Newly Funded – A Question-Answering Dataset for AI Agents in Modern Chip Design

Vijay Janapa Reddi (Harvard John A. Paulson School of Engineering and Applied Sciences) and the Edge Computing Lab aim to expand QuArch v0.5, the first dataset of 1,500 validated question-answer pairs designed to test and improve language models specifically for microprocessor design. The expansion will cover more advanced topics and establish a public leaderboard, creating a critical resource for AI-driven hardware design that has already demonstrated up to 8% accuracy improvements when used for fine-tuning models.

This project, and others like it, is made possible by the HDSI Special Projects Fund