Mathematical Reasoning on GSM8K (AL & TPS)
256.28Inference Speed (TPS)PRISM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PRISMModel=LLaMA-2, Temp=T = 02026.02 | 256.28 | 4.68 | |
| HASSModel=LLaMA-2, Temp=T = 02026.02 | 238.27 | 4.51 | |
| PRISMModel=LLaMA-3, Temp=T = 02026.02 | 235.85 | 5.06 | |
| PRISMModel=LLaMA-2, Temp=T = 12026.02 | 234.73 | 4.63 | |
| EAGLE-2Model=LLaMA-2, Temp=T = 02026.02 | 222.95 | 4.26 | |
| HASSModel=LLaMA-2, Temp=T = 12026.02 | 218.77 | 4.5 | |
| HASSModel=LLaMA-3, Temp=T = 02026.02 | 212.79 | 4.74 | |
| EAGLE-2Model=LLaMA-2, Temp=T = 12026.02 | 205.44 | 4.24 | |
| EAGLE-2Model=LLaMA-3, Temp=T = 02026.02 | 186.82 | 4.26 | |
| PRISMModel=LLaMA-3, Temp=T = 12026.02 | 170.26 | 4.71 | |
| StandardModel=LLaMA-3, Temp=T = 02026.02 | 166.41 | 5.75 | |
| HASSModel=LLaMA-3, Temp=T = 12026.02 | 157.93 | 4.45 | |
| StandardModel=LLaMA-3, Temp=T = 12026.02 | 152.54 | 5.31 | |
| EAGLE-2Model=LLaMA-3, Temp=T = 12026.02 | 142.84 | 4.02 | |
| StandardModel=LLaMA-2, Temp=T = 02026.02 | 140.77 | 2.88 | |
| StandardModel=LLaMA-2, Temp=T = 12026.02 | 138.21 | 2.85 | |
| VanillaModel=LLaMA-2, Temp=T = 02026.02 | 104.25 | — | |
| VanillaModel=LLaMA-2, Temp=T = 12026.02 | 102.54 | — | |
| VanillaModel=LLaMA-3, Temp=T = 02026.02 | 91.42 | — | |
| VanillaModel=LLaMA-3, Temp=T = 12026.02 | 90.83 | — |