mlsanityData to 7 Oct 2026

CAPER: Clause-Aligned Process Supervision for Text-to-SQL

arXiv 2606.03327 ↗First posted 2 Jun 2026cs.DB cs.CLcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Process reward models for reasoning

Abstract (arXiv)

Text-to-SQL systems are typically evaluated by query-level execution correctness, but this terminal signal provides little guidance about which intermediate SQL decision caused success or failure. Token-level dense supervision is also ill-suited: SQL tokens do not align with complete semantic decisions, can penalize execution-equivalent queries, and are difficult to label reliably at scale. We therefore propose CAPER, which automatically derives clause-level supervision via counterfactual intervention on the SQL abstract syntax tree, enabling root-cause error localization for reward modeling; the resulting data is used to train CAPER-9B, a lightweight Clause-PRM that provides clause-boundary feedback for policy optimization and candidate verification. Experiments on BIRD and Spider show that clause-aligned supervision not only improves execution accuracy, achieving up to a 15.3% relative EX improvement over GPT-5.4, but also strengthens failure-localization capability, reaching 84.53% accuracy and 90.60% MRR on held-out failures. Our project page is at https://github.com/banrichard/RL-NL2SQL.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Process reward models for reasoning, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 36 of its 52 references that are in the map. First posted on arXiv 2 Jun 2026.

Builds on

36 of 52 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025Graph-Reward-SQL: Execution-Free Reinforcement Learning for Text-to-SQL via Graph Matching and Stepwise Reward18 May
  2. 2025Reward-SQL: Boosting Text-to-SQL via Stepwise Execution-Aware Reasoning and Process-Supervised Rewards7 May
  3. 2022Solving math word problems with process- and outcome-based feedback25 Nov

Other references · 33

  1. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,655
  2. 2023Let's Verify Step by Stepcited by 3,598
  3. 2018Spider: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to-SQL Taskcited by 1,120
  4. 2017Counterfactual Multi-Agent Policy Gradientscited by 733
  5. 2017Seq2SQL: Generating Structured Queries from Natural Language using Reinforcement Learningcited by 679
  6. 2023Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLscited by 667
  7. 2023DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-Correctioncited by 465
  8. 2019RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parserscited by 420
  9. 2021PICARD: Parsing Incrementally for Constrained Auto-Regressive Decoding from Language Modelscited by 402
  10. 2023Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluationcited by 378
  11. 2024Spider 2.0: Evaluating Language Models on Real-World Enterprise Text-to-SQL Workflowscited by 223
  12. 2019Towards Complex Text-to-SQL in Cross-Domain Database with Intermediate Representationcited by 210
21 more
  1. 2024Next-Generation Database Interfaces: A Survey of LLM-based Text-to-SQLcited by 118
  2. 2019Counterfactual Off-Policy Evaluation with Gumbel-Max Structural Causal Modelscited by 114
  3. 2025OmniSQL: Synthesizing High-quality Text-to-SQL Data at Scalecited by 90
  4. 2025SQL-R1: Training Natural Language to SQL Reasoning Model By Reinforcement Learningcited by 67
  5. 2024A Survey of Text-to-SQL in the Era of LLMs: Where are we, and where are we going?cited by 54
  6. 2022Recent Advances in Text-to-SQL: A Survey of What We Have and What We Expectcited by 47
  7. 2025XiYan-SQL: A Novel Multi-Generator Framework For Text-to-SQLcited by 40
  8. 2025Alpha-SQL: Zero-Shot Text-to-SQL using Monte Carlo Tree Searchcited by 33
  9. 2018Bayesian Counterfactual Risk Minimizationcited by 28
  10. 2020Photon: A Robust Cross-Domain Text-to-SQL Systemcited by 26
  11. 2025LinkAlign: Scalable Schema Linking for Real-World Large-Scale Multi-Database Text-to-SQLcited by 26
  12. 2025DeepEye-SQL: A Software-Engineering-Inspired Text-to-SQL Frameworkcited by 26
  13. 2023CRUSH4SQL: Collective Retrieval Using Schema Hallucination For Text2SQLcited by 23
  14. 2025STaR-SQL: Self-Taught Reasoner for Text-to-SQLcited by 16
  15. 2021Weakly Supervised Text-to-SQL Parsing through Question Decompositioncited by 15
  16. 2025SHARE: An SLM-based Hierarchical Action CorREction Assistant for Text-to-SQLcited by 15
  17. 2025SWE-SQL: Illuminating LLM Pathways to Solve User SQL Issues in Real-World Applicationscited by 15
  18. 2025ExeSQL: Self-Taught Text-to-SQL Models with Execution-Driven Bootstrapping for SQL Dialectscited by 11
  19. 2023Sequential Counterfactual Risk Minimizationcited by 6
  20. 2025Abacus-SQL: A Text-to-SQL System Empowering Cross-Domain and Open-Domain Database Retrievalcited by 3
  21. 2025STARQA: A Question Answering Dataset for Complex Analytical Reasoning over Structured Databasescited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Progress-SQL: Improving Reinforcement Learning for Text-to-SQL via Progressive Rewards
  2. 2026ACTS-SQL: Agentic and Critic-Oriented Tree-Structured SQL Correctness with Large Language Models
  3. 2025SQLCritic: Correcting Text-to-SQL Generation via Clause-wise Critic
  4. 2026Every Step Counts: Step-Level Credit Assignment for Tool-Integrated Text-to-SQL
  5. 2026Test-Time Verification for Text-to-SQL via Outcome Reward Models
  6. 2026SQLConductor: Search-to-Policy Learning for Step-wise Text-to-SQL Orchestration
  7. 2026LIMIT: Less Is More for Instruction Tuning in Text-to-SQL
  8. 2025Arctic-Text2SQL-R1: Simple Rewards, Strong Reasoning in Text-to-SQL