AI Compute Investment
Nathan and Vivek discuss the relative compute investment in AI models, highlighting the small token numbers compared to base models and the impact of chip configurations on training speed. The conversation sheds light on the one to 10% realm of training the base model and the significance of access to different compute resources.In this clip
From this podcast
The Cognitive Revolution: How AI Changes Everything
Google’s Multimodal Med-PaLM with Vivek Natarajan and Tao Tu
Related Questions