CHANGELOG

0.1.2
 - ref: split the trainer class into submodules to improve readability (#5)
 - ref: remove dump_test_scores method from trainer
 - enh: verify image & mask pairs before training
 - fix: run divided group inference only when data dir available
 - enh: exclude data.path from output dir name
 - enh: pinned specific versions of dependencies
0.1.1
 - ref: add dataset & training submodules (#7)
0.1.0
 - ref: implement hydra based cli (#1 and #3)
 - ref: renamed package semanticsegmenter to cytosegment
0.0.4
 - fix: skip crop & pad for images with target shape in dataset
 - enh: load model metadata safely
 - enh: save models with metadata during training
 - ref: make experiment names short
 - ref: migrate to raghava's augmentation
0.0.3
 - enh: implement function to add meta data to jit model
 - ref: migrate utils submodule to helper_functions module
 - ref: move model converter to model utils
 - enh: load model along with meta data
 - enh: enable cuda for divided group inference
 - enh: add inference script for divided groups
0.0.2
 - enh: remove sigmoid from inference
 - fix: pass target shape for onnx conversion
 - fix: remove min_max from assertion
 - fix: remove concatenation for without augmentation
 - ref: include sigmoid in model architecture
 - ref: resize samples in unet dataset
0.0.1
 - enh: save checkpoint in model eval mode
 - ref: unzipping data properly
 - ref: migrate model summary to model_utils.py
 - enh: update data augmentation and loading
 - ref: create a module for models