Multi-turn Text-to-SQL on CoSQL Out-of-domain
74Execution Accuracy[Long-Horizon] Warm-Start SFT + RL (Outcome Only)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| [Long-Horizon] Warm-Start SFT + RL (Outcome Only)Model Size=4B, Horizon Setting=Long-Horizon2025.10 | 74 | 63 | |
| [Long-Horizon] Warm-Start SFT + RL (Outcome + Process)Model Size=4B, Horizon Setting=Long-Horizon2025.10 | 74 | 63 | |
| [Long-Horizon] + Warm-Start SFT Only (Round 2)Model Size=4B, Horizon Setting=Long-Horizon, Learning Round=Round 22025.10 | 73.5 | 61.2 | |
| [Long-Horizon] Warm-Start SFT Only (Round 1)Model Size=4B, Horizon Setting=Long-Horizon, Learning Round=Round 12025.10 | 72.7 | 58.7 | |
| [Long-Horizon] + Warm-Start SFT Only (Round 3)Model Size=4B, Horizon Setting=Long-Horizon, Learning Round=Round 32025.10 | 72.3 | 61.8 | |
| [Short-Horizon] Direct RLModel Size=4B, Horizon Setting=Short-Horizon2025.10 | 71.4 | 61.3 | |
| [Long-Horizon] Warm-Start SFT + RL (Outcome + Process)Model Size=1.7B, Horizon Setting=Long-Horizon2025.10 | 70.4 | 59.8 | |
| [Long-Horizon] Warm-Start SFT + RL (Outcome Only)Model Size=1.7B, Horizon Setting=Long-Horizon2025.10 | 70.3 | 59.8 | |
| [Short-Horizon] SFTModel Size=4B, Horizon Setting=Short-Horizon2025.10 | 70.2 | 61 | |
| [Long-Horizon] + Warm-Start SFT Only (Round 3)Model Size=1.7B, Horizon Setting=Long-Horizon, Learning Round=Round 32025.10 | 68.8 | 56.2 | |
| [Long-Horizon] + Warm-Start SFT Only (Round 2)Model Size=1.7B, Horizon Setting=Long-Horizon, Learning Round=Round 22025.10 | 67.2 | 54.2 | |
| [Long-Horizon] Warm-Start SFT Only (Round 1)Model Size=1.7B, Horizon Setting=Long-Horizon, Learning Round=Round 12025.10 | 67.1 | 55.5 | |
| [Short-Horizon] Direct RLModel Size=1.7B, Horizon Setting=Short-Horizon2025.10 | 66.3 | 55.9 | |
| [Short-Horizon] SFTModel Size=1.7B, Horizon Setting=Short-Horizon2025.10 | 64.1 | 55.2 | |
| CoE-SQLModel Size=Closed-Source, Prompting Strategy=Few-shot, 16-shot2025.10 | 58.5 | 49.6 | |
| RASAT+PICARDModel Size=3B2025.10 | 55.8 | 48 |