metadata
license: apache-2.0
base_model: Qwen/Qwen2.5-Coder-3B-Instruct
tags:
- text-to-sql
- self-play
- grpo
- sql-zero
sql-zero-3b-solver-iter2-s100
Solver 3B iter2 c/ repetition penalty, checkpoint canonico (best-dev s100/139). Spider 0.728, BIRD 0.376.
Parte do projeto SQL-Zero (self-play zero-anotacao para Text-to-SQL, adaptacao do Dr. Zero). Protocolo de avaliacao: greedy@1, schema completo (DDL+3 rows), execution accuracy. Treinado com verl (GRPO); reward de execucao 3-tier (1.0/0.1/0.0).