In this task we will analyze the SEC 10K reports of NVIDIA over the last 5 years and derive insights and conclude whether the company grew over the years or not. We will be using the RAG(Retrieval Augmented Generation) approach for this task. We extract the data from SEC's Official Website using the API service provided by SEC-API. We extract only some sections of the filings in html format. We will be using 🦙 llama parse to parse the data. Since llama parse only accepts pdf files as input we will be converting the html files into pdf. After that we will merge all the parsed data into one single .MD file. We will be using RAG approach. We first split the data into small chunks using RecursiveCharacterTextSplitter(), Then we embed the data using 'BAAI/bge-base-en-v1.5' model. We then use qdrant to create a vector database which also contains a vector search engine for RAG. We use flashrankrerank to rerank the data. Finally we will be using llama 3.1 llm with the help of Groq API to derive insights.
