Track trials, findings, and clinical evidence as it lands.
Search papers, labs, and topics across Lattice.
FALCON addresses the oversimplification of existing NL-to-SQL synthetic data by introducing a framework that generates ambiguity-aware, structurally complex queries at low cost using compact open models. The method combines reserved-word SQL seeding and persona-based prompting, with alignment-based filtering to preserve difficulty and distinguish valid complex queries from errors. Training on FALCON-generated data significantly improves model performance on increasingly complex queries, outperforming baselines and recovering performance on simpler queries when mixed with existing benchmarks.
Ambiguity-aware NL2SQL generation framework produces realistic, complex queries that boost model performance on challenging real-world benchmarks, outperforming existing synthetic data.
Relational databases are among the most widely deployed forms of structured knowledge, and natural language access to them requires grounding language onto schema entities and relations while handling the ambiguity inherent in how people phrase requests. Existing synthetic NL-to-SQL data generation methods largely ignore this ambiguity and produce oversimplified queries that fail to prepare models for the complexity of real-world structured knowledge access. We present FALCON, a framework that generates realistic, ambiguity-aware NL-to-SQL data matching the complexity of challenging real-world benchmarks, at low cost using compact open models. Our approach combines reserved-word SQL seeding and persona-based prompting to generate structurally complex queries, while alignment-based filtering preserves difficulty by distinguishing genuinely incorrect examples from complex but valid queries. Human evaluation confirms consistent high quality across model sizes, and our generated data exceeds existing benchmarks in both SQL complexity and natural language richness. Difficulty-stratified analysis shows models trained on FALCON data increasingly outperform baseline-trained models as query complexity increases, validating our pipeline's success in generating challenging training data. When combined with a small proportion of existing benchmark data, mixed training recovers performance on simpler queries while preserving these advantages on complex ones. The model- and database-agnostic design enables organizations to generate high-complexity NL-to-SQL training data locally without external APIs.