Use ChatGPT With Custom Data Using Embeddings
One of the ways to use ChatGPT to provide responses on our custom data is using embeddings. Though its not actual fine tuning of any of the OpenAI models but using word embeddings we will be able to get responses from ChatGPT on our custom data. Again these word embeddings and responses will be created and provided using OpenAI models.
Make sure you have valid OpenAI API key to leverage its free limit of $5. It will be required to do this tutorial. Also in the video I have mentioned that I am not sure how it will work with other file formats apart from .pdf file but later on I have put .pdf, .csv, .md, .txt files in the same folder, and it has worked quite well.
All the scripts used in this video is available on the following repository.
https://gitlab.com/Nayan.1989/chatgpt...
Also, in case if you want to use other embedding algorithms or wants to know more about Langhian and Llama index packages use below URLs.
https://python.langchain.com/docs/get...
https://gpt-index.readthedocs.io/en/l...