Model checkpoints and mlflow #7
Description
Current implementation lacks of saving routines for checkpoints during optimization.
- Study how mlflow stores them
- Determine how easy it would be to retrieve them from mlflow database
- Evaluate ease of transforming them back to torch models ready for use