benchmark modified

  if model in ['graphconv']:

    featurizer = 'GraphConv'

    n_features = 71

  elif model in ['rf', 'tf', 'tf_robust', 'logreg']:

  elif model in ['tf', 'tf_robust', 'logreg', 'rf']:

    featurizer = 'ECFP'

    n_features = 1024

  else:

  for dname in dataset_name:

    print('-------------------------------------')

    print('Benchmark test on dataset: '+dname)

    print('Benchmark %s on dataset: %s' % (model, dname))

    print('-------------------------------------')

    base_dir = os.path.join(base_dir_o, dname)

    time_start = time.time()

    #loading datasets     

    tasks,datasets,transformers = loading_functions[dname](featurizer=featurizer)

    tasks,datasets,transformers = loading_functions[dname](

        featurizer=featurizer)

    train_dataset, valid_dataset, test_dataset = datasets

    time_finish_loading = time.time()

    #time_finish_loading-time_start is the time(s) used for dataset loading

def benchmark_train_and_valid(base_dir, train_dataset, valid_dataset, tasks,

                              transformers, hyper_parameters, 

                              n_features, model='tf',

                              n_features, model='tf', seed=123, 

                              verbosity='high'):

  """

  Calculate performance of different models on the specific dataset & tasks

    K.set_session(sess)

    # Building graph convolution model

    with g.as_default():

      if seed is not None:

      tf.set_random_seed(seed)

      graph_model = dc.nn.SequentialGraph(n_features)

      graph_model.add(dc.nn.GraphConv(int(n_filters), activation='relu'))