The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Stable Diffusion and LLMs at the Edge with Jilei Hou - #633
10 snips
Jun 12, 2023 Jilei Hou, VP of Engineering at Qualcomm Technologies, specializes in information theory and signal processing. He discusses the rise of generative AI and the advancement of deploying these models on edge devices. Challenges like model size and inference latency are highlighted, alongside solutions like quantization for optimizing performance. The conversation also dives into local optimization techniques that drastically reduce computation times for diffusion models. Jilei emphasizes the need for multimodal models, reshaping AI interactions and future innovations.
AI Snips
Chapters
Transcript
Episode notes
Information Theory and AI Synergy
- Information theory and AI have a natural synergy, especially at Qualcomm.
- Qualcomm focuses on applying machine learning algorithms to signal processing, a natural area for information theory.
Qualcomm AI Research Formation
- Qualcomm AI Research was formed after several AI startup acquisitions around 2018.
- The group focuses on on-device AI power efficiency and personalization, with an early focus on quantization research.
Generative AI Breakthrough
- Generative AI, while newly popular, relies on established generative modeling techniques.
- The recent breakthrough comes from efficiently feeding these models with internet-scale data and massive model capacity.