Remove requirement of having the type of model as a command line argu…

…ment (it will simply be a hyperparameter) since we currently need to be able to run the model script like 'python MY_MODEL_SCRIPT.py' with no arguments

options.py

@@ -1,7 +1,7 @@
 import os, sys
 import argparse
 
-def parse_training_arguments():
+def parse_training_arguments(model_name):
     parser = argparse.ArgumentParser()
     required_args = parser.add_argument_group('required arguments')
     parser.add_argument('-d', '--depth', dest="depth", type=str, 
@@ -13,8 +13,8 @@ def parse_training_arguments():
     parser.add_argument('-k', '--kappa', type=str, dest="kappa", 
         help='How fast to linearly ramp up KL loss', 
             metavar="kappa", default='1.')
-    required_args.add_argument("-m", "--model_name", dest="model_name", 
-        help="model name: adage | vae", metavar="model_name")
+    parser.add_argument("-m", "--model_name", dest="model_name", 
+        help="model name: adage | vae", metavar="model_name", default=model_name)
     parser.add_argument('-N', '--noise', type=float, dest="noise",
         help='How much Gaussian noise to add during training',
             metavar="noise", default=0.05)
@@ -87,7 +87,7 @@ def parse_prediction_arguments():
     args = parser.parse_args()
     return args
 
-def parse_command_line_arguments(task):
+def parse_command_line_arguments(task, model_name):
     if task == "predict":
         opt = parse_prediction_arguments()
         checkpoint_combined = 'checkpoints/' + ".".join(["tybalt",   
@@ -96,7 +96,7 @@ def parse_command_line_arguments(task):
 
     elif task == "train":
         # task == "train"
-        opt = parse_training_arguments()
+        opt = parse_training_arguments(model_name)
         if not os.path.isdir(os.path.join(os.getcwd(), 'checkpoints')):
            os.mkdir(os.path.join(os.getcwd(), 'checkpoints'))
         if not opt.model_name:

train.py

@@ -47,7 +47,7 @@ def save_training_performance(pd, args):
 
 if __name__ == '__main__':
     opt, checkpoint_combined, checkpoint_encoder, checkpoint_decoder = \
-        parse_command_line_arguments("train")
+        parse_command_line_arguments("train", candle_params['tybalt_model_name'])
 
     opt, rnaseq_df, train_df, test_df = get_data(opt)