Workshop Goals
Recently, machine learning methods like generative AI, Large Language Models (LLMs) for example, and reinforcement learning (RL), have shown remarkable ability in performing a wide range of tasks. Their applications in the design of computing stacks promise to revolutionize software and hardware code generation, system-level and architecture design, electronic design automation flow, test and verification, and beyond. RL methods have shown potential in a range of EDA optimization tasks. To harness this potential and spur research in this rapidly evolving area, we need datasets, benchmarks and competitions—much like Imagenet catalyzed the current era of deep learning, we also think community-wide datasets and competitions will do the same for ML for EDA.
Our goal is to bring together major stakeholders from various sectors (industry, academia, governmment) at a single venue for focused discussions and activities:
- Identify Gaps in Existing Datasets: Through discussions and analyses, pinpoint areas where current datasets are lacking and highlight the need for more comprehensive data.
Identify key areas in which the paucity of data is critical. - Solicit Community Datasets Development Efforts: Make an open call to the community—academia, industry and government—to contribute to datasets in these key areas, aiming to enrich the resources available for ML-aided design community.
- Organize Benchmarking Competitions: Decide on a plan to host benchmarking competitions to evaluate the quality of datasets and the effectiveness of designs generated using these datasets, fostering innovation and improvements.