Ask-EDA: A Design Assistant Empowered by LLM, Hybrid RAG and Abbreviation De-hallucination

TL;DR

Ask-EDA addresses the challenge of locating relevant, up-to-date design information by integrating LLMs with a hybrid retrieval-augmented generation (RAG) system and abbreviation de-hallucination (ADH). It fuses dense and sparse retrieval through Reciprocal Rank Fusion to produce context for the LLM, and augments prompts with a dedicated abbreviation dictionary to mitigate term-related hallucinations. Evaluations on three IBM-design-domain datasets show significant recall gains from hybrid RAG (over 40% on q2a-100 and over 60% on cmds-100) and notable improvements from ADH on abbr-100. A Slack-based chat interface demonstrates practical deployment, enabling 24x7 design guidance with improved accuracy and contextual relevance for engineers.

Abstract

Electronic design engineers are challenged to find relevant information efficiently for a myriad of tasks within design construction, verification and technology development. Large language models (LLM) have the potential to help improve productivity by serving as conversational agents that effectively function as subject-matter experts. In this paper we demonstrate Ask-EDA, a chat agent designed to serve as a 24x7 expert available to provide guidance to design engineers. Ask-EDA leverages LLM, hybrid retrieval augmented generation (RAG) and abbreviation de-hallucination (ADH) techniques to deliver more relevant and accurate responses. We curated three evaluation datasets, namely q2a-100, cmds-100 and abbr-100. Each dataset is tailored to assess a distinct aspect: general design question answering, design command handling and abbreviation resolution. We demonstrated that hybrid RAG offers over a 40% improvement in Recall on the q2a-100 dataset and over a 60% improvement on the cmds-100 dataset compared to not using RAG, while ADH yields over a 70% enhancement in Recall on the abbr-100 dataset. The evaluation results show that Ask-EDA can effectively respond to design-related inquiries.

Figures (5)

• Figure 1: Document ingestion into hybrid database.
• Figure 2: Pipeline to produce a response based on user query.
• Figure 3: Results on q2a-100 and cmds-100 datasets.
• Figure 4: Evaluation results on abbreviation de-hallucination.
• Figure A.1: Example screenshot of Slack interface.