Mathematical Reasoning on GSM8K (Tokens, Pass@5, Cons@5)
456.9Tokens UsedBasel. P^{(0)}_{system}
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Basel. P^{(0)}_{system}Model=DS-8B2026.04 | 456.9 | 52.32 | 76.3 | 89.5 | 57.4 | |
| Basel. P^{(0)}_{system}Model=DS-14B2026.04 | 470.9 | 28.96 | 87.3 | 94.9 | 71 | |
| TRACESModel=DS-14B, Threshold=0.92026.04 | 471.5 | 19.97 | 89.1 | 94.8 | 79.5 | |
| TRACESModel=DS-14B, Threshold=0.82026.04 | 485.9 | 18.47 | 89.9 | 95.1 | 81.2 | |
| TRACESModel=DS-14B, Threshold=0.72026.04 | 501.7 | 16.82 | 90.4 | 95.1 | 82.6 | |
| Basel. P^{(0)}_{user}Model=DS-14B2026.04 | 505.1 | 23.8 | 85.6 | 95.6 | 66.2 | |
| TRACESModel=DS-8B, Threshold=0.92026.04 | 506.1 | 47.18 | 77.5 | 93.6 | 53.4 | |
| Basel. P^{(0)}_{user}Model=DS-8B2026.04 | 525.8 | 45.13 | 77.1 | 91.7 | 57.9 | |
| TRACESModel=DS-14B, Threshold=0.62026.04 | 526.5 | 14.23 | 90.6 | 95.1 | 83.2 | |
| TRACESModel=DS-8B, Threshold=0.82026.04 | 544.8 | 43.15 | 79.2 | 93.7 | 55.9 | |
| TRACESModel=DS-14B, Threshold=0.52026.04 | 553.4 | 11.43 | 90.9 | 95.1 | 84.1 | |
| Basel. P^{(1)}_{system}Model=DS-8B2026.04 | 560.8 | 41.48 | 75.4 | 91.4 | 53.7 | |
| Basel. P^{(1)}_{system}Model=DS-14B2026.04 | 566.5 | 14.54 | 83.8 | 95.2 | 62.9 | |
| TRACESModel=DS-14B, Threshold=0.42026.04 | 567.8 | 9.92 | 90.9 | 95.1 | 84 | |
| TRACESModel=DS-8B, Threshold=0.72026.04 | 602.3 | 37.18 | 80.4 | 93.9 | 58.5 | |
| StandardModel=DS-14B2026.04 | 662.9 | — | 91 | 95.2 | 84.3 | |
| TRACESModel=DS-8B, Threshold=0.62026.04 | 681.8 | 28.85 | 81.5 | 94.2 | 61.4 | |
| TRACESModel=DS-8B, Threshold=0.52026.04 | 761.6 | 20.53 | 82.3 | 94.2 | 63.5 | |
| TRACESModel=QwQ-32B, Threshold=0.92026.04 | 777.9 | 62.52 | 88.9 | 95.9 | 76.6 | |
| TRACESModel=DS-8B, Threshold=0.42026.04 | 806.8 | 15.81 | 82.5 | 94.3 | 64.1 | |
| Basel. P^{(3)}_{system}Model=DS-8B2026.04 | 830.5 | 13.34 | 74.8 | 90.4 | 54.1 | |
| Basel. P^{(0)}_{system}Model=QwQ-32B2026.04 | 833.3 | 59.85 | 94 | 97.4 | 86.9 | |
| Basel. P^{(3)}_{system}Model=DS-14B2026.04 | 839.6 | -26.65 | 84.1 | 95.2 | 63.1 | |
| Basel. P^{(1)}_{system}Model=QwQ-32B2026.04 | 871.2 | 58.02 | 94.3 | 97.5 | 87.6 | |
| TRACESModel=QwQ-32B, Threshold=0.82026.04 | 895.9 | 56.84 | 90.9 | 95.9 | 80.4 | |
| StandardModel=DS-8B2026.04 | 958.3 | — | 82.9 | 94.3 | 65.1 | |
| Basel. P^{(0)}_{user}Model=QwQ-32B2026.04 | 988 | 52.4 | 95.2 | 96.8 | 93.7 | |
| TRACESModel=QwQ-32B, Threshold=0.72026.04 | 1,035.3 | 50.12 | 92.5 | 96.3 | 84.4 | |
| TRACESModel=QwQ-32B, Threshold=0.62026.04 | 1,270.1 | 38.81 | 93.6 | 96.3 | 87.9 | |
| Basel. P^{(3)}_{system}Model=QwQ-32B2026.04 | 1,387.3 | 33.16 | 93.5 | 97.4 | 85.5 | |
| TRACESModel=QwQ-32B, Threshold=0.52026.04 | 1,544.7 | 25.58 | 94.6 | 96.5 | 90.4 | |
| TRACESModel=QwQ-32B, Threshold=0.42026.04 | 1,685.9 | 18.79 | 94.8 | 96.6 | 91.2 | |
| StandardModel=QwQ-32B2026.04 | 2,075.7 | — | 95.3 | 96.5 | 93.4 |