Cloud Versions Save

The TensorFlow Cloud repository provides APIs that will allow to easily go from debugging and training your Keras and TensorFlow code in a local environment to distributed training in the cloud.

0.1.6

3 years ago
  • New module CloudTuner - Implementation of a library for hyperparameter tuning that is built into the KerasTuner and creates a seamless integration with Cloud AI Platform Optimizer as a backend to get suggestions of hyperparameters and run trials.
  • New application Monitoring - TensorFlow extension that exports its metrics to Stackdriver backend, allowing users to monitor the training and inference jobs in real time.
  • New experimental project cloud_fit - an experimental module that enables training keras models on Cloud AI Platform Training by serializing the model, and datasets for remote execution.
  • Small bug fixes

0.1.5

3 years ago
  • Restructuring of source code for new projects
  • Multi-file code example
  • Integration test example
  • Small bug fixes

0.1.4

3 years ago
  • New API remote() to detect if currently in a remote cloud env.
  • CI using Github Action.
  • Updated README.
  • Some minor bug fixes.

0.1.3

3 years ago

New features

  • Support for single node Keras tuner workflow.
  • Support for TPU training.

Fixes

  • Fixed docker build decode errors.
  • Default to Py3 for TF docker images.

Others

  • New colab notebook example.
  • New Auto Keras example.
  • Improved ReadMe docs.
  • Improved error messages.

0.1.2

3 years ago
  • Support for passing colab notebook as entry_point.
  • Support for cloud docker build and colab workflow.
  • Support for log streaming in colab

0.1.1

3 years ago
  • Detailed README with setup instructions and examples
  • Support for running run API from within a python script which contains a Keras model

0.1.0

3 years ago

First release

  • Initial release with support for running a python script on GCP
  • Examples for basic workflows in Keras