COAL-SQL: Coverage-Guided Augmentation and Failure-Driven Learning for Text-to-SQL Post-Training

What changed

COAL-SQL combines coverage-guided augmentation and failure-driven learning for text-to-SQL post-training, achieving 64.9% execution accuracy on the BIRD development set with 12,600 examples.

Why you should care

Coverage-guided data augmentation and failure-driven learning can improve text-to-SQL post-training efficiency for open-source large language models.

Your move

Watch. Monitor further independent replication of COAL-SQL across diverse SQL generation benchmarks.

Evidence

PreliminaryarXiv — cs.CL ↗

What to watch next

Independent benchmarks or reproduction attempts evaluating COAL-SQL on alternative text-to-SQL datasets beyond BIRD.

Event

research

Event date

Sep 21, 2026

Relevant to

General AI readers