TPU Virtual Machines (VMs) for artificial intelligence applications are now generally available on Google Cloud.
A new TPU embedding API is included in the general availability version, which Google Cloud claims helps speed up large-scale machine learning-based ranking and recommendation workloads.
Google Cloud claims that embedding acceleration with Cloud TPU can help organizations save money on ranking and recommendation use-cases that typically rely on expensive deep neural network-based techniques. The TPU VMs also allow input data pipelines to be run directly on the TPU hosts, according to Google Cloud.
Read More: https://www.zdnet.com/article/google-announces-cloud-tpu-virtual-machines-for-ai-workloads/
Check Out The New Enterprisetalk Podcast. For more such updates follow us on Google News Enterprisetalk News.
