Numerical Reasoning on DROP (Non-football 500 test subset)
74.2AccuracyLeast-to-Most
Evaluation Results
| Method | Links | |
|---|---|---|
| Least-to-MostPrompting method=Least-to-Most, Model=text-davinci-0022022.05 | 74.2 | |
| Chain-of-ThoughtPrompting method=Chain-of-Thought, Model=text-davinci-0022022.05 | 60.8 | |
| Standard promptingPrompting method=Standard prompting, Model=text-davinci-0022022.05 | 49.4 | |
| Zero-ShotPrompting method=Zero-Shot, Model=text-davinci-0022022.05 | 27 |