BIRD

The BIRD dataset is a benchmark that contains a collection of natural language questions and their corresponding SQL queries, used to evaluate the performance of models in the Text-to-SQL generation task.

Papers using BIRD (9)

A State-of-the-Art SQL Reasoning Model using RLVR2025 · 3 citesGraph-Reward-SQL: Execution-Free Reinforcement Learning for Text-to-SQL via Graph Matching and Stepwise Reward2025 · 1 citesReasoning-SQL: Reinforcement Learning with SQL Tailored Partial Rewards for Reasoning-Enhanced Text-to-SQL2025 · 1 citesProgress-SQL: Improving Reinforcement Learning for Text-to-SQL via Progressive Rewards2026SQL-ASTRA: Alleviating Sparse Feedback in Agentic SQL via Column-Set Matching and Trajectory Aggregation2026Learning to Self-Evolve2026PaVeRL-SQL: Text-to-SQL via Partial-Match Rewards and Verbal Reinforcement Learning2025Arctic-Text2SQL-R1: Simple Rewards, Strong Reasoning in Text-to-SQL2025 · 2 citesHES-SQL: Hybrid Reasoning for Efficient Text-to-SQL with Structural Skeleton Guidance2025