Trying to replicate MIR results.

Main scripts:

  • models/ToyModel.py -- does most of the heavy lifting for the model. Includes generating the dataset features.
  • models/model_utils/compute_mir_n_migs.py -- computes MIR.
  • core_scripts/ -- sets up the experiments, other admin overhead.
  • core_params/ -- has model and dataset parameters.
  • experiment_generator.py -- has the main experiment tested.