Response Performance Evaluations of ChatGPT Models on Large Language Model Frameworks

In recent years, LLMs have become increasingly popular. These models are being used in various fields and applications. While the increase in the number of models is positive in one aspect, it may not be easy to determine which LLM and LLM framework to use for a specific application. Finding a suitable model for its performance and response time may require many trials. In this study, the success of answering questions was measured using a combination of two different LLMs developed by OpenAI (GPT-3.5 & GPT-4) and five different methods, including LLM frameworks, on three different datasets. Additionally, the established structures were compared in terms of time. Consequently, this study establishes a base for developers planning to use these LLM frameworks to examine five different structures, laying the groundwork for future studies.