The provided code will train single NICE models across all data missingness mechanisms and rates listed within the experiment description from the paper.

To train, run "python train.py".

To generate test set completions, run "python mnist_impute.py".

To gather RMSE statistics, replace line 9 of nice_rmse_test.py with desired test folder, and run "python nice_rmse_test.py".  
Final outputs will be mean and standard deviation of individual and average PL-MCMC imputation RMSE.

To gather FID statistics, replace line 277 of nice_rmse_test.py with desired test folder, and run "python nice_rmse_test.py".
Final outputs will be mean and standard deviation of individual and average PL-MCMC imputation FID