removed returns from molgan, test clean-up

    Take noise data as an input and processes it through number of

    dense and dropout layers. Then data is converted into two forms

    one used for training and other for generation of compounds.

    Returns

    -------

    keras.Model

        Returns generator model.

        There are four output of this model:

    The model has four outputs:

      1. edges logits used during training

      2. nodes logits used during training

      3. edges logits used for compound generation

    The input vectors need to be in one-hot encoding format.

    Use MolGAN featurizer for that purpose. It will be simplified

    in the future release.

    Returns

    -------

    keras.Model

        Returns disctriminator model

    """

    adjacency_tensor = layers.Input(

        shape=(self.vertices, self.vertices, self.edges))

from deepchem.models import BasicMolGANModel as MolGAN

from deepchem.models.optimizers import ExponentialDecay

from tensorflow import one_hot

from tensorflow.errors import InternalError

from tensorflow.keras.backend import clear_session as keras_clear_session

      # generate sample

      g = gan.predict_gan_generator(1000)

      # check how many valid molecules were created and add to list

      generated_molecules = feat.defeaturize(g)

      valid_molecules_count = len(

          list(filter(lambda x: x is not None, generated_molecules)))

      if valid_molecules_count:

        valid_attempts = valid_attempts + 1

    # finally test if there was at least one valid training session

    # as the model structure improves this should become more and more strict

    assert valid_attempts > 0