Diving Deep into ChatGPT: My Journey from Awe to Personal Implementation

When I first encountered ChatGPT, I was nothing short of amazed. The seamless interaction, the nuanced understanding, and the vast knowledge it demonstrated made me wonder: How is this technological marvel even possible?

My curiosity led me on a quest to understand the models that power such chatbots. I immediately found Hugging Face, a platform teeming with a plethora of models dedicated to Natural Language Processing (NLP). While many of these models were adept at generative text tasks, not all were tailored for dynamic chat-based interactions.

The real turning point came when I stumbled upon an open-source project named FastChat. It promised a swift deployment of NLP models, including the intriguing Vicuna (Llama based model from Meta), into a graphical chat interface. With FastChat, I embarked on a journey to deploy Llama, and soon enough, I had crafted my very own chatbot, "akin" to ChatGPT!

Disclaimer: this blog article is a totally true story about my recent study of ChatGPT and NLP and it was written at 80% by chatGPT4. :) Most images are also created by Chatgpt :)