Added in seeds

      feed_dict['{}/{}:0'.format(placeholder_root, name)] = value

    return feed_dict

#class TensorflowGraphModel(object):

class TensorflowGraphModel(Model):

  """Parent class for deepchem Tensorflow models.

               weight_init_stddevs=[.02], bias_init_consts=[1.], penalty=0.0,

               penalty_type="l2", dropouts=[0.5], learning_rate=.001,

               momentum=".9", optimizer="adam", batch_size=50, n_classes=2,

               train=True, verbosity=None, **kwargs):

               train=True, verbosity=None, seed=None, **kwargs):

    """Constructs the computational graph.

    Parameters

    self.n_classes = n_classes

    self.train = train

    self.verbosity = verbosity

    self.seed = seed

    if logdir is not None:

      if not os.path.exists(logdir):

    # replicated supervisor's default path.

    self._save_path = os.path.join(logdir, 'model.ckpt')

    self.train_graph = self.construct_graph(training=True)

    self.eval_graph = self.construct_graph(training=False)

    self.train_graph = self.construct_graph(training=True, seed=self.seed)

    self.eval_graph = self.construct_graph(training=False, seed=self.seed)

  def save(self):

    """

  def get_num_tasks(self):

    return self.n_tasks

  def construct_graph(self, training):

  def construct_graph(self, training, seed):

    """Returns a TensorflowGraph object."""

    graph = tf.Graph() 

    # Setup graph

    with graph.as_default():

      if seed is not None:

        tf.set_random_seed(seed)

      output = self.build(graph, name_scopes, training)

      labels = self.add_label_placeholders(graph, name_scopes)

      weights = self.add_example_weight_placeholders(graph, name_scopes)

          for ind, (X_b, y_b, w_b, ids_b) in enumerate(

              ############################################################ DEBUG

              ## hardcode pad_batches=True to work around limitations in Tensorflow

              dataset.iterbatches(batch_size, pad_batches=True)):

              #dataset.iterbatches(batch_size, pad_batches=False)):

              #dataset.iterbatches(batch_size, pad_batches=True)):

              dataset.iterbatches(batch_size, pad_batches=False)):

              #dataset.iterbatches(batch_size, pad_batches=pad_batches)):

              ############################################################ DEBUG

            if ind % log_every_N_batches == 0:

        outputs = np.array(from_one_hot(

            np.squeeze(np.concatenate(output)), axis=-1))

    outputs = np.copy(outputs)

    return outputs[:len_unpadded]

    outputs = np.reshape(outputs, (len(X), n_tasks))

    outputs = outputs[:len_unpadded]

    return outputs

  def predict_proba_on_batch(self, X, pad_batch=False):

    """Return model output for the provided input.