0.81.0-dev

• Added lab show functionality that visualises the directed flow of the lab project
• Various bug fixes

Lab 0.7.9-dev

• Added artifact logging
• Tensorboard support
• Further under the hood refinements and bug fixes

Lab 0.7.8-dev

New features:

• Add automated checking for sync with minio
• Simplified integration with Jupyter notebooks / Jupyter lab
• Models are now persisted using cloudpickle to facilitate cross-domain portability
• lab run and lab ls are able to parse and reflect additional arguments passed to train scripts via sys.argv

A number of bug fixes and improved error handling.

Lab 0.7.0-dev

• Improved error handling and updates
• Configuration of multiple minio backends
• Easier removal of experiments
• Refactoring that simplifies further development of the common line interface
• Various bug fixes

Lab 0.6.4-dev

Lab now supports training experiments in nested directories. For example:

lab run experiment1/train_dir1/train.py

Added a .gitignore file that excludes virtual environment from lab projects uploaded to GitHub

Breaking Changes

experiments is now a standalone module and is not a part of projects module. A new experiments is now set up as:

from lab.experiment import Experimet

Lab 0.6.3-dev

• Changes to the logging API, mainly marking sklearn and keras modules under a single umbrella class.
• Documentation of the core concepts

Lab 0.6.1-dev

New features in this release:

• Integration with private Minio cloud for centralised project storage
• Improved efficiencies around virtualenv transfers
• General bug fixes and error handling