ChatGPT, a general-purpose AI model launched on November 30, 2022 by OpenAI has come as a very beneficial technology in providing major outbreaks in education to health sectors to our daily lifestyle. ChatGPT have been trained on the OpenAI's Generative Pre-trained Transformer 3.5 model and have been developed to perform well on conversational Natural Language Processing(NLP) tasks. ChatGPT also leverages other models developed by OpenAI such as clip, dall-e and codex to optimize its performance and to provide outputs in different forms/mediums.

Taking the amount of beneficiary resources that ChatGPT provides, knowing about its fundamental technologies will serve you to understand even more about its mechanisms. ChatGPT has been mainly trained on the Large Language Models(LLMs), which uses deep learning algorithms to process, and understand natural language. They are coined as “large” considering the amount of parameters they contain which are about 175 billion in the case of ChatGPT. The fundamental deep learning algorithm used in ChatGPT are neural networks. Neural Networks are the machine learning models which mimics the complex functioning of the human brain by using nodes to learn from data.

Large Language Models are specifically based on a type of neural network architecture known as Transformers, which are capable of processing sequential data such as text or speech using attention mechanisms. Attention mechanisms are capable of pinpointing and focusing on the important aspects of input or context for better computation. Alongside LLMs, ChatGPT also uses a learning mechanism known as reinforcement learning from the response available from the users. Reinforcement Learning is a type of un-supervised iterative learning method that allows machines to learn from responses and make better predictions or provide better output in the future.

Along the models, an equally crucial part of training ChatGPT is data. Data for training ChatGPT have been collected from the available data on the internet, licensed data from the third parties, from the users and the human trainers which have contributed in an effective optimization and implementation of models. So, these were the major resources that have contributed to the creation of ChatGPT.