ForwardAdaptation

This readme contains the istructions to run the provided code

The packages can be installed using the environment.yml file provided.

- Make sure you already have conda installed on your machine 
- After installing conda, use conda env create -f environment.yml
- Environment with name **magmaenv** should be created now 

To run training, please make sure you have all the correct paths in the Constants.py

- Please make sure you have a Model_Weights folder to save trained models, if not please create one 
- The parameter values in the Constants.py are same as the ones used for the experiments
- Once the training is complete, the models will be saved in the Model_Weights folder
- Only a subset of the original training and validation data is provided because of size constraints

To run optimization

- First make sure you have Data/Compare_Points/EGMs, Data/Compare_Points/Val_folder, and Data/Sim_L.mat
- Then call ECGI_optimize() from ECGI.py - it takes two arguments
- First argument is log folder and the second argument is model filename
- Example call ECGI_optimize ('ECGI_logs_sim_data/', 'Model_name'). The model name will be searched in Model_Weights Folder