Text-to-SQL on Spider Robustness Suite SYN REALISTIC DK (dev)
78.9Execution Accuracy (SYN)SQLForge-LM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SQLForge-LMBackbone=CodeLlama-13B, Multiple Calls=false2025.05 | 78.9 | 84.8 | 82.1 | 81.9 | |
| SENSEBackbone=CodeLlama-13B, Multiple Calls=false2025.05 | 77.6 | 84.1 | 80.2 | 80.6 | |
| CodeSBackbone=StarCoder-15B, Multiple Calls=false2025.05 | 77 | 83.1 | 70.7 | 76.9 | |
| RESDSQLBackbone=ROBERTa-3B, Multiple Calls=true2025.05 | 76.9 | 81.9 | 66 | 74.9 | |
| SQLForge-LMBackbone=CodeLlama-7B, Multiple Calls=false2025.05 | 74.6 | 83.3 | 78.9 | 78.8 | |
| SENSEBackbone=CodeLlama-7B, Multiple Calls=false2025.05 | 72.6 | 82.7 | 77.9 | 77.7 |