Mathematical Reasoning on AMC (Accuracy, C_mem, C_time)
0.1C_mem (Ratio)T2
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| T2Backbone model=Qwen3-4B2026.01 | 0.1 | 0.52 | 0.8187 | |
| ACBackbone model=Qwen3-4B2026.01 | 0.14 | 2.67 | 0.8173 | |
| T2Backbone model=Llama3.1-8B2026.01 | 0.16 | 0.37 | 0.27 | |
| ACBackbone model=Llama3.1-8B2026.01 | 0.21 | 2.27 | 0.2673 | |
| T2Backbone model=Deepseek-R1-Distill-Qwen-7B2026.01 | 0.22 | 0.48 | 0.8597 | |
| T2Backbone model=Qwen2.5-7B2026.01 | 0.23 | 0.44 | 0.4902 | |
| ESCBackbone model=Qwen3-4B2026.01 | 0.23 | 1.84 | 0.8177 | |
| ESCBackbone model=Deepseek-R1-Distill-Qwen-7B2026.01 | 0.27 | 1.72 | 0.8603 | |
| ACBackbone model=Qwen2.5-7B2026.01 | 0.28 | 2.32 | 0.4867 | |
| DSCBackbone model=Qwen3-4B2026.01 | 0.28 | 1.71 | 0.8132 | |
| DSCBackbone model=Qwen2.5-7B2026.01 | 0.31 | 2.16 | 0.4885 | |
| ACBackbone model=Deepseek-R1-Distill-Qwen-7B2026.01 | 0.32 | 2.77 | 0.8582 | |
| DeepConfBackbone model=Qwen3-4B2026.01 | 0.32 | 3.27 | 0.8165 | |
| DSCBackbone model=Llama3.1-8B2026.01 | 0.33 | 2.18 | 0.2656 | |
| ESCBackbone model=Qwen2.5-7B2026.01 | 0.34 | 1.9 | 0.4877 | |
| DeepConfBackbone model=Qwen2.5-7B2026.01 | 0.38 | 2.67 | 0.4767 | |
| DSCBackbone model=Deepseek-R1-Distill-Qwen-7B2026.01 | 0.39 | 2.89 | 0.8128 | |
| DeepConfBackbone model=Llama3.1-8B2026.01 | 0.48 | 2.61 | 0.2682 | |
| ESCBackbone model=Llama3.1-8B2026.01 | 0.53 | 2.03 | 0.2656 | |
| DeepConfBackbone model=Deepseek-R1-Distill-Qwen-7B2026.01 | 0.58 | 3.56 | 0.8552 | |
| Std-PT.Backbone model=Qwen2.5-7B2026.01 | 1 | 1 | 0.4936 | |
| Std-PT.Backbone model=Llama3.1-8B2026.01 | 1 | 1 | 0.2737 | |
| Std-PT.Backbone model=Deepseek-R1-Distill-Qwen-7B2026.01 | 1 | 1 | 0.8678 | |
| Std-PT.Backbone model=Qwen3-4B2026.01 | 1 | 1 | 0.826 |