50 epoch training w eval

10 epoch training with evaluation

Fixed problem with eval, metrics now make sense. This is a run with 1 epoch of training

Finalized evaluation step, which now works. Ready to merge into master

Training for 1 epoch, still need to fix evaluation step

First finalized pipeline

Changes to almost all of the pipeline:

Revert change to dataloader that uses is_test flag.

Preparing logging for training and evaluation, and adding evaluation step [WIP]

reformatting the training file

train model on colab after fixing normalization bug

remove secondary requirements (i.e. not things that are explicitly installed by the user), fix normalization problem, and use tqdm for image processing progress bar

Successfully configured the dataloader and trained for one epoch. Results are not so good, but it's something. Still the Fastaiv1 looked better qualitatively

Fixed a bug in the training stage where the model was not saved, commiting before training on colab

Training stage seems to work, creating a non-run commit to use colab as an orchestration machine

Seems like we are now using the correct format for fastai2. Still there is a strange bug where the signal is killed in training

Migrated to fastai2, creating the DataLoader now works but I'm stuck on not being able to change the batch_size or num_workers as the interface seems to have changed

Finished data import and processing setup, bug in training step

Transition to MLWorkspace docker and setup makefile with environment commands