{"url":"/sota/text-to-sql-on-bird-big-bench-for-large-scale","task":{"name":"Text-To-SQL","url":"/task/text-to-sql","note":null},"dataset":{"name":"BIRD (BIg Bench for LaRge-scale Database Grounded Text-to-SQL Evaluation)","url":"/dataset/bird-sql"},"category":"Natural Language Processing","categories":["Computer Code","Natural Language Processing"],"category_note":null,"description":"**Text-to-SQL** is a task in natural language processing (NLP) where the goal is to automatically generate SQL queries from natural language text. The task involves converting the text input into a structured representation and then using this representation to generate a semantically correct SQL query that can be executed on a database.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [SyntaxSQLNet](https://arxiv.org/pdf/1810.05237v2.pdf) )</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Execution Accuracy % (Test)","Execution Accuracy % (Dev)","Execution Accurarcy (Human)"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Execution Accuracy % (Test)":"higher","Execution Accuracy % (Dev)":"higher","Execution Accurarcy (Human)":null}},"counts":{"rows":41,"rows_with_code":14,"rows_with_paper_page":16,"rows_dated":16,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"XiYan-SQL","metrics":{"Execution Accuracy % (Dev)":"73.34","Execution Accuracy % (Test)":"75.63"},"uses_additional_data":false,"paper_date":"2024-11-13","paper":"/paper/xiyan-sql-a-multi-generator-ensemble","paper_url":"https://arxiv.org/abs/2411.08599v3","paper_title":"A Preview of XiYan-SQL: A Multi-Generator Ensemble Framework for Text-to-SQL","code":"https://github.com/XGenerationLab/XiYan-SQL","n_code_links":5,"syntology":null},{"rank_in_archive_order":2,"model":"DSAIR + GPT-4o","metrics":{"Execution Accuracy % (Dev)":"74.32","Execution Accuracy % (Test)":"74.12"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":3,"model":"CHASE-SQL + Gemini","metrics":{"Execution Accuracy % (Dev)":"73.14","Execution Accuracy % (Test)":"74.06"},"uses_additional_data":false,"paper_date":"2024-10-02","paper":"/paper/chase-sql-multi-path-reasoning-and-preference","paper_url":"https://arxiv.org/abs/2410.01943v1","paper_title":"CHASE-SQL: Multi-Path Reasoning and Preference Optimized Candidate Selection in Text-to-SQL","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":4,"model":"ExSL + granite-34b-code","metrics":{"Execution Accuracy % (Dev)":"72.43","Execution Accuracy % (Test)":"73.17"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"OpenSearch-SQL+ v2 + GPT-4o","metrics":{"Execution Accuracy % (Dev)":"69.3","Execution Accuracy % (Test)":"72.28"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":6,"model":"Distillery + GPT-4o","metrics":{"Execution Accuracy % (Dev)":"67.21","Execution Accuracy % (Test)":"71.83"},"uses_additional_data":false,"paper_date":"2024-08-14","paper":"/paper/the-death-of-schema-linking-text-to-sql-in","paper_url":"https://arxiv.org/abs/2408.07702v2","paper_title":"The Death of Schema Linking? Text-to-SQL in the Age of Well-Reasoned Language Models","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":7,"model":"Insights AI","metrics":{"Execution Accuracy % (Dev)":"72.16","Execution Accuracy % (Test)":"70.26"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":8,"model":"PURPLE + RED + GPT-4o","metrics":{"Execution Accuracy % (Dev)":"68.12","Execution Accuracy % (Test)":"70.21"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":9,"model":"MCTS-SQL","metrics":{"Execution Accuracy % (Dev)":"68.91","Execution Accuracy % (Test)":"69.40"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":10,"model":"RECAP + Gemini","metrics":{"Execution Accuracy % (Dev)":"66.95","Execution Accuracy % (Test)":"69.03"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":11,"model":"ByteBrain","metrics":{"Execution Accuracy % (Dev)":"65.45","Execution Accuracy % (Test)":"68.87"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":12,"model":"ExSL + granite-20b-code","metrics":{"Execution Accuracy % (Dev)":"65.38","Execution Accuracy % (Test)":"67.86"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":13,"model":"CHESS","metrics":{"Execution Accuracy % (Dev)":"65","Execution Accuracy % (Test)":"66.69"},"uses_additional_data":false,"paper_date":"2024-05-27","paper":"/paper/chess-contextual-harnessing-for-efficient-sql","paper_url":"https://arxiv.org/abs/2405.16755v3","paper_title":"CHESS: Contextual Harnessing for Efficient SQL Synthesis","code":"https://github.com/shayantalaei/chess","n_code_links":2,"syntology":{"n_ran":6,"n_unverified":0,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":14,"model":"Arcwise + GPT-4o","metrics":{"Execution Accuracy % (Dev)":"67.99","Execution Accuracy % (Test)":"66.21"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":15,"model":"MCS-SQL + GPT-4","metrics":{"Execution Accuracy % (Dev)":"63.36","Execution Accuracy % (Test)":"65.45"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":16,"model":"SCL-SQL","metrics":{"Execution Accuracy % (Dev)":"64.73","Execution Accuracy % (Test)":"65.23"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":17,"model":"OpenSearch-SQL v1 + GPT-4","metrics":{"Execution Accuracy % (Dev)":"61.34","Execution Accuracy % (Test)":"64.95"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":18,"model":"PB-SQL v1","metrics":{"Execution Accuracy % (Dev)":"60.5","Execution Accuracy % (Test)":"64.84"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":19,"model":"PURPLE + GPT-4o","metrics":{"Execution Accuracy % (Dev)":"62.97","Execution Accuracy % (Test)":"64.51"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":20,"model":"MSL-SQL + DeepSeek-V2.5","metrics":{"Execution Accuracy % (Dev)":"66.82","Execution Accuracy % (Test)":"64.00"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":21,"model":"SENSE-13B","metrics":{"Execution Accuracy % (Dev)":"55.48","Execution Accuracy % (Test)":"63.39"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":22,"model":"SENSE","metrics":{"Execution Accuracy % (Dev)":"55.48","Execution Accuracy % (Test)":"63.39"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":23,"model":"GRA-SQL","metrics":{"Execution Accuracy % (Dev)":"62.58","Execution Accuracy % (Test)":"63.22"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":24,"model":"SuperSQL","metrics":{"Execution Accuracy % (Dev)":"58.5","Execution Accuracy % (Test)":"62.66"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":25,"model":"Dubo-SQL, v1","metrics":{"Execution Accuracy % (Dev)":"59.71","Execution Accuracy % (Test)":"60.71"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":26,"model":"SFT CodeS-15B","metrics":{"Execution Accuracy % (Dev)":"58.47","Execution Accuracy % (Test)":"60.37"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":27,"model":"MAC-SQL + GPT-4","metrics":{"Execution Accuracy % (Dev)":"57.56","Execution Accuracy % (Test)":"59.59"},"uses_additional_data":false,"paper_date":"2023-12-18","paper":"/paper/mac-sql-multi-agent-collaboration-for-text-to","paper_url":"https://arxiv.org/abs/2312.11242v5","paper_title":"MAC-SQL: A Multi-Agent Collaborative Framework for Text-to-SQL","code":"https://github.com/wbbeyourself/mac-sql","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":28,"model":"SFT CodeS-7B","metrics":{"Execution Accuracy % (Dev)":"57.17","Execution Accuracy % (Test)":"59.25"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":29,"model":"DAIL-SQL + GPT-4","metrics":{"Execution Accuracy % (Dev)":"54.76","Execution Accuracy % (Test)":"57.41"},"uses_additional_data":false,"paper_date":"2023-08-29","paper":"/paper/text-to-sql-empowered-by-large-language","paper_url":"https://arxiv.org/abs/2308.15363v4","paper_title":"Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation","code":"https://github.com/beachwang/dail-sql","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":2,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":30,"model":"DIN-SQL + GPT-4","metrics":{"Execution Accuracy % (Dev)":"50.72","Execution Accuracy % (Test)":"55.90"},"uses_additional_data":false,"paper_date":"2023-04-21","paper":"/paper/din-sql-decomposed-in-context-learning-of-1","paper_url":"https://arxiv.org/abs/2304.11015v3","paper_title":"DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-Correction","code":"https://github.com/mohammadrezapourreza/few-shot-nl2sql-with-prompting","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":4,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":31,"model":"GPT-4 (Baseline)","metrics":{"Execution Accuracy % (Dev)":"46.35","Execution Accuracy % (Test)":"54.89"},"uses_additional_data":false,"paper_date":"2023-09-28","paper":"/paper/can-llms-effectively-leverage-structural","paper_url":"https://arxiv.org/abs/2309.16595v4","paper_title":"Can LLMs Effectively Leverage Graph Structural Information through Prompts, and Why?","code":"https://github.com/CurryTang/Graph-LLM","n_code_links":2,"syntology":{"n_ran":10,"n_unverified":5,"n_samples":15,"n_pointer_only_licence":2}},{"rank_in_archive_order":32,"model":"Claude-2 (Baseline)","metrics":{"Execution Accuracy % (Dev)":"42.70","Execution Accuracy % (Test)":"49.02"},"uses_additional_data":false,"paper_date":"2023-09-28","paper":"/paper/can-llms-effectively-leverage-structural","paper_url":"https://arxiv.org/abs/2309.16595v4","paper_title":"Can LLMs Effectively Leverage Graph Structural Information through Prompts, and Why?","code":"https://github.com/CurryTang/Graph-LLM","n_code_links":2,"syntology":{"n_ran":10,"n_unverified":5,"n_samples":15,"n_pointer_only_licence":2}},{"rank_in_archive_order":33,"model":"Open SQL-7B","metrics":{"Execution Accuracy % (Dev)":"37.68","Execution Accuracy % (Test)":"47.74"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":34,"model":"CoT + ChatGPT","metrics":{"Execution Accuracy % (Dev)":"36.64","Execution Accuracy % (Test)":"40.08"},"uses_additional_data":false,"paper_date":"2023-05-04","paper":"/paper/can-llm-already-serve-as-a-database-interface","paper_url":"https://arxiv.org/abs/2305.03111v3","paper_title":"Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs","code":"https://github.com/bird-bench/mini_dev","n_code_links":1,"syntology":{"n_ran":6,"n_unverified":0,"n_samples":6,"n_pointer_only_licence":6}},{"rank_in_archive_order":35,"model":"ChatGPT (Baseline)","metrics":{"Execution Accuracy % (Dev)":"37.22","Execution Accuracy % (Test)":"39.30"},"uses_additional_data":false,"paper_date":"2023-05-04","paper":"/paper/can-llm-already-serve-as-a-database-interface","paper_url":"https://arxiv.org/abs/2305.03111v3","paper_title":"Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs","code":"https://github.com/bird-bench/mini_dev","n_code_links":1,"syntology":{"n_ran":6,"n_unverified":0,"n_samples":6,"n_pointer_only_licence":6}},{"rank_in_archive_order":36,"model":"Codex (Baseline)","metrics":{"Execution Accuracy % (Dev)":"34.35","Execution Accuracy % (Test)":"36.47"},"uses_additional_data":false,"paper_date":"2023-05-04","paper":"/paper/can-llm-already-serve-as-a-database-interface","paper_url":"https://arxiv.org/abs/2305.03111v3","paper_title":"Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs","code":"https://github.com/bird-bench/mini_dev","n_code_links":1,"syntology":{"n_ran":6,"n_unverified":0,"n_samples":6,"n_pointer_only_licence":6}},{"rank_in_archive_order":37,"model":"Palm-2 (Baseline)","metrics":{"Execution Accuracy % (Dev)":"27.38","Execution Accuracy % (Test)":"33.04"},"uses_additional_data":false,"paper_date":"2023-05-04","paper":"/paper/can-llm-already-serve-as-a-database-interface","paper_url":"https://arxiv.org/abs/2305.03111v3","paper_title":"Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs","code":"https://github.com/bird-bench/mini_dev","n_code_links":1,"syntology":{"n_ran":6,"n_unverified":0,"n_samples":6,"n_pointer_only_licence":6}},{"rank_in_archive_order":38,"model":"MSc-SQL","metrics":{"Execution Accuracy % (Dev)":"65.6"},"uses_additional_data":false,"paper_date":"2024-10-16","paper":"/paper/msc-sql-multi-sample-critiquing-small","paper_url":"https://arxiv.org/abs/2410.12916v1","paper_title":"MSc-SQL: Multi-Sample Critiquing Small Language Models For Text-To-SQL Translation","code":"https://github.com/layer6ai-labs/msc-sql","n_code_links":1,"syntology":{"n_ran":9,"n_unverified":0,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":39,"model":"SFT CodeS-15B + SQLFixAgent","metrics":{"Execution Accuracy % (Dev)":"64.62"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":40,"model":"DELLM + MAC-SQL","metrics":{"Execution Accuracy % (Dev)":"48.92"},"uses_additional_data":false,"paper_date":"2024-02-18","paper":"/paper/knowledge-to-sql-enhancing-sql-generation","paper_url":"https://arxiv.org/abs/2402.11517v3","paper_title":"Knowledge-to-SQL: Enhancing SQL Generation with Data Expert LLM","code":"https://github.com/Rcrossmeister/Knowledge-to-SQL","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":1,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":41,"model":"Human Performance","metrics":{"Execution Accurarcy (Human)":"92.96"},"uses_additional_data":false,"paper_date":"2023-05-04","paper":"/paper/can-llm-already-serve-as-a-database-interface","paper_url":"https://arxiv.org/abs/2305.03111v3","paper_title":"Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs","code":"https://github.com/bird-bench/mini_dev","n_code_links":1,"syntology":{"n_ran":6,"n_unverified":0,"n_samples":6,"n_pointer_only_licence":6}}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,885 of the 9,623 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9623,"papers_checked":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":13,"rows_with_any_sample_ran":12,"distinct_papers_with_graph_line":8,"distinct_papers_with_any_sample_ran":7,"samples_over_distinct_papers":{"n_ran":37,"n_unverified":12,"n_samples":49,"n_pointer_only_licence":11,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":71,"n_unverified":17,"n_samples":88,"n_pointer_only_licence":37,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}