Skip to content

Latest commit

 

History

History
10 lines (9 loc) · 1.1 KB

README.md

File metadata and controls

10 lines (9 loc) · 1.1 KB

generative model experiments

generate_parameters.py: produces the set of hyperparameters and models we want to grid search over.
run_experiments.py: is the main loop which loads the datasets and models -> fits and trains the model according to the train and validation sets -> evaluates the model according to the test set -> generates 1000 samples and counts the mismatches from the wild type
optimizer.py: searches over the set of hyperparameters generated from generate_parameters.py and calls run_experiments.py to fit and valuate each model using a train, valid, and test set. (this is also the part that can be parallelized). It then sorts each model by their test set score.
models.py: all models inherit from this class and contain an init, fit, evaluate, sample, show_model, plot_model, save, and load function. all models contain an internal object called self.model which is the actual underlying model.
utils.py: helper functions
vae.py: the vae implementation
rnn.py: the rnn implementation
hmm.py: the hmm implementation