Text-to-SQL on Spider2-Lite BigQuery
62.07Pass@1DIVSKILL-SQL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DIVSKILL-SQLBackbone LLM=Opus 4.62026.05 | 62.07 | 73.17 | 64.88 | |
| DIVSKILL-SQLModel=Opus 4.62026.05 | 62.07 | 73.17 | 64.88 | |
| CHASE-SQLBackbone LLM=Opus 4.62026.05 | 59.02 | 71.71 | 56.59 | |
| CHASE-SQLModel=Opus 4.62026.05 | 59.02 | 71.71 | 56.59 | |
| DIVSKILL-SQLBackbone=GPT-5.4, K (number of candidates)=82026.05 | 57.8 | 77.56 | 63.41 | |
| CHASE-SQLBackbone=GPT-5.4, K (number of candidates)=82026.05 | 53.9 | 74.15 | 59.51 | |
| REFORCEBackbone LLM=Opus 4.62026.05 | 51.1 | — | 55.12 | |
| REFORCEModel=Opus 4.62026.05 | 51.1 | — | 55.12 | |
| REFORCEBackbone=GPT-5.4, K (number of candidates)=dynamic2026.05 | 48.29 | — | 51.22 | |
| DIN-SQLBackbone LLM=Opus 4.62026.05 | 18.54 | — | — | |
| DIN-SQLModel=Opus 4.62026.05 | 18.54 | — | — | |
| DIN-SQLBackbone=GPT-5.42026.05 | 2.93 | — | — |