How to use Jupyter notebooks, langchain and Kaggle.com to create an AI chatbot on any topic
In this video, I do a deep dive on the two Jupyter notebooks which I built as part of my office oracle project.
Both notebooks are now open source:
I talk through what I learned, why Jupyter notebooks were such a handy tool for getting my data quality to where I needed it to be, before worrying about application logic.
I also demonstrate langchain DocumentLoaders, how to store secrets in Jupyter notebooks when open sourcing them, and much more.
What's involved in building an AI chatbot that is trained on a custom corpus of knowledge? In this video I breakdown the data preparation, training and app development components and explain why Jupyter notebooks were such a handy tool while creating this app and tweaking my model.
App is open source at https://github.com/zackproser/office-... and a demo is currently available at https://office-oracle.vercel.app.