GSoC 2024 Contributions ‐ Palaniappan R - The-OpenROAD-Project/ORAssistant GitHub Wiki

The key contributions from Palaniappan R’s work as a GSoC contributor on the OpenROAD project are:

1. Project Objective: The project aimed to enhance user experience within OpenROAD by developing an ORAssistant chatbot using Large Language Models (LLMs). The chatbot helps users resolve FAQs and common issues, particularly around installation and command usage, reducing the number of support tickets on platforms like GitHub.

2. RAG Architecture: Palaniappan employed Retrieval-Augmented Generation (RAG) to combine LLMs with external factual data from a knowledge base. This method enhances the chatbot’s ability to generate informed and relevant responses to user queries.

3. Knowledge Base Development: The knowledge base consists of official documentation from OpenROAD and related tools, along with annotated data from GitHub discussions. The entire data-building process has been automated to keep the knowledge base up-to-date.

4. Tool-Based Architecture: The system uses a tool-based approach where different tools are applied to various vector databases, allowing efficient retrieval of documents. Queries are rephrased to retain context based on chat history.

5. Future Plans: Palaniappan plans to expand the chatbot’s functionality by adding flow script generation and supporting more tools like KLayout. ORAssistant will also be integrated into OpenROAD’s CLI and GUI interfaces.