0.2.1

What's Changed

• remove managers - training is based on PyTorch lightning
• Added debug operations in data pipeline
• Added more imaging operations
• Model should return now batch_dict and not batch_dict["model"]
• Minor bugfixes

Full Changelog: 0.2.0...0.2.1

0.2.0

Major revision that significantly improves flexibility and code sharing.
Read the new main README file for a detailed description.

0.1.12

• Enhance and fix examples
• minor bug fixes

0.1.11

• KNIGHT challenge - baseline implementation for task 2
• Fix single process caching
• Fix multimodality clinical and imaging tutorial notebook

0.1.10

• Fix training mode with cpu only
• Evaluation for BRIGHT challenge - including implementation for BSS (Brier Skill Score)
• Evaluation for KNIGHT challenge
• Baseline implementation for KNIGHT challenge
• Simple tutorial for multimodality: imaging and clinical

0.1.9

• loss warmup
• metric from loss
• minor bugfix (missing lr_sch_target)

0.1.8

What changed:

• option to set a random seed which allows reproducing results
• option to set a learning rate scheduler target
• minor bug fixes

0.1.7

Fix broken import in metrics module

0.1.6

citeable release with zendo

0.1.5

Update landing page for PyPI