List your company

Mathematical Reasoning on GSM8K (Accuracy, Pass@8)

97.8AccuracyQwen3-8B (Thinking)
27.18445.51763.8582.183May 16, 2026

Evaluation results

MethodLinks
Qwen3-8B (Thinking)Model=Qwen3-8B, Variant=Thinking2026.0597.898
Qwen3-1.7B (Thinking)Model=Qwen3-1.7B, Variant=Thinking2026.0590.997
Qwen3-1.7B + Guided (ρ¯ ≈ 0.16)Model=Qwen3-1.7B, Intervention Rate (rho)=≈ 0.162026.0588.998
Qwen3-0.6B (Thinking)Model=Qwen3-0.6B, Variant=Thinking2026.0581.694
Qwen3-0.6B + Guided (ρ¯ ≈ 0.13)Model=Qwen3-0.6B, Intervention Rate (rho)=≈ 0.132026.0580.499
Qwen3-1.7B-BaseModel=Qwen3-1.7B, Variant=Base2026.0575.493
Qwen3-1.7B + Guided (ρ¯ ≈ 0.06)Model=Qwen3-1.7B, Intervention Rate (rho)=≈ 0.062026.057297
Qwen3-0.6B + Guided (ρ¯ ≈ 0.04)Model=Qwen3-0.6B, Intervention Rate (rho)=≈ 0.042026.0556.596
Qwen3-0.6B-BaseModel=Qwen3-0.6B, Variant=Base2026.0529.982