Datalab 1.2.20180109

Changes:
Upgrade to Tensorflow 1.4.1
NVidia P100 GPUs are now supported on GPU images (due to upgrade to CUDA 6)
Various Bugfixes.

Datalab 1.2.20171023

Bug fixes:
Upgraded Tensorflow to version 1.2.1
Fixed issues with BigQuery Results table not being displayed.
Various other bug fixes.

Datalab 1.2.20170818

Features and improvements:

• Python 3 Support! Now you have a drop down list to pick PY2/PY3! (#1344, #1351, #1398).
• Auto Idle Shutdown: If not active for certain time the VM is shut down automatically to save cost (#1395, #1425, #1502, #1506, #1573, #1578).
• Faster Datalab startup: Start the Datalab container using a systemd service (#1531).
• Package updates:
  • TensorFlow 1.0 -> 1.2.
  • Apache Beam 0.6 -> 2.0 (#1478).
  • Updated Scikit-learn.
  • Added LIME.
  • Addd scikit-image (#1529) .

Bug fixes:

Various fixes in MLToolbox.

Datalab 1.2.20170525

THIS RELEASE WAS ROLLED BACK DUE TO ISSUES INTERACTING WITH GOOGLE CLOUD SHELL: #1388 The last good release continues to be https://github.com/googledatalab/datalab/releases/tag/v1.1.20170421

This release adds the following:

Features and improvements:

• Add a terminal functionality (#1341)
• Switch to UTF-8 filesystem encoding (#1346)
• Fix Datalab version string to use full semver format (#1319)
• Self-hosted material icons and web fonts (#1313)

Bug fixes:

• Fix bug in the list command that filtered new instances. (#1372)
• Don't attempt to reconnect if the instance is no longer RUNNING. (#1368)
• Do not set SSH options on Windows. (#1357)

Dev cycle improvements

• Implement --live option for faster development. (#1321, #1322)
• Add mochajs framework for automated UI and notebook tests (#1336)
• Add starter unit tests for Datalab (#1369)

Datalab 1.1.20170421

Bug fixes:

Fix MLToolbox Image Classification Package preprocessing and training service run (googledatalab/pydatalab@2f40e2e)

Fix google.datalab.ml.Summary and google.datalab.ml.FeatureSliceView code for python 3 compatibility (same commit as above)

Storage object deletion waits for consistency (googledatalab/pydatalab@3c7e44d)

Fix google.datalab.ml.CsvDataSet not taking bq.Schema object (googledatalab/pydatalab@703fe51)

Fix check for datalab version to use entire version from gcs file (#1319)

Datalab 1.1.20170411

Patch update to fix the perf bug below.

Features and improvements:

Update the color for logical operators to match BigQuery keywords (#1305)

Bug fixes:

Fixed a bug where code cells with very long outputs would cause the browser to be very slow when switching cells or clicking inside them. (#1315)

Datalab 1.1.20170406

First update after GA release brings the following features and bug fixes:

Features and improvements:

• Add a new settings pane under Account dropdown (#1271)
• Don't prompt for a zone when the user specifies a non-existant instance. (#1273)
• Verify the /tmp directory is writeable at startup. (#1267)
• Support selecting a zone by number. (#1277)
• Map the entire /content directory to the disk instead of just /content/datalab (#1281)
• Add .gitignore and README.md when the repo is brand new. (#1270)
• Improve the user experience around accessing help information and notebook navigation (#1284)
• Add new "Jump to file" functionality (#1282)
• Add "Interrupt Execution" (stop cell) button (#1289)
• Upgrade DataFlow to 0.6.0. Remove CloudML SDK (deprecated). (#1299)
• Update ipywidgets to 6.0.0 (#1302)

Bug fixes:

• Bring back Notebook down chevron (#1252)
• Fix link from help panel to Readme noteboook. (#1258)
• Fix editor load errors, restored syntax colors for markdown file (#1260)
• Don't let the datalab tool manage non-datalab instances (#1269)
• Fix css cut-off issue on cell mini-toolbar (#1287)
• Fix appContent height bug that shows on Safari (#1291)
• Fixes for the race condition in setting up the /tmp dir (#1296)
• Only store directories as default tree locations. (#1304)
• Fix some indentation; no code changes. (#1306)

This release also deprecates the Kernel Gateway scenario in:

• Remove support for kernel gateway backends. (#1288)

Datalab GA Release

Google Cloud Datalab is now Generally Available for production use! It features:

• Easy to set up in Google Cloud Platform using the Command Line Interface (CLI)
  commands in Cloud SDK (https://cloud.google.com/datalab/docs/quickstarts). No need for local Docker.

• Two machine learning toolbox solutions (https://github.com/googledatalab/pydatalab/tree/master/solutionbox) (structured data and image classification) and various supporting libraries(https://github.com/googledatalab/pydatalab/tree/master/google/datalab/ml) which provides end-to-end and scalable machine learning solutions.

• A set of new improved BigQuery APIs under namespace “google.datalab” and commands.

• Cell-level menu for ease of use. Cleaner styling and layout.

• Notebook backup to GCS. Streamlined git repo creation and commit workflow.

• New notebooks showing APIs and commands of Storage, BigQuery, and Machine Learning.

With this release, the following things have been deprecated:

• The mode of running Datalab against a kernel gateway. Instead, use the datalab create command to create instances.

• The "datalab.*" APIs. Instead, use the "google.datalab" APIs

Please see the documentation at https://cloud.google.com/datalab/

Special thanks to all the contributors!

Beta release

Initial public release of Google Cloud DataLab (http://cloud.google.com/datalab).

Support for working with BigQuery SQL and Python for interactive and exploratory data analysis, visualization and transformation.