Inspiration
Extracting information from pdfs and large documents was very hectic task for many of us if the pdf is of more than 50 pages and contain very large textual information hence, I thought of a tool which can solve this issue.
What it does
It parses the pdf into smaller chunks and stores it as vectors for providing context to the Gemini model and when user asks any question it try to answer that with the information provided from the document.
How I built it
I have used streamlit to fire up the User interface and used Lang chain to utilize the Faiss indexing to create vectors from the pdf and google.generativeai package for gemini.
Challenges I ran into
Main challenge was to effectively parse the large pdf and provide relevant information context to the AI model.
Accomplishments that I am proud of
This tool is open source with 43 forks and counting and This app has had 1,964 unique viewers on streamlit which shows how this tool is benefiting others
What I learned
how powerful Gemini is in understanding context provided and utilization of Faiss indexing from langchain.
What's next for Gemini MultiPdf chat
I am working to create external API with this tool to extend its integration with other Applications.
Log in or sign up for Devpost to join the conversation.