@ayushdeva-ai @avtansh1317 @sansiddh
This PR is somewhat ready. Please take a look and provide feedback.
Here are the key points:
The IHME config is now quite similar to SEIR.
The IHME model is internally quite different from SIR but the fitting.py script mirrors the one for SIR, and much of the utils, plotting etc. overlap.
There are new scripts to run the IHME and SEIR models.
There are still some TODOs scattered throughout the code which I'll be addressing.
There is a driver config in configs/other/ (can be renamed) which can be used to generate multiple configs to run experiments. I personally found this useful for hyperparameter tuning and multiple train period experiments.
Ignore any function called create_output, it just dumps out all components of predictions_dict individually and will be removed before merging.
This PR does not have any scripts which combine IHME and SEIR, that will be in a separate PR.