v4.0.0

Fix

• Update mil config to reflect new signature (c4dc0ed)
• Filter h5 dataset by filenames (b3e1770)
• Safely bypass sigint or sigterm (388d368)
• Remove trainer arguments for pl2 (99194e0)

Breaking

• train|val|test_path keywords now do not refer to the the target train|val|test paths to output h5 files to, but to the files providing paths_and_targets. file_path is the new target file, containing all images. (b3e1770)

v3.2.1

Fix

• Remove pl2 removed trainer arguments (817beb8)

v3.2.0

Feature

• Automate masking with dlup entropy_masker (1066ea1)
• Compile the pytorch lightning models (c009004)

Fix

• Lightning 2.0 ignore missing type imports (08b3e85)
• Ddp default find_unused_parameters=False (3e2a15d)
• Use recommended precision (6b654e2)
• Rename epoch_end to on_epoch_end (37e0c52)
• Add torch/lightning v2 deps (2afdf05)
• Simclr-16-3 update (29fdf58)

Performance

• Do not log to progress bar (da4e3b4)

v3.1.1

Fix

• Pretrained config typo (e6a3778)

v3.1.0

Feature

• Add num_subfolds parameter to create_splits (dfe7ceb)

Fix

• Splits tests with num_subfolds (3a9d44e)

v3.0.1

Fix

• Update viz of untrained model (e32d9f5)

v3.0.0

Feature

• From torchvision.models to pytorchcv models (bb5732d)
• Show targets/img_id/case_id distribution (c757995)

Fix

• Pytorchcv not typed ignore missing imports (2b33f70)

Breaking

• in the config files, models must be specified by corresponding pytorchcv models, not torchvision. (bb5732d)

v2.7.3

Performance

• Speed up feature compilation (772a0bd)

v2.7.2

Fix

• Create feature directory if not exists (674c8c4)

v2.7.1

Fix

• Bypass oom error (0177032)
• Add ipywidgets dependency (3c26dcd)