Skip to content

Latest commit

 

History

History
19 lines (15 loc) · 1.92 KB

README.md

File metadata and controls

19 lines (15 loc) · 1.92 KB

Tensor Processing Units (TPUs) are Google’s custom-developed ASICs used to accelerate machine-learning workloads. You can run your training jobs on Cloud Machine Learning Engine, using Cloud TPU. Cloud ML Engine provides a job management interface so that you don't need to manage the TPU yourself. Instead, you can use the Cloud ML Engine jobs API in the same way as you use it for training on a CPU or a GPU.

For TPU training, we recommend users to use the sample from Cloud TPU Demos. And you can follow the doc to get started with TPU training on CloudML Engine.

Samples:

  • Tensorflow: ResNet - Using the ResNet-50 dataset with Cloud TPUs on ML Engine.
  • Tensorflow: HP Tuning - ResNet - How to run hyperparameter tuning jobs on Cloud Machine Learning Engine with Cloud TPUs using TensorFlow's tf.metrics.
  • Tensorflow: Hypertune - ResNet - How to run hyperparameter tuning jobs on Cloud Machine Learning Engine with Cloud TPUs using the cloudml-hypertune package.
  • Tensorflow: Templates - A collection of minimal templates that can be run on Cloud TPUs on Compute Engine, Cloud Machine Learning, and Colab.

If you’re looking for samples for how to use Cloud TPU, check out the guides here.

Note: These guides do not use ML Engine