Cleaning RL dataset boosts Text-to-SQL performance significantly
joecole · x · 2026-08-28
A research team achieved a significant boost in text-to-SQL post-training performance by meticulously cleaning up bad samples in a popular RL dataset. The final model even beats Sol and Fable on some benchmarks.
More from Research
- Priming Models: Teacher-Free One-Step Generation Hits FID 1.19 With 12x Less Compute — du_yilun · 2026-10-15
- vLLM Achieves Bitwise-Consistent RL Training, Fixing KL Blowups and Score Collapse — vllm_project · 2026-10-15
- David Deutsch: Proofs Are No Longer a Useful Proxy for Mathematical Progress — francoisfleuret · 2026-10-15
- STEMBench launches: 6 STEM domains, zero full-target successes across all runs — mariofilhoml · 2026-10-15
- OpenAI gives $5M to Simons Foundation's $15M push to help mathematicians navigate AI-era results — OpenAI · 2026-10-15
- Qwen releases BlenderWorld, a 223-task benchmark for 3D creation in Blender — yshan2u · 2026-10-15