Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
119 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Training Table Question Answering via SQL Query Decomposition (2402.13288v1)

Published 19 Feb 2024 in cs.DB and cs.AI

Abstract: Table Question-Answering involves both understanding the natural language query and grounding it in the context of the input table to extract the relevant information. In this context, many methods have highlighted the benefits of intermediate pre-training from SQL queries. However, while most approaches aim at generating final answers from inputs directly, we claim that there is better to do with SQL queries during training. By learning to imitate a restricted portion of SQL-like algebraic operations, we show that their execution flow provides intermediate supervision steps that allow increased generalization and structural reasoning compared with classical approaches of the field. Our study bridges the gap between semantic parsing and direct answering methods and provides useful insights regarding what types of operations should be predicted by a generative architecture or be preferably executed by an external algorithm.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (35)
  1. A survey on table question answering: recent advances. In China Conference on Knowledge Graph and Semantic Computing, pages 174–186. Springer, 2022.
  2. On the potential of lexico-logical alignments for semantic parsing to sql queries. arXiv preprint arXiv:2010.11246, 2020.
  3. Compositional semantic parsing on semi-structured tables. arXiv preprint arXiv:1508.00305, 2015.
  4. Tapex: Table pre-training via learning a neural sql executor. arXiv preprint arXiv:2107.07653, 2021.
  5. Tapas: Weakly supervised table parsing via pre-training. arXiv preprint arXiv:2004.02349, 2020.
  6. Unirpg: Unified discrete reasoning over table and text as program generation. arXiv preprint arXiv:2210.08249, 2022a.
  7. Intermediate-task transfer learning with pretrained models for natural language understanding: When and why does it work? arXiv preprint arXiv:2005.00628, 2020.
  8. Injecting numerical reasoning skills into language models. arXiv preprint arXiv:2004.04487, 2020.
  9. Grappa: Grammar-augmented pre-training for table semantic parsing. arXiv preprint arXiv:2009.13845, 2020.
  10. Mate: multi-view attention for table transformer efficiency. arXiv preprint arXiv:2109.04312, 2021.
  11. Time travel in llms: Tracing data contamination in large language models. arXiv preprint arXiv:2308.08493, 2023.
  12. Tuta: Tree-based transformers for generally structured table pre-training. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pages 1780–1790, 2021.
  13. Wenhu Chen. Large language models are few (1)-shot table reasoners. arXiv preprint arXiv:2210.06710, 2022.
  14. Binding language models in symbolic languages. arXiv preprint arXiv:2210.02875, 2022.
  15. Chain-of-table: Evolving tables in the reasoning chain for table understanding. arXiv preprint arXiv:2401.04398, 2024.
  16. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685, 2021.
  17. Qlora: Efficient finetuning of quantized llms. Advances in Neural Information Processing Systems, 36, 2024.
  18. Omnitab: Pretraining with natural and synthetic data for few-shot table-based question answering. arXiv preprint arXiv:2207.03637, 2022.
  19. Toolformer: Language models can teach themselves to use tools. Advances in Neural Information Processing Systems, 36, 2024.
  20. Rat-sql: Relation-aware schema encoding and linking for text-to-sql parsers. arXiv preprint arXiv:1911.04942, 2019.
  21. A discrete hard em approach for weakly supervised question answering. arXiv preprint arXiv:1909.04849, 2019.
  22. Seq2sql: Generating structured queries from natural language using reinforcement learning. arXiv preprint arXiv:1709.00103, 2017.
  23. Answering conversational questions on structured data without logical forms. arXiv preprint arXiv:1908.11787, 2019.
  24. Understanding tables with intermediate pre-training. arXiv preprint arXiv:2010.00571, 2020.
  25. Fortap: Using formulas for numerical-reasoning-aware table pretraining. arXiv preprint arXiv:2109.07323, 2021.
  26. Tacube: Pre-computing data cubes for answering numerical-reasoning questions over tabular data. arXiv preprint arXiv:2205.12682, 2022b.
  27. Tat-qa: A question answering benchmark on a hybrid of tabular and textual content in finance. arXiv preprint arXiv:2105.07624, 2021.
  28. Edgar F Codd. A relational model of data for large shared data banks. Communications of the ACM, 13(6):377–387, 1970.
  29. Bart: Denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension. arXiv preprint arXiv:1910.13461, 2019.
  30. Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems, 35:24824–24837, 2022.
  31. Tabert: Pretraining for joint understanding of textual and tabular data. arXiv preprint arXiv:2005.08314, 2020.
  32. Tableformer: Robust transformer modeling for table-text encoding. arXiv preprint arXiv:2203.00274, 2022.
  33. Dot: An efficient double transformer for nlp tasks with tables. arXiv preprint arXiv:2106.00479, 2021.
  34. Reastap: Injecting table reasoning skills during pre-training via synthetic reasoning examples. arXiv preprint arXiv:2210.12374, 2022.
  35. Large language models are versatile decomposers: Decompose evidence and questions for table-based reasoning. arXiv preprint arXiv:2301.13808, 2023.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (3)
  1. Raphaël Mouravieff (2 papers)
  2. Benjamin Piwowarski (38 papers)
  3. Sylvain Lamprier (40 papers)

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now
X Twitter Logo Streamline Icon: https://streamlinehq.com

Tweets