train parsers to convert natural language questions into SQL across tasks with limited annotated examples, adapting to the real-world scenario of dynamically updated databases. Previous studies mitigate this challenge by replaying historical data or employing parameter-efficient tuning (PET), but they often violate data privacy or rely on ideal continual learning settings. To address these problems, we propose a new Large Language Model (LLM)-Enhanced Continuous Semantic Parsing method, named LECSP, which alleviates forgetting while encouraging generalization, without requiring real data replay or ideal settings. Specifically, it first analyzes the commonalities and differences between tasks from the SQL syntax perspective to guide LLMs in reconstructing key memories and improving memory accuracy through a calibration strategy. Then, it uses a task-aware dual-teacher distillation framework to promote the accumulation and transfer of knowledge during sequential training. Experimental results on two CSP benchmarks show that our method significantly outperforms existing methods, even those utilizing data replay or ideal settings. Additionally, we achieve generalization performance beyond the upper limits, better adapting to unseen tasks.

Figures

Figures reproduced from arXiv: 2412.07246 by the authors.

Figure 1
Figure 1. (a) Description of the CSP. (b) The average num [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Memory reconstruction on task t via LLMs. First, domain information in the questions and SQL pairs are removed, which means replacing entity-link results uniformly with symbols such as [COL] and [VAL]. Next, K-means clustering is used to obtain the set of SQL skeleton A(t) for task t, and the component bias ∆A(t) is derived by taking the difference with the saved sets from previous tasks A(1) , ..., A(t−1). Note tha… view at source ↗
Figure 3
Figure 3. The task-aware dual-teacher distillation learning [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Results (EX) till the seen tasks based on Spider-stream-semi dataset (T5-large). [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: (a) Accuracy of synthetic data execution across [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: The prompt case of inter-task memory completion based on SQL skeleton. [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: The prompt case of inter-task memory completion based on SQL skeleton (with nested structures). [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: The prompt case of self-correction. Execution result and Error message are returned by the SQL executor. like PEFT and C3, which simplifies the continual learn￾ing scenario. Specifically, they assume the task identity for each test sample is known and load the correspo…
Figure 10
Figure 10. Figure 10: Statistics of pseudo-samples constructed by L (a) Spider-stream [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: Performance (EX) comparison at different difficulty levels on Spider-stream-semi dataset (T5-large). [PITH_FULL_IMAGE:figures/full_fig_p017_11.png]
Figure 12
Figure 12. Figure 12: Results (EX) till the seen tasks on Spider-stream-semi dataset (T5-base). [PITH_FULL_IMAGE:figures/full_fig_p018_12.png]
Figure 13
Figure 13. Figure 13: Results (EX) till the seen tasks on Combined-stream dataset (T5-base). [PITH_FULL_IMAGE:figures/full_fig_p019_13.png]
Figure 14
Figure 14. Figure 14: Results (EX) till the seen tasks on Combined-stream dataset (T5-large). [PITH_FULL_IMAGE:figures/full_fig_p020_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 38 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]

    D.; and Thanisch, P

    Androutsopoulos, I.; Ritchie, G. D.; and Thanisch, P. 1995. Natural language interfaces to databases--an introduction. Natural language engineering, 1(1): 29--81

    work page 1995

  4. [4]

    M.; Firat, O.; Johnson, M.; Lepikhin, D.; Passos, A.; Shakeri, S.; Taropa, E.; Bailey, P.; Chen, Z.; and et al., E

    Anil, R.; Dai, A. M.; Firat, O.; Johnson, M.; Lepikhin, D.; Passos, A.; Shakeri, S.; Taropa, E.; Bailey, P.; Chen, Z.; and et al., E. C. 2023. PaLM 2 Technical Report. arXiv:2305.10403

    arXiv 2023

  5. [5]

    V.; Xu, Y.; and Fung, P

    Bang, Y.; Cahyawijaya, S.; Lee, N.; Dai, W.; Su, D.; Wilie, B.; Lovenia, H.; Ji, Z.; Yu, T.; Chung, W.; Do, Q. V.; Xu, Y.; and Fung, P. 2023. A Multitask, Multilingual, Multimodal Evaluation of ChatGPT on Reasoning, Hallucination, and Interactivity. arXiv:2302.04023

    arXiv 2023

  6. [6]

    Cai, Z.; Li, X.; Hui, B.; Yang, M.; Li, B.; Li, B.; Cao, Z.; Li, W.; Huang, F.; Si, L.; and Li, Y. 2022. STAR : SQL Guided Pre-Training for Context-dependent Text-to- SQL Parsing. In Goldberg, Y.; Kozareva, Z.; and Zhang, Y., eds., Findings of the Association for Computational Linguistics: EMNLP 2022, 1235--1247. Abu Dhabi, United Arab Emirates: Associati...

    work page 2022

  7. [7]

    Chang, S.; and Fosler-Lussier, E. 2023 a . How to Prompt LLM s for Text-to- SQL : A Study in Zero-shot, Single-domain, and Cross-domain Settings. In NeurIPS 2023 Second Table Representation Learning Workshop

    work page 2023

  8. [8]

    Chang, S.; and Fosler-Lussier, E. 2023 b . Selective Demonstrations for Cross-domain Text-to- SQL . In Bouamor, H.; Pino, J.; and Bali, K., eds., Findings of the Association for Computational Linguistics: EMNLP 2023, 14174--14189. Singapore: Association for Computational Linguistics

    work page 2023

Show all 53 references
  1. [9]

    Chen, X.; Lin, M.; Sch \"a rli, N.; and Zhou, D. 2024. Teaching Large Language Models to Self-Debug. In The Twelfth International Conference on Learning Representations

    2024

  2. [10]

    Chen, Y.; Guo, X.; Wu, T.; Qi, G.; Li, Y.; and Dong, Y. 2023 a . Learn from Yesterday: a semi-supervised continual learning method for supervision-limited text-to-SQL task streams. In Proceedings of the Thirty-Seventh AAAI Conference on Artificial Intelligence and Thirty-Fifth...

    2023

  3. [11]

    Chen, Y.; Zhang, S.; Qi, G.; and Guo, X. 2023 b . Parameterizing Context: Unleashing the Power of Parameter-Efficient Fine-Tuning and In-Context Tuning for Continual Table Semantic Parsing. In Oh, A.; Neumann, T.; Globerson, A.; Saenko, K.; Hardt, M.; and Levine, S., eds., Adv...

    2023

  4. [12]

    d'Autume, C. d. M.; Ruder, S.; Kong, L.; and Yogatama, D. 2019. Episodic memory in lifelong language learning. In Proceedings of the 33rd International Conference on Neural Information Processing Systems. Red Hook, NY, USA: Curran Associates Inc

    2019

  5. [13]

    Dou, L.; Gao, Y.; Pan, M.; Wang, D.; Che, W.; Lou, J.-G.; and Zhan, D. 2023. UniSAr: a unified structure-aware autoregressive language model for text-to-SQL semantic parsing. International Journal of Machine Learning and Cybernetics, 14(12): 4361--4376

    2023

  6. [14]

    Guo, C.; Tian, Z.; Tang, J.; Li, S.; Wen, Z.; Wang, K.; and Wang, T. 2024. Retrieval-Augmented GPT-3.5-Based Text-to-SQL Framework with Sample-Aware Prompting and Dynamic Revision Chain. In Luo, B.; Cheng, L.; Wu, Z.-G.; Li, H.; and Li, C., eds., Neural Information Processing,...

    2024

  7. [15]

    Guo, C.; Tian, Z.; Tang, J.; Wang, P.; Wen, Z.; Yang, K.; and Wang, T. 2023. Prompting GPT-3.5 for Text-to-SQL with De-semanticization and Skeleton Retrieval. In PRICAI 2023: Trends in Artificial Intelligence: 20th Pacific Rim International Conference on Artificial Intelligenc...

    2023

  8. [16]

    Han, X.; Dai, Y.; Gao, T.; Lin, Y.; Liu, Z.; Li, P.; Sun, M.; and Zhou, J. 2020. Continual Relation Learning via Episodic Memory Activation and Reconsolidation. In Jurafsky, D.; Chai, J.; Schluter, N.; and Tetreault, J., eds., Proceedings of the 58th Annual Meeting of the Asso...

    2020

  9. [17]

    He, J.; and Yao, X. 2002. From an individual to a population: An analysis of the first hitting time of population-based evolutionary algorithms. IEEE Transactions on Evolutionary Computation, 6(5): 495--511

    2002

  10. [18]

    R.; Campello, R

    Hruschka, E. R.; Campello, R. J.; Freitas, A. A.; et al. 2009. A survey of evolutionary algorithms for clustering. IEEE Transactions on systems, man, and cybernetics, Part C (applications and reviews), 39(2): 133--155

    2009

  11. [19]

    Hsieh, C.-Y.; Li, C.-L.; Yeh, C.-k.; Nakhost, H.; Fujii, Y.; Ratner, A.; Krishna, R.; Lee, C.-Y.; and Pfister, T. 2023. Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes. In Rogers, A.; Boyd-Graber, J.; and Okazaki, N...

    2023

  12. [20]

    Hu, C.; Fu, J.; Du, C.; Luo, S.; Zhao, J.; and Zhao, H. 2023. ChatDB: Augmenting LLMs with Databases as Their Symbolic Memory. arXiv:2306.03901

    2023 arXiv

  13. [21]

    J.; Madotto, A.; and Fung, P

    Ji, Z.; Lee, N.; Frieske, R.; Yu, T.; Su, D.; Xu, Y.; Ishii, E.; Bang, Y. J.; Madotto, A.; and Fung, P. 2023. Survey of Hallucination in Natural Language Generation. ACM Comput. Surv., 55(12)

    2023

  14. [22]

    Q.; Sablayrolles, A.; Roux, A.; Mensch, A.; Savary, B.; Bamford, C.; Chaplot, D

    Jiang, A. Q.; Sablayrolles, A.; Roux, A.; Mensch, A.; Savary, B.; Bamford, C.; Chaplot, D. S.; de las Casas, D.; Hanna, E. B.; Bressand, F.; Lengyel, G.; Bour, G.; Lample, G.; Lavaud, L. R.; Saulnier, L.; Lachaux, M.-A.; Stock, P.; Subramanian, S.; Yang, S.; Antoniak, S.; Scao...

    2024 arXiv

  15. [23]

    Jung, D.; Han, D.; Bang, J.; and Song, H. 2023. Generating Instance-level Prompts for Rehearsal-free Continual Learning. In 2023 IEEE/CVF International Conference on Computer Vision (ICCV), 11813--11823

    2023

  16. [24]

    Lester, B.; Al-Rfou, R.; and Constant, N. 2021. The Power of Scale for Parameter-Efficient Prompt Tuning. In Moens, M.-F.; Huang, X.; Specia, L.; and Yih, S. W.-t., eds., Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, 3045--3059. Online...

    2021

  17. [25]

    Li, H.; Zhang, J.; Li, C.; and Chen, H. 2023 a . RESDSQL: decoupling schema linking and skeleton parsing for text-to-SQL. In Proceedings of the Thirty-Seventh AAAI Conference on Artificial Intelligence and Thirty-Fifth Conference on Innovative Applications of Artificial Intell...

    2023

  18. [26]

    Li, H.; Zhang, J.; Liu, H.; Fan, J.; Zhang, X.; Zhu, J.; Wei, R.; Pan, H.; Li, C.; and Chen, H. 2024. CodeS: Towards Building Open-source Language Models for Text-to-SQL. Proc. ACM Manag. Data, 2(3)

    2024

  19. [27]

    Li, J.; Hui, B.; Cheng, R.; Qin, B.; Ma, C.; Huo, N.; Huang, F.; Du, W.; Si, L.; and Li, Y. 2023 b . Graphix-T5: Mixing Pre-trained Transformers with Graph-Aware Layers for Text-to-SQL Parsing. Proceedings of the AAAI Conference on Artificial Intelligence, 37(11): 13076--13084

    2023

  20. [28]

    Li, J.; Hui, B.; Qu, G.; Yang, J.; Li, B.; Li, B.; Wang, B.; Qin, B.; Geng, R.; Huo, N.; Zhou, X.; Chenhao, M.; Li, G.; Chang, K.; Huang, F.; Cheng, R.; and Li, Y. 2023 c . Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQL...

    2023

  21. [29]

    Li, Z.; Qu, L.; and Haffari, G. 2021. Total Recall: a Customized Continual Learning Method for Neural Semantic Parsers. In Moens, M.-F.; Huang, X.; Specia, L.; and Yih, S. W.-t., eds., Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, 3816...

    2021

  22. [30]

    Lialin, V.; Goel, R.; Simanovsky, A.; Rumshisky, A.; and Shah, R. 2021. Update Frequently, Update Fast: Retraining Semantic Parsing Systems in a Fraction of Time. arXiv:2010.07865

    2021 arXiv

  23. [31]

    Liang, Y.; Meng, F.; Wang, J.; Xu, J.; Chen, Y.; and Zhou, J. 2024. Continual Learning with Semi-supervised Contrastive Distillation for Incremental Neural Machine Translation. In Ku, L.-W.; Martins, A.; and Srikumar, V., eds., Proceedings of the 62nd Annual Meeting of the Ass...

    2024

  24. [32]

    Mao, W.; Wang, R.; Guo, J.; Zeng, J.; Gao, C.; Han, P.; and Liu, C. 2024. Enhancing Text-to- SQL Parsing through Question Rewriting and Execution-Guided Refinement. In Ku, L.-W.; Martins, A.; and Srikumar, V., eds., Findings of the Association for Computational Linguistics ACL...

    2024

  25. [33]

    L.; Almeida, D.; Altenschmidt, J.; Altman, S.; and Shyamal Anadkat, e

    OpenAI; :; Achiam, J.; Adler, S.; Agarwal, S.; Ahmad, L.; Akkaya, I.; Aleman, F. L.; Almeida, D.; Altenschmidt, J.; Altman, S.; and Shyamal Anadkat, e. a. 2024. GPT-4 Technical Report. arXiv:2303.08774

    2024 arXiv

  26. [34]

    Pourreza, M.; and Rafiei, D. 2023. DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-Correction. In Oh, A.; Neumann, T.; Globerson, A.; Saenko, K.; Hardt, M.; and Levine, S., eds., Advances in Neural Information Processing Systems, volume 36, 36339--36348. Curra...

    2023

  27. [35]

    Qin, C.; and Joty, S. 2022. LFPT 5: A Unified Framework for Lifelong Few-shot Language Learning Based on Prompt Tuning of T5. In International Conference on Learning Representations

    2022

  28. [36]

    Razdaibiedina, A.; Mao, Y.; Hou, R.; Khabsa, M.; Lewis, M.; and Almahairi, A. 2023. Progressive Prompts: Continual Learning for Language Models. In The Eleventh International Conference on Learning Representations

    2023

  29. [37]

    Shazeer, N.; and Stern, M. 2018. Adafactor: Adaptive Learning Rates with Sublinear Memory Cost. In Dy, J.; and Krause, A., eds., Proceedings of the 35th International Conference on Machine Learning, volume 80 of Proceedings of Machine Learning Research, 4596--4604. PMLR

    2018

  30. [38]

    Wang, D.; Dou, L.; Zhang, X.; Zhu, Q.; and Che, W. 2024 a . Improving Demonstration Diversity by Human-Free Fusing for Text-to-SQL. arXiv:2402.10663

    2024 arXiv

  31. [39]

    Wang, L.; Zhang, X.; Li, Q.; Zhang, M.; Su, H.; Zhu, J.; and Zhong, Y. 2023 a . Incorporating neuro-inspired adaptability for continual learning in artificial intelligence. Nature Machine Intelligence, 5(12): 1356--1368

    2023

  32. [40]

    Wang, L.; Zhang, X.; Su, H.; and Zhu, J. 2024 b . A Comprehensive Survey of Continual Learning: Theory, Method and Application. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1--20

    2024

  33. [41]

    Wang, L.; Zhang, X.; Yang, K.; Yu, L.; Li, C.; HONG, L.; Zhang, S.; Li, Z.; Zhong, Y.; and Zhu, J. 2022. Memory Replay with Data Compression for Continual Learning. In International Conference on Learning Representations

    2022

  34. [42]

    Wang, X.; Zhang, S.; Qing, Z.; Shao, Y.; Gao, C.; and Sang, N. 2021 a . Self-Supervised Learning for Semi-Supervised Temporal Action Proposal. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 1905--1914

    2021

  35. [43]

    Wang, Y.; Wang, W.; Joty, S.; and Hoi, S. C. 2021 b . C ode T 5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation. In Moens, M.-F.; Huang, X.; Specia, L.; and Yih, S. W.-t., eds., Proceedings of the 2021 Conference on Empirical ...

    2021

  36. [44]

    Wang, Z.; Liu, Y.; Ji, T.; Wang, X.; Wu, Y.; Jiang, C.; Chao, Y.; Han, Z.; Wang, L.; Shao, X.; and Zeng, W. 2023 b . Rehearsal-free Continual Language Learning via Efficient Parameter Isolation. In Rogers, A.; Boyd-Graber, J.; and Okazaki, N., eds., Proceedings of the 61st Ann...

    2023

  37. [45]

    Xiao, R.; Dong, Y.; Zhao, J.; Wu, R.; Lin, M.; Chen, G.; and Wang, H. 2023. F ree AL : Towards Human-Free Active Learning in the Era of Large Language Models. In Bouamor, H.; Pino, J.; and Bali, K., eds., Proceedings of the 2023 Conference on Empirical Methods in Natural Langu...

    2023

  38. [46]

    Xue, S.; Jiang, C.; Shi, W.; Cheng, F.; Chen, K.; Yang, H.; Zhang, Z.; He, J.; Zhang, H.; Wei, G.; Zhao, W.; Zhou, F.; Qi, D.; Yi, H.; Liu, S.; and Chen, F. 2024. DB-GPT: Empowering Database Interactions with Private Large Language Models. arXiv:2312.17449

    2024 arXiv

  39. [47]

    K.; Bansal, M.; and Xiang, B

    Yadav, P.; Sun, Q.; Ding, H.; Li, X.; Zhang, D.; Tan, M.; Bhatia, P.; Ma, X.; Nallapati, R.; Ramanathan, M. K.; Bansal, M.; and Xiang, B. 2023. Exploring Continual Learning for Code Generation Models. In Rogers, A.; Boyd-Graber, J.; and Okazaki, N., eds., Proceedings of the 61...

    2023

  40. [48]

    Yogatama, D.; de Masson d'Autume, C.; Connor, J.; Kocisky, T.; Chrzanowski, M.; Kong, L.; Lazaridou, A.; Ling, W.; Yu, L.; Dyer, C.; and Blunsom, P. 2019. Learning and Evaluating General Linguistic Intelligence. arXiv:1901.11373

    2019 arXiv

  41. [49]

    V.; bailin wang; Tan, Y

    Yu, T.; Wu, C.-S.; Lin, X. V.; bailin wang; Tan, Y. C.; Yang, X.; Radev, D.; richard socher; and Xiong, C. 2021. GraPPa: Grammar-Augmented Pre-Training for Table Semantic Parsing. In International Conference on Learning Representations

    2021

  42. [50]

    Yu, T.; Zhang, R.; Yang, K.; Yasunaga, M.; Wang, D.; Li, Z.; Ma, J.; Li, I.; Yao, Q.; Roman, S.; Zhang, Z.; and Radev, D. 2018. S pider: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to- SQL Task. In Riloff, E.; Chiang, D.; Hockenma...

    2018

  43. [51]

    Zhao, W.; Wang, S.; Hu, Y.; Zhao, Y.; Qin, B.; Zhang, X.; Yang, Q.; Xu, D.; and Che, W. 2024. SAPT: A Shared Attention Framework for Parameter-Efficient Continual Learning of Large Language Models. arXiv:2401.08295

    2024 arXiv

  44. [52]

    Zhong, R.; Yu, T.; and Klein, D. 2020. Semantic Evaluation for Text-to- SQL with Distilled Test Suites. In Webber, B.; Cohn, T.; He, Y.; and Liu, Y., eds., Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), 396--411. Online: Associa...

    2020

  45. [53]

    Zhong, V.; Xiong, C.; and Socher, R. 2018. Seq2 SQL : Generating Structured Queries From Natural Language Using Reinforcement Learning

    2018

Pith tools

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