Mathematical Reasoning on AMC 23 (Accuracy, Pass@8)
90.9AccuracyQwen3-8B (Thinking)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen3-8B (Thinking)Model=Qwen3-8B, Variant=Thinking2026.05 | 90.9 | 100 | |
| Qwen3-1.7B (Thinking)Model=Qwen3-1.7B, Variant=Thinking2026.05 | 80 | 95 | |
| Qwen3-1.7B + Guided (ρ¯ ≈ 0.16)Model=Qwen3-1.7B, Intervention Rate (rho)=≈ 0.162026.05 | 71.9 | 97.5 | |
| Qwen3-0.6B + Guided (ρ¯ ≈ 0.13)Model=Qwen3-0.6B, Intervention Rate (rho)=≈ 0.132026.05 | 58.4 | 90 | |
| Qwen3-1.7B + Guided (ρ¯ ≈ 0.06)Model=Qwen3-1.7B, Intervention Rate (rho)=≈ 0.062026.05 | 50.6 | 95 | |
| Qwen3-0.6B (Thinking)Model=Qwen3-0.6B, Variant=Thinking2026.05 | 46.9 | 80 | |
| Qwen3-0.6B + Guided (ρ¯ ≈ 0.04)Model=Qwen3-0.6B, Intervention Rate (rho)=≈ 0.042026.05 | 35.9 | 77.5 | |
| Qwen3-1.7B-BaseModel=Qwen3-1.7B, Variant=Base2026.05 | 28.1 | 65 | |
| Qwen3-0.6B-BaseModel=Qwen3-0.6B, Variant=Base2026.05 | 13.4 | 42.5 |