Mathematical Reasoning on GSM8K (test) (Acc, Steer, UTS, Entropy)
91.8AccuracyUTS-guided system
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| UTS-guided systemTemperature (T)=1.02026.02 | 91.8 | 1.72 | 1.612 | 0.0006 | |
| UTS-guided systemTemperature (T)=0.52026.02 | 91.65 | 1.45 | 1.613 | 0.0001 | |
| UTS-guided systemTemperature (T)=1.52026.02 | 91.11 | 1.85 | 1.61 | 0.0015 | |
| UTS-guided systemTemperature (T)=2.02026.02 | 90.32 | 2.1 | 1.61 | 0.0018 | |
| UTS-guided systemTemperature (T)=2.52026.02 | 90.31 | 2.28 | 1.605 | 0.0068 | |
| UTS-guided systemTemperature (T)=3.02026.02 | 88.84 | 2.44 | 1.603 | 0.0058 |