Backlog

Possible and planned features:

• SEML portable mode for publishing source code
• Proper documentation via readthedocs
• Tests
• integrate with Tensorboard HParams for nicer evaluation
• detect any Slurm state instead of only whether the job got killed
• Recommend using separate DB for each user. Maybe provide installation instructions?
• Make Sacred optional (makes SEML easier for beginners, and Sacred might be discontinued at some point)

Low priority:

• Integrate with PyTorch Lightning (what would this even mean? Some convenience functions?)
• Automatic hyperparameter optimization (via Sherpa, hyperopt, Optuna?) -> parallel, on Cluster
• Job chaining via sbatch --dependency
• suspend and restart experiments