List your company

Mathematical Reasoning on MATH 500 (Accuracy Variance, Consistency, Token Variance)

79.7Mean AccuracyGemma-3-12B-Instruct
22.70837.50452.367.096Dec 2, 2025

Evaluation results

MethodLinks
Gemma-3-12B-InstructSetting=Optimized2025.1279.70.001655.4144,471.24
Qwen2.5-7B-InstructSetting=Optimized2025.1268.60.00735.4122,866.7
Gemma-3-12B-InstructSetting=Random2025.12670.00249.5153,144.84
Qwen2.5-7B-InstructSetting=Random2025.1258.50.00735.4305,133.92
Llama-3.1-8B-InstructSetting=Optimized2025.1228.50.0261.11,547,849
Llama-3.1-8B-InstructSetting=Random2025.1224.90.0152.71,512,173.79