v0.8.0

[0.8.0] - 2020-03.10

Added

Fixed

• Weights access when pruning in pytorch 1.4

Removed

• torchvision dependency
• sacred dependency
• msd_pytorch.relu_inplace
• command-line interface
• old MSDModule
• stitch functions and modules

v0.7.2

[0.7.2] - 2019-07-30

Added

• Support for multi-gpu execution. Use parallel=True when
  constructing a MSDRegressionModel or MSDSegmentationModel.

Fixed

• Make model.forward() more memory-efficient.

v0.7.1

Added

• Add weights_path command-line argument to msd executable to indicate
  where to store final weights.
• Add MSDBlock2d: this is a faster and slightly more memory efficient
  implementation of the same MSD component. Many thanks to Jonas
  Adler for suggesting this way of structuring the code!

Changed

• The MSD models use MSDBlock2d implementation by default now.