Mathematical Problem Solving on AIME 25 (Time/Efficiency)
15.69Average TimeCoT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CoTModel=Qwen3-1.7B2025.07 | 15.69 | 9.91 | — | |
| LLM As a CriticModel=Qwen3-1.7B, Variant=Per-Step Scale2025.07 | 19.62 | 15.46 | — | |
| LLM As a CriticModel=Qwen3-1.7B, Variant=Avg uncertainty2025.07 | 21.13 | 16.94 | 1.48 | |
| CoTModel=Qwen3-4B2025.07 | 21.9 | 15.63 | — | |
| Guided SearchModel=Qwen3-1.7B, Variant=MUR2025.07 | 25.11 | 16.8 | -35.81 | |
| LLM As a CriticModel=Qwen3-1.7B, Variant=MUR2025.07 | 25.4 | 17.75 | 2.29 | |
| Guided SearchModel=Qwen3-1.7B, Variant=Avg uncertainty2025.07 | 31.67 | 22.41 | -30.2 | |
| ϕ-DecodingModel=Qwen3-1.7B, Variant=MUR2025.07 | 36.6 | 21.81 | -21.57 | |
| ϕ-DecodingModel=Qwen3-1.7B, Variant=Avg uncertainty2025.07 | 38.61 | 23.55 | -19.83 | |
| LLM As a CriticModel=Qwen3-1.7B, Variant=SMART2025.07 | 44.93 | 26.66 | 11.2 | |
| Guided SearchModel=Qwen3-4B, Variant=MUR2025.07 | 46.5 | 34.11 | -60.24 | |
| ϕ-DecodingModel=Qwen3-1.7B, Variant=SMART2025.07 | 50.21 | 34.45 | -8.93 | |
| Guided SearchModel=Qwen3-1.7B, Variant=SMART2025.07 | 55.03 | 37.37 | -15.24 | |
| LLM As a CriticModel=Qwen3-4B, Variant=MUR2025.07 | 56.4 | 38.41 | -20.81 | |
| LLM As a CriticModel=Qwen3-4B, Variant=SMART2025.07 | 61.63 | 37.26 | -21.96 | |
| ϕ-DecodingModel=Qwen3-1.7B, Variant=Per-Step Scale2025.07 | 62.56 | 43.38 | — | |
| CoTModel=Qwen3-8B2025.07 | 65.92 | 45.38 | — | |
| LLM As a CriticModel=Qwen3-4B, Variant=Per-Step Scale2025.07 | 70.9 | 59.22 | — | |
| LLM As a CriticModel=Qwen3-4B, Variant=Avg uncertainty2025.07 | 71.17 | 47.86 | -11.36 | |
| Guided SearchModel=Qwen3-1.7B, Variant=Per-Step Scale2025.07 | 73.34 | 52.61 | — | |
| ϕ-DecodingModel=Qwen3-4B, Variant=Avg uncertainty2025.07 | 74.53 | 59.94 | -35.07 | |
| Guided SearchModel=Qwen3-4B, Variant=Avg uncertainty2025.07 | 76.8 | 62.78 | -31.57 | |
| ϕ-DecodingModel=Qwen3-4B, Variant=MUR2025.07 | 92.73 | 58.06 | -36.95 | |
| LLM As a CriticModel=Qwen3-8B, Variant=MUR2025.07 | 100.02 | 72.22 | -6.65 | |
| LLM As a CriticModel=Qwen3-8B, Variant=SMART2025.07 | 107.29 | 77.09 | -1.78 | |
| LLM As a CriticModel=Qwen3-8B, Variant=Per-Step Scale2025.07 | 108.81 | 78.87 | — | |
| LLM As a CriticModel=Qwen3-8B, Variant=Avg uncertainty2025.07 | 122.64 | 78.8 | -0.07 | |
| ϕ-DecodingModel=Qwen3-4B, Variant=Per-Step Scale2025.07 | 123.57 | 95.01 | — | |
| Guided SearchModel=Qwen3-4B, Variant=Per-Step Scale2025.07 | 125.8 | 94.35 | — | |
| ϕ-DecodingModel=Qwen3-4B, Variant=SMART2025.07 | 137.2 | 91.93 | -3.08 | |
| Guided SearchModel=Qwen3-8B, Variant=MUR2025.07 | 172.6 | 113.52 | -93.68 | |
| Guided SearchModel=Qwen3-8B, Variant=Avg uncertainty2025.07 | 179.73 | 114.81 | -92.39 | |
| Guided SearchModel=Qwen3-4B, Variant=SMART2025.07 | 181.07 | 96.38 | 2.03 | |
| Guided SearchModel=Qwen3-8B, Variant=SMART2025.07 | 207.56 | 137.16 | -70.04 | |
| ϕ-DecodingModel=Qwen3-8B, Variant=MUR2025.07 | 253.98 | 160.16 | -79.94 | |
| ϕ-DecodingModel=Qwen3-8B, Variant=Avg uncertainty2025.07 | 267.33 | 163.97 | -76.13 | |
| ϕ-DecodingModel=Qwen3-8B, Variant=SMART2025.07 | 306.03 | 207.32 | -32.78 | |
| Guided SearchModel=Qwen3-8B, Variant=Per-Step Scale2025.07 | 338.58 | 207.2 | — | |
| ϕ-DecodingModel=Qwen3-8B, Variant=Per-Step Scale2025.07 | 390.91 | 240.1 | — |