Clarifying Questions on OpenBookQA (test)
28AccuracyRefit
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| RefitMode=Thinking Mode2025.06 | 28 | 26 | 4.55 | 5.55 | 5.32 | 4.37 | 4.95 | 4.94 | |
| SwiftMode=Thinking Mode2025.06 | 24 | 46 | 4.67 | 5.7 | 5.39 | 4.41 | 5.13 | 5.23 | |
| StepDPOMode=Thinking Mode2025.06 | 20 | 26 | 4.27 | 5.31 | 5.01 | 4.07 | 4.67 | 4.73 | |
| BaseMode=Thinking Mode2025.06 | 10 | 47 | 4.89 | 5.93 | 5.63 | 4.6 | 5.37 | 5.35 | |
| DPOMode=Thinking Mode2025.06 | 10 | 30 | 4.27 | 5.29 | 5.05 | 4.04 | 4.75 | 4.87 | |
| STaR-GATE-DMode=Thinking Mode2025.06 | 10 | 4 | 4.11 | 5.13 | 4.85 | 3.81 | 4.42 | 4.59 |