Mathematical Reasoning on AMC23 (Accuracy, Length)
90.83AccuracyStandardp
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| StandardpBackbone=DeepSeek-R1-7B2026.02 | 90.83 | 6,224.23 | |
| S3-CoTBackbone=DeepSeek-R1-7B2026.02 | 90.83 | 5,715.74 | |
| EffReaBackbone=DeepSeek-R1-7B2026.02 | 88.33 | 4,649.27 | |
| LASERDEBackbone=DeepSeek-R1-7B2026.02 | 88.33 | 2,766.08 | |
| AutoTHINKBackbone=DeepSeek-R1-7B2026.02 | 88.33 | 3,846.43 | |
| C3oTBackbone=DeepSeek-R1-7B2026.02 | 87.5 | 6,820.08 | |
| S3-CoTscBackbone=DeepSeek-R1-7B2026.02 | 87.5 | 5,654.8 | |
| LC-R1Backbone=DeepSeek-R1-7B2026.02 | 85 | 2,788.19 | |
| EfficientpBackbone=DeepSeek-R1-7B2026.02 | 83.33 | 6,062.81 | |
| ShorterBetterBackbone=DeepSeek-R1-7B2026.02 | 66.67 | 1,613.66 | |
| CoT-ValveBackbone=DeepSeek-R1-7B2026.02 | 65 | 1,940.33 | |
| C3oTBackbone=Qwen2.5-7B2026.02 | 51.67 | 788.6 | |
| S3-CoTscBackbone=Qwen2.5-7B2026.02 | 46.67 | 687.33 | |
| S3-CoTBackbone=Qwen2.5-7B2026.02 | 45.83 | 678.8 | |
| TokenSkipBackbone=Qwen2.5-7B2026.02 | 45 | 766.95 | |
| EfficientpBackbone=Qwen2.5-7B2026.02 | 44.17 | 573.44 | |
| StandardpBackbone=Qwen2.5-7B2026.02 | 43.33 | 846.49 | |
| CoT-ValveBackbone=Qwen2.5-7B2026.02 | 37.5 | 900.23 |