NL2SQL on Archer Chinese zh (dev)
84.6Execution Accuracy (EX)SOMA-SQL
Evaluation Results
| Method | Links | |
|---|---|---|
| SOMA-SQLBackbone LLM=GPT-52026.06 | 84.6 | |
| SOMA-SQLBackbone LLM=GPT-5.3 Codex2026.06 | 78.9 | |
| SOMA-SQLBackbone LLM=Gemma-4-31B2026.06 | 75 | |
| AutoLinkBackbone LLM=GPT-5.3 Codex2026.06 | 73.1 | |
| AutoLinkBackbone LLM=GPT-52026.06 | 70.2 | |
| AutoLinkBackbone LLM=Gemma-4-31B2026.06 | 54.8 | |
| DeepEye-SQLBackbone LLM=GPT-5.3 Codex2026.06 | 50 | |
| DeepEye-SQLBackbone LLM=GPT-52026.06 | 49 | |
| DeepEye-SQLBackbone LLM=Gemma-4-31B2026.06 | 42.3 |