Inspiration
We are both passionate students and love the topics of data science and AI. Recently our favorite professor Dr.Che recommend us a book on deep learning. We wanted a study buddy to learn more about this book and hence this project.
What it does
This takes in text books as a pdf, places it in a vector database, and then queries using a LLM. It also has the capacity to take and save notes for the user on request.
How we built it
We used streamlit for the UI. Then LLama Index to process the data into a database and manage LLM interaction. Then we used OpenAI for the embeddings. this also uses the transformers library for a summary function in addition to saving notes.
Challenges we ran into
We initially had trouble getting the LLM to integrate with the database. These issue include windows firewall blocking locally ran LLM and OpenAI required placement of the api key in environment variables. In addition the program is resource intensive so only one of our computers could run the program.
Accomplishments that we're proud of
We have created a rags chatbot before and we think we might even use this project in future for our own studies. The ability to have a custom bot for classes is amazing.
What we learned
We learned openai can be a pain and more about LLMs. We also learned stream lit and rags chatbot.
What's next for Study Buddy
We would like to add more functionality for different types of file formats that LLama index supports. We would also like to switch to a different llm so this could be ran locally.
btw for the code use the failed code branch. we thought the code was bad and then we realize it was good and used it as main.
Log in or sign up for Devpost to join the conversation.