Mathematical Reasoning on OlympiadBench (Accuracy, Pass@8)
60AccuracyQwen3-8B (Thinking)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Qwen3-8B (Thinking)Model=Qwen3-8B, Variant=Thinking2026.05 | 60 | 69 | |
| Qwen3-1.7B (Thinking)Model=Qwen3-1.7B, Variant=Thinking2026.05 | 52.5 | 63 | |
| Qwen3-1.7B + Guided (ρ¯ ≈ 0.16)Model=Qwen3-1.7B, Intervention Rate (rho)=≈ 0.162026.05 | 48.9 | 67 | |
| Qwen3-0.6B + Guided (ρ¯ ≈ 0.13)Model=Qwen3-0.6B, Intervention Rate (rho)=≈ 0.132026.05 | 42.5 | 65 | |
| Qwen3-0.6B (Thinking)Model=Qwen3-0.6B, Variant=Thinking2026.05 | 34.3 | 54 | |
| Qwen3-1.7B + Guided (ρ¯ ≈ 0.06)Model=Qwen3-1.7B, Intervention Rate (rho)=≈ 0.062026.05 | 33.2 | 64 | |
| Qwen3-0.6B + Guided (ρ¯ ≈ 0.04)Model=Qwen3-0.6B, Intervention Rate (rho)=≈ 0.042026.05 | 20.2 | 49 | |
| Qwen3-1.7B-BaseModel=Qwen3-1.7B, Variant=Base2026.05 | 19.2 | 41 | |
| Qwen3-0.6B-BaseModel=Qwen3-0.6B, Variant=Base2026.05 | 6.8 | 21 |