Text-to-SQL on Spider Realistic (EX, TS)
88.6Execution Accuracy (EX)FineStep
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FineStepModel Size=4B2026.05 | 88.6 | 80.9 | |
| GIGPOModel Size=4B2026.05 | 87.6 | 80.5 | |
| DAPOModel Size=4B2026.05 | 87.2 | 78.5 | |
| GSPOModel Size=4B2026.05 | 86 | 78.1 | |
| FineStepModel Size=8B2026.05 | 86 | 77.9 | |
| GIGPOModel Size=8B2026.05 | 85.6 | 77 | |
| DAPOModel Size=8B2026.05 | 85.4 | 76 | |
| GSPOModel Size=8B2026.05 | 84.8 | 77 | |
| FineStepModel Size=30B-A3B2026.05 | 84.1 | 76 | |
| Progress-SQL-7BReward=Progressive2026.06 | 84.1 | 76.4 | |
| Progress-SQL-14BReward=Progressive2026.06 | 84.1 | 76.6 | |
| GRPOModel Size=4B2026.05 | 83.7 | 75.2 | |
| GRPOModel Size=8B2026.05 | 83.7 | 75.8 | |
| SQL-Trail-7BReward=One-shot2026.06 | 83.7 | 74.4 | |
| EXPO-SQLDetail=RL, Base Model=Qwen2.5-Coder-7B2026.04 | 83.4 | — | |
| GRPOModel Size=30B-A3B2026.05 | 83.3 | 73.6 | |
| SQL-R1Detail=RL, Base Model=Qwen2.5-Coder-7B2026.04 | 83.3 | — | |
| SFT CodeS-7BReward=–2026.06 | 82.9 | 77.2 | |
| SENSE-7BDetail=SFT, Base Model=Qwen2.5-Coder-7B2026.04 | 82.7 | — | |
| No-TrainModel Size=4B2026.05 | 82.5 | 66.3 | |
| No-TrainModel Size=30B-A3B2026.05 | 82.5 | 64.6 | |
| Qwen2.5-Coder-14B-InstructReward=–2026.06 | 81.5 | 74.6 | |
| SkyRL-SQL-7BReward=One-shot2026.06 | 81.1 | — | |
| No-TrainModel Size=8B2026.05 | 80.9 | 65.4 | |
| OmniSQL-7BReward=–2026.06 | 80.5 | 75 | |
| KE-ICLLLM=Gemini-Pro-1.52026.05 | 80.11 | — | |
| SQL-R1-7BReward=One-shot2026.06 | 80.1 | 61.6 | |
| MAC-SQLDetail=Prompt, Base Model=GPT-42026.04 | 79.9 | — | |
| CodeSLLM=Gemini-Pro-1.5, Setting=ICL2026.05 | 79.72 | — | |
| DAIL-SQLLLM=Gemini-Pro-1.5, Setting=ICL2026.05 | 78.3 | — | |
| KE-ICLLLM=GPT-4o2026.05 | 78.14 | — | |
| OmniSQL-7BDetail=SFT, Base Model=Qwen2.5-Coder-7B2026.04 | 78 | — | |
| KE-RLLLM=Qwen2.5-Coder-7B2026.05 | 77.58 | — | |
| CodeSLLM=GPT-4o, Setting=ICL2026.05 | 76.77 | — | |
| ACT-SQLDetail=Prompt, Base Model=GPT-42026.04 | 75.8 | — | |
| DAIL-SQLLLM=GPT-4o, Setting=ICL2026.05 | 75.51 | — | |
| Qwen2.5-Coder-7BDetail=-, Base Model=Qwen2.5-Coder-7B2026.04 | 75.4 | — | |
| Qwen2.5-Coder-7B-InstructReward=–2026.06 | 74.4 | 66.9 | |
| OmniLLM=Qwen2.5-Coder-7B, Setting=RL2026.05 | 72.43 | — | |
| KE-ICLLLM=Qwen2.5-Coder-7B2026.05 | 71.45 | — | |
| SQL-GENLLM=Qwen2.5-Coder-7B, Setting=RL2026.05 | 70.35 | — | |
| KE-RLLLM=Deepseek-Coder-7B2026.05 | 68.73 | — | |
| KE-ICLLLM=Deepseek-Coder-7B2026.05 | 66.2 | — | |
| OmniLLM=Deepseek-Coder-7B, Setting=RL2026.05 | 64.85 | — | |
| CodeSLLM=Qwen2.5-Coder-7B, Setting=ICL2026.05 | 64.8 | — | |
| DAIL-SQLLLM=Qwen2.5-Coder-7B, Setting=ICL2026.05 | 63.71 | — | |
| SQL-GENLLM=Deepseek-Coder-7B, Setting=RL2026.05 | 62.72 | — | |
| KE-RLLLM=Granite-3.1-8B2026.05 | 60.93 | — | |
| CodeSLLM=Deepseek-Coder-7B, Setting=ICL2026.05 | 57.13 | — | |
| SQL-GENLLM=Granite-3.1-8B, Setting=RL2026.05 | 57.01 | — | |
| DAIL-SQLLLM=Deepseek-Coder-7B, Setting=ICL2026.05 | 56.02 | — | |
| OmniLLM=Granite-3.1-8B, Setting=RL2026.05 | 55.1 | — | |
| KE-ICLLLM=Granite-3.1-8B2026.05 | 50.13 | — | |
| CodeSLLM=Granite-3.1-8B, Setting=ICL2026.05 | 47.43 | — | |
| DAIL-SQLLLM=Granite-3.1-8B, Setting=ICL2026.05 | 46.1 | — |