WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct

Benchmark Model Rank Results
math-word-problem-solving-on-mathWizardMath-7B-V1.1#131Accuracy: 33.0Parameters (Billions): 7
math-word-problem-solving-on-mathWizardMath-70B-V1.0#154Accuracy: 22.7Parameters (Billions): 70
math-word-problem-solving-on-mathWizardMath-13B-V1.0#170Accuracy: 14.0Parameters (Billions): 13
math-word-problem-solving-on-mathWizardMath-7B-V1.0#178Accuracy: 10.7Parameters (Billions): 7
arithmetic-reasoning-on-gsm8kWizardMath-7B-V1.1#137Accuracy: 83.2Parameters (Billion): 7
arithmetic-reasoning-on-gsm8kWizardMath-70B-V1.0#151Accuracy: 81.6Parameters (Billion): 70
arithmetic-reasoning-on-gsm8kWizardMath-13B-V1.0#211Accuracy: 63.9Parameters (Billion): 13
arithmetic-reasoning-on-gsm8kWizardMath-7B-V1.0#228Accuracy: 54.9Parameters (Billion): 7