Skip to content

Latest commit

 

History

History
125 lines (84 loc) · 3.16 KB

RELEASE.md

File metadata and controls

125 lines (84 loc) · 3.16 KB

Current Version (Still in Development)

  • Add notes for next release here.

Release 0.1.16

  • Dump Python stack when Segmentation fault occurs during remote environment.

Release 0.1.15

  • Update API documentation.

Release 0.1.14

  • Remove upper bound dependency on tensorflow_datasets.
  • Fix dataset auto shard policy.
  • Fix machine_config for DistributingCloudTuner.
  • Additional updates in samples and documentation.

Release 0.1.13

  • Upgrad the default Cloud Build machine to N1_HIGHCPU_8

Release 0.1.12

  • Support for launching concurrent tuning jobs.
  • Code samples for running concurrent tuning jobs.
  • Project setup instruction notebooks.
  • Updated privacy policy notice for telemetry.
  • Small bug fixes in CloudTuner.

Release 0.1.11

  • Telemetry additions
  • Address test failures/flakiness

Release 0.1.10

  • Py 3.5 support removed.
  • Small bug fixes.

Release 0.1.9

  • Added Kaggle integration.

Release 0.1.8

  • cloud_fit now moved to a sub-module under Tuner
  • HParams plugin integration with DistributingCloudTuner
  • Added integration tests
  • Small bug fixes.

Release 0.1.7

  • cloud_fit uses pickle instead of cloudpickle.
  • Better integration tests checking for job status.
  • Small bug fixes.

Release 0.1.6

  • 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.

Release 0.1.5

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

Release 0.1.4

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

Release 0.1.3

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 ReadMedocs.
  • Improved error messages.

Release 0.1.2

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

Release 0.1.1

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

Release 0.1.0

First release

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