Mathematical Reasoning on Math 500 (Accuracy, #Tokens)
87.87Math 500 AccuracySPS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SPSModel=Qwen2.5-7B-Instruct, N=82026.05 | 87.87 | 9,076 | |
| PB-SMCModel=Qwen2.5-7B-Instruct, N=82026.05 | 87.47 | 10,628 | |
| SPSModel=Qwen2.5-3B-Instruct, N=82026.05 | 83.2 | 9,359 | |
| SPSModel=Phi-4-mini-Instruct, N=82026.05 | 82.67 | 7,959 | |
| PB-SMCModel=Qwen2.5-3B-Instruct, N=82026.05 | 82.07 | 11,727 | |
| PB-SMCModel=Phi-4-mini-Instruct, N=82026.05 | 82.07 | 10,422 | |
| Base ModelModel=Qwen2.5-7B-Instruct, N=12026.05 | 76 | 584 | |
| Base ModelModel=Qwen2.5-3B-Instruct, N=12026.05 | 64.47 | 607 | |
| Base ModelModel=Phi-4-mini-Instruct, N=12026.05 | 62.8 | 525 |