Mathematical Reasoning Repair on GSM8K (test)
1,162Pass CountDirect
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DirectTotal=13192026.05 | 1,162 | 157 | 0 | 1 | |
| Self-ReflectionTotal=13192026.05 | 1,155 | 164 | 35 | 0.88 | |
| Self-RefineTotal=13192026.05 | 1,148 | 171 | 5 | 0.99 | |
| CausalFlowTotal=13192026.05 | 989 | 330 | 173 | 0.87 |