Reasoning on GSM8K (test) using Pass@1
61.03Pass@1A2D2
Evaluation Results
| Method | Links | |
|---|---|---|
| A2D2Sampling Steps=2562026.06 | 61.03 | |
| A2D2Sampling Steps=1282026.06 | 60.96 | |
| A2D2Sampling Steps=10242026.06 | 57.62 | |
| A2D2Sampling Steps=5122026.06 | 57.54 | |
| Pre-trained + IFTSampling Steps=10242026.06 | 41.39 | |
| Pre-trained + IFTSampling Steps=5122026.06 | 39.88 | |
| Pre-trained + IFTSampling Steps=2562026.06 | 37.98 | |
| Pre-trained + IFTSampling Steps=1282026.06 | 35.71 |