Deep Learning Training on AWS Inferentia

Yet another money-saving AI-model training hack

The topic of this post is AWS’s home-grown AI chip, AWS Inferentia — more specifically, the second-generation AWS Inferentia2. This is a sequel to our post from last year on AWS Trainium and joins a series of posts on the topic of dedicated AI accelerators. Contrary to the chips we have explored in our previous posts in the series, AWS Inferentia was designed for AI model inference and …

ai chip artificial intelligence aws aws inferentia chip cloud-ml-engine deep learning deep learning training home inferentia joins machine learning money saving series training trainium