Enhancing Automotive PDF Chatbots: A Graph RAG Approach with Custom Function Calling for Locally Deployed Ollama Models
Fei Liu, Huanhuan Ren, Yu Guan, Na Li · 2024
This research explores state-of-the-art retrieval augmented generation (RAG) methodologies utilizing a locally hosted Ollama model to address the rising demand for streamlined offline PDF chatbots in the automotive industry. In this paper we present a Langchain-based optimization of the a forementioned framework for modeling all input text from automotive documents. For example, using JoinChpy which use specialized finetuning to a base model where it is fine-tuned adding specialized modifications, the first on PDF and retrieval algorithms as well as context compression strictly for automotive literature We define embedding pipeline classes and graph RAG agents.