v0.2.0
Here are key updates in this release:
Configurable GR-ConvNets
Configurable IOU threshold in evaluation
Updated train/val split method to RandomSampler based method
Support for configurable optimizer
Updated logging to single directory
Save training logs and args
Updated evaluate.py to support multiple networks