List your company

Mathematical Reasoning on GSM8K (Accuracy & Average interaction number)

95.5AccuracyGPT-4 + PHP
82.29285.72189.1592.579Apr 19, 2023

Evaluation results

MethodLinks
GPT-4 + PHPPHP=true, Prompting Strategy=Complex CoT, Decoding Strategy=greedy2023.0495.52.0507
GPT-4PHP=false, Prompting Strategy=Complex CoT, Decoding Strategy=greedy2023.0494.9—
Previous SOTAPHP=false, Prompting Strategy=Complex CoT, Decoding Strategy=greedy2023.0492—
GPT-3.5 Turbo + PHPPHP=true, Prompting Strategy=Complex CoT, Decoding Strategy=greedy2023.0485.12.3426
GPT-3.5 TurboPHP=false, Prompting Strategy=Complex CoT, Decoding Strategy=greedy2023.0482.8—