Brief Summary of development objectives

h5 data module -permenant train/test splits HIPNN interface

Facilitate dynamic loading+ error handling

1. Better h5 management 1a) Data reducing process 1b) Parsl data integration
2. pySCF + PyPI
3. Multiple QM workers on a block
4. Fix training: each model is its own task
5. make molecule custom data time
6. automatically held out test set
7. make simple bootstrap dataset:
8. Write a parsl configuration for a dedicated nodelist or a local system.
9. get cpu and GPU tasks to run simultaniously on a node.