Neural networks have grown exponentially in recent years, from 2018 state-of-the-art neural networks of 100 million parameters to the famous GPT-3 with 175 billion parameters. However, this Grand ML Demand Challenge must be addressed by making substantial improvements — an order of magnitude or more — across a broad spectrum of multiple different components. Read Sean's blog for a written version of this Hot Chips 34 talk. It gives a deep dive into the Cerebras hardware to show you how our revolutionary approaches in core architecture, scale-up, and scale-out are designed to meet this ML demand: https://www.cerebras.net/blog/cerebra...

Learn more about Cerebras: https://cerebras.net

#deeplearning #ai #artificialintelligence #hotchips #hotchips34 #hc34