Session

DescriptionLarge Language Models (LLMs) are fundamental machine learning models that use advanced algorithms to comprehend, process, and optimize natural language. These models are typically trained on vast amounts of data to learn patterns and use that in the processing and optimization phase. Due to the advent of LLMs, computing platforms are revolutionizing. This includes traditional Semiconductors as well as computing at both Cloud and Edge. In this session, we propose to have three complementary technologies that will cover this spectrum. In the first talk, the authors will discuss the LLM techniques that can be deployed for High-Level Synthesis Optimization and Verification. This approach enhances the efficiency and accessibility of AI accelerator development and serves as a bridge between AI algorithmic advancements and hardware innovation. The researchers in the second talk believe in and observe the importance of High-quality HDL in the design process. They will present that careful selection, filtering, and augmentation of data across HDLs can yield powerful LLM (and other) models that surpass current state-of-the-art models. This technique can be used to train various models for circuit design related activities. The final talk is about leveraging the extraordinary capabilities of Large Language Models to revolutionize AI accelerator design and enhance its accessibility.

Organizer: Sabya Das, Synopsys