Remove uses of deprecated APIs

contrib/atomicconv/models/atomicnet_ops.py

@@ -80,8 +80,8 @@ def AtomicNNLayer(tensor, size, weights, biases, name=None):
   """
 
   if len(tensor.get_shape()) != 2:
-    raise ValueError('Dense layer input must be 2D, not %dD' %
-                     len(tensor.get_shape()))
+    raise ValueError(
+        'Dense layer input must be 2D, not %dD' % len(tensor.get_shape()))
   with tf.name_scope(name, 'fully_connected', [tensor, weights, biases]):
     return tf.nn.xw_plus_b(tensor, weights, biases)
 
@@ -111,8 +111,8 @@ def gather_neighbors(X, nbr_indices, B, N, M, d):
   example_tensors = tf.unstack(X, axis=0)
   example_nbrs = tf.unstack(nbr_indices, axis=0)
   all_nbr_coords = []
-  for example, (example_tensor,
-                example_nbr) in enumerate(zip(example_tensors, example_nbrs)):
+  for example, (example_tensor, example_nbr) in enumerate(
+      zip(example_tensors, example_nbrs)):
     nbr_coords = tf.gather(example_tensor, example_nbr)
     all_nbr_coords.append(nbr_coords)
   neighbors = tf.stack(all_nbr_coords)
@@ -149,7 +149,7 @@ def DistanceTensor(X, Nbrs, boxsize, B, N, M, d):
       nbrs_tensors = tf.unstack(nbrs, axis=1)
       for nbr, nbr_tensor in enumerate(nbrs_tensors):
         _D = tf.subtract(nbr_tensor, atom_tensor)
-        _D = tf.subtract(_D, boxsize * tf.round(tf.div(_D, boxsize)))
+        _D = tf.subtract(_D, boxsize * tf.round(tf.math.divide(_D, boxsize)))
         D.append(_D)
   else:
     for atom, atom_tensor in enumerate(atom_tensors):

contrib/atomicconv/models/legacy.py

@@ -278,7 +278,7 @@ def add_training_cost(self, graph, name_scopes, output, labels, weights):
               # non-zero weight examples in the batch.  Also, instead of using
               # tf.reduce_mean (which can put ops on the CPU) we explicitly
               # calculate with div/sum so it stays on the GPU.
-              gradient_cost = tf.div(
+              gradient_cost = tf.math.divide(
                   tf.reduce_sum(weighted_cost), self.batch_size)
               gradient_costs.append(gradient_cost)
 
@@ -416,8 +416,8 @@ def add_example_weight_placeholders(self, graph, name_scopes):
     feeding and fetching the same tensor.
     """
     weights = []
-    placeholder_scope = TensorflowGraph.get_placeholder_scope(graph,
-                                                              name_scopes)
+    placeholder_scope = TensorflowGraph.get_placeholder_scope(
+        graph, name_scopes)
     with placeholder_scope:
       for task in range(self.n_tasks):
         weights.append(
@@ -617,8 +617,8 @@ def add_label_placeholders(self, graph, name_scopes):
     Placeholders are wrapped in identity ops to avoid the error caused by
     feeding and fetching the same tensor.
     """
-    placeholder_scope = TensorflowGraph.get_placeholder_scope(graph,
-                                                              name_scopes)
+    placeholder_scope = TensorflowGraph.get_placeholder_scope(
+        graph, name_scopes)
     with graph.as_default():
       batch_size = self.batch_size
       n_classes = self.n_classes
@@ -770,8 +770,8 @@ def add_label_placeholders(self, graph, name_scopes):
     Placeholders are wrapped in identity ops to avoid the error caused by
     feeding and fetching the same tensor.
     """
-    placeholder_scope = TensorflowGraph.get_placeholder_scope(graph,
-                                                              name_scopes)
+    placeholder_scope = TensorflowGraph.get_placeholder_scope(
+        graph, name_scopes)
     with graph.as_default():
       batch_size = self.batch_size
       labels = []
@@ -859,8 +859,8 @@ def build(self, graph, name_scopes, training):
         batch_size x n_features.
     """
     n_features = self.n_features
-    placeholder_scope = TensorflowGraph.get_placeholder_scope(graph,
-                                                              name_scopes)
+    placeholder_scope = TensorflowGraph.get_placeholder_scope(
+        graph, name_scopes)
     with graph.as_default():
       with placeholder_scope:
         self.mol_features = tf.placeholder(
@@ -906,8 +906,8 @@ def build(self, graph, name_scopes, training):
                     weight_init=tf.truncated_normal(
                         shape=[prev_layer_size, 1],
                         stddev=weight_init_stddevs[i]),
-                    bias_init=tf.constant(value=bias_init_consts[i], shape=[1
-                                                                           ]))))
+                    bias_init=tf.constant(value=bias_init_consts[i],
+                                          shape=[1]))))
       return output
 
   def construct_feed_dict(self, X_b, y_b=None, w_b=None, ids_b=None):

contrib/one_shot_models/multitask_classifier.py

@@ -85,8 +85,9 @@ def __init__(self,
                pad_batches=True,
                verbose=True):
 
-    warnings.warn("MultitaskGraphClassifier is deprecated. "
-                  "Will be removed in DeepChem 1.4.", DeprecationWarning)
+    warnings.warn(
+        "MultitaskGraphClassifier is deprecated. "
+        "Will be removed in DeepChem 1.4.", DeprecationWarning)
     super(MultitaskGraphClassifier, self).__init__(
         model_dir=logdir, verbose=verbose)
     self.n_tasks = n_tasks
@@ -191,16 +192,17 @@ def add_training_loss(self, final_loss, logits):
       task_label_vector = task_labels[task]
       task_weight_vector = task_weights[task]
       # Convert the labels into one-hot vector encodings.
-      one_hot_labels = tf.to_float(
-          tf.one_hot(tf.to_int32(tf.squeeze(task_label_vector)), 2))
+      one_hot_labels = tf.cast(
+          tf.one_hot(tf.cast(tf.squeeze(task_label_vector), tf.int32), 2),
+          tf.float32)
       # Since we use tf.nn.softmax_cross_entropy_with_logits note that we pass in
       # un-softmaxed logits rather than softmax outputs.
       task_loss = loss_fn(logits[task], one_hot_labels, task_weight_vector)
       task_losses.append(task_loss)
     # It's ok to divide by just the batch_size rather than the number of nonzero
     # examples (effect averages out)
     total_loss = tf.add_n(task_losses)
-    total_loss = tf.div(total_loss, self.batch_size)
+    total_loss = tf.math.divide(total_loss, self.batch_size)
     return total_loss
 
   def add_softmax(self, outputs):

contrib/one_shot_models/multitask_regressor.py

@@ -44,8 +44,9 @@ def __init__(self,
                pad_batches=True,
                verbose=True):
 
-    warnings.warn("MultitaskGraphRegressor is deprecated. "
-                  "Will be removed in DeepChem 1.4.", DeprecationWarning)
+    warnings.warn(
+        "MultitaskGraphRegressor is deprecated. "
+        "Will be removed in DeepChem 1.4.", DeprecationWarning)
 
     super(MultitaskGraphRegressor, self).__init__(
         model_dir=logdir, verbose=verbose)
@@ -154,14 +155,13 @@ def add_training_loss(self, final_loss, outputs):
     for task in range(self.n_tasks):
       task_label_vector = task_labels[task]
       task_weight_vector = task_weights[task]
-      task_loss = loss_fn(outputs[task],
-                          tf.squeeze(task_label_vector),
+      task_loss = loss_fn(outputs[task], tf.squeeze(task_label_vector),
                           tf.squeeze(task_weight_vector))
       task_losses.append(task_loss)
     # It's ok to divide by just the batch_size rather than the number of nonzero
     # examples (effect averages out)
     total_loss = tf.add_n(task_losses)
-    total_loss = tf.div(total_loss, self.batch_size)
+    total_loss = tf.math.divide(total_loss, self.batch_size)
     return total_loss
 
   def fit(self,
@@ -222,8 +222,9 @@ class DTNNMultitaskGraphRegressor(MultitaskGraphRegressor):
 
   def build(self):
     # Create target inputs
-    warnings.warn("DTNNMultitaskGraphRegressor is deprecated. "
-                  "Will be removed in DeepChem 1.4.", DeprecationWarning)
+    warnings.warn(
+        "DTNNMultitaskGraphRegressor is deprecated. "
+        "Will be removed in DeepChem 1.4.", DeprecationWarning)
     self.label_placeholder = tf.placeholder(
         dtype='float32', shape=(None, self.n_tasks), name="label_placeholder")
     self.weight_placeholder = tf.placeholder(

contrib/tensorflow_models/__init__.py

@@ -51,8 +51,9 @@ class TensorflowGraph(object):
 
   def __init__(self, graph, session, name_scopes, output, labels, weights,
                loss):
-    warnings.warn("TensorflowGraph is deprecated. "
-                  "Will be removed in DeepChem 1.4.", DeprecationWarning)
+    warnings.warn(
+        "TensorflowGraph is deprecated. "
+        "Will be removed in DeepChem 1.4.", DeprecationWarning)
     self.graph = graph
     self.session = session
     self.name_scopes = name_scopes
@@ -178,8 +179,9 @@ def __init__(self,
     seed: int
       If not none, is used as random seed for tensorflow.
     """
-    warnings.warn("TensorflowGraphModel is deprecated. "
-                  "Will be removed in DeepChem 1.4.", DeprecationWarning)
+    warnings.warn(
+        "TensorflowGraphModel is deprecated. "
+        "Will be removed in DeepChem 1.4.", DeprecationWarning)
 
     # Save hyperparameters
     self.n_tasks = n_tasks
@@ -278,7 +280,7 @@ def add_training_cost(self, graph, name_scopes, output, labels, weights):
               # non-zero weight examples in the batch.  Also, instead of using
               # tf.reduce_mean (which can put ops on the CPU) we explicitly
               # calculate with div/sum so it stays on the GPU.
-              gradient_cost = tf.div(
+              gradient_cost = tf.math.divide(
                   tf.reduce_sum(weighted_cost), self.batch_size)
               gradient_costs.append(gradient_cost)
 

contrib/tensorflow_models/progressive_joint.py

@@ -12,6 +12,7 @@
 from deepchem.metrics import from_one_hot
 from deepchem.nn import model_ops
 
+
 class ProgressiveJointRegressor(TensorflowMultiTaskRegressor):
   """Implements a progressive multitask neural network.
   
@@ -39,8 +40,9 @@ def __init__(self, n_tasks, n_features, alpha_init_stddevs=[.02], **kwargs):
     alpha_init_stddevs: list
       List of standard-deviations for alpha in adapter layers.
     """
-    warnings.warn("ProgressiveJointRegressor is deprecated. "
-                  "Will be removed in DeepChem 1.4.", DeprecationWarning)
+    warnings.warn(
+        "ProgressiveJointRegressor is deprecated. "
+        "Will be removed in DeepChem 1.4.", DeprecationWarning)
     self.alpha_init_stddevs = alpha_init_stddevs
     super(ProgressiveJointRegressor, self).__init__(n_tasks, n_features,
                                                     **kwargs)
@@ -312,7 +314,7 @@ def add_training_costs(self, graph, name_scopes, output, labels, weights):
               # non-zero weight examples in the batch.  Also, instead of using
               # tf.reduce_mean (which can put ops on the CPU) we explicitly
               # calculate with div/sum so it stays on the GPU.
-              gradient_cost = tf.div(
+              gradient_cost = tf.math.divide(
                   tf.reduce_sum(weighted_cost), self.batch_size)
               gradient_costs.append(gradient_cost)
 

contrib/tensorflow_models/progressive_multitask.py

@@ -12,6 +12,7 @@
 from deepchem.metrics import from_one_hot
 from deepchem.nn import model_ops
 
+
 class ProgressiveMultitaskRegressor(TensorflowMultiTaskRegressor):
   """Implements a progressive multitask neural network.
   
@@ -39,8 +40,9 @@ def __init__(self, n_tasks, n_features, alpha_init_stddevs=[.02], **kwargs):
     alpha_init_stddevs: list
       List of standard-deviations for alpha in adapter layers.
     """
-    warnings.warn("ProgressiveMultitaskRegressor is deprecated. "
-                  "Will be removed in DeepChem 1.4.", DeprecationWarning)
+    warnings.warn(
+        "ProgressiveMultitaskRegressor is deprecated. "
+        "Will be removed in DeepChem 1.4.", DeprecationWarning)
     self.alpha_init_stddevs = alpha_init_stddevs
     super(ProgressiveMultitaskRegressor, self).__init__(n_tasks, n_features,
                                                         **kwargs)
@@ -293,7 +295,7 @@ def add_training_costs(self, graph, name_scopes, output, labels, weights):
               # non-zero weight examples in the batch.  Also, instead of using
               # tf.reduce_mean (which can put ops on the CPU) we explicitly
               # calculate with div/sum so it stays on the GPU.
-              gradient_cost = tf.div(
+              gradient_cost = tf.math.divide(
                   tf.reduce_sum(weighted_cost), self.batch_size)
               gradient_costs.append(gradient_cost)
 
@@ -487,7 +489,8 @@ def add_task_training_costs(self, graph, name_scopes, outputs, labels,
           # non-zero weight examples in the batch.  Also, instead of using
           # tf.reduce_mean (which can put ops on the CPU) we explicitly
           # calculate with div/sum so it stays on the GPU.
-          task_cost = tf.div(tf.reduce_sum(weighted_cost), self.batch_size)
+          task_cost = tf.math.divide(
+              tf.reduce_sum(weighted_cost), self.batch_size)
           task_costs[task] = task_cost
 
     return task_costs

contrib/tensorflow_models/test_utils.py

@@ -66,7 +66,7 @@ def PrepareMask(self, features, mask):
     return masked_features, mask_t
 
   def Check(self, func, features, expected, axis=None, mask=None):
-    with self.test_session() as sess:
+    with self.session() as sess:
       features, features_t = self.PrepareFeatures(features)
       if mask is not None:
         features, mask = self.PrepareMask(features, mask)
@@ -127,7 +127,7 @@ def testVarianceWithMask(self):
         mask=[[1, 0], [0, 1]])
 
   def testMoment(self):
-    with self.test_session() as sess:
+    with self.session() as sess:
       features = np.random.random((3, 4, 5))
       features_t = tf.constant(features, dtype=tf.float32)
 
@@ -168,7 +168,7 @@ def testMoment(self):
             atol=1e-5)
 
   def testSkewness(self):
-    with self.test_session() as sess:
+    with self.session() as sess:
       features = np.random.random((3, 4, 5))
       features_t = tf.constant(features, dtype=tf.float32)
       self.assertAllClose(
@@ -183,7 +183,7 @@ def testSkewness(self):
           atol=1e-5)
 
   def testKurtosis(self):
-    with self.test_session() as sess:
+    with self.session() as sess:
       features = np.random.random((3, 4, 5))
       features_t = tf.constant(features, dtype=tf.float32)
       self.assertAllClose(

contrib/tensorflow_models/utils.py

@@ -36,8 +36,9 @@ def ParseCheckpoint(checkpoint):
   Returns:
     The path to an actual checkpoint file.
   """
-  warnings.warn("ParseCheckpoint is deprecated. "
-                "Will be removed in DeepChem 1.4.", DeprecationWarning)
+  warnings.warn(
+      "ParseCheckpoint is deprecated. "
+      "Will be removed in DeepChem 1.4.", DeprecationWarning)
   with open(checkpoint) as f:
     try:
       cp = checkpoint_state_pb2.CheckpointState()
@@ -67,8 +68,8 @@ def Mask(t, mask):
   if mask is None:
     return t
   if not t.get_shape()[:-1].is_compatible_with(mask.get_shape()):
-    raise ValueError('Shapes do not match: %s vs. %s' % (t.get_shape(),
-                                                         mask.get_shape()))
+    raise ValueError(
+        'Shapes do not match: %s vs. %s' % (t.get_shape(), mask.get_shape()))
   return tf.multiply(t, tf.expand_dims(mask, -1))
 
 
@@ -184,12 +185,12 @@ def Moment(k, tensor, standardize=False, reduction_indices=None, mask=None):
 
   # compute the requested central moment
   # note that mean is a raw moment, not a central moment
-  mean = tf.div(
+  mean = tf.math.divide(
       tf.reduce_sum(tensor, axis=reduction_indices, keep_dims=True), divisor)
   delta = tensor - mean
   if mask is not None:
     delta = Mask(delta, mask)
-  moment = tf.div(
+  moment = tf.math.divide(
       tf.reduce_sum(
           math_ops.pow(delta, k), axis=reduction_indices, keep_dims=True),
       divisor)

contrib/vina_model/test_vina_model.py

@@ -46,7 +46,7 @@ def test_get_cells(self):
     start = 0
     stop = 4
     nbr_cutoff = 1
-    with self.test_session() as sess:
+    with self.session() as sess:
       ndim = 3
       cells = get_cells(start, stop, nbr_cutoff, ndim=ndim).eval()
       assert len(cells.shape) == 2
@@ -71,7 +71,7 @@ def test_compute_neighbor_list(self):
     # The number of cells which we should theoretically have
     n_cells = int(((stop - start) / nbr_cutoff)**ndim)
 
-    with self.test_session() as sess:
+    with self.session() as sess:
       coords = start + np.random.rand(N, ndim) * (stop - start)
       coords = tf.stack(coords)
       nbr_list = compute_neighbor_list(
@@ -90,7 +90,7 @@ def test_put_atoms_in_cells(self):
     # The number of cells which we should theoretically have
     n_cells = ((stop - start) / nbr_cutoff)**ndim
 
-    with self.test_session() as sess:
+    with self.session() as sess:
       cells = get_cells(start, stop, nbr_cutoff, ndim=ndim)
       coords = np.random.rand(N, ndim)
       _, atoms_in_cells = put_atoms_in_cells(coords, cells, N, n_cells, ndim, k)
@@ -113,7 +113,7 @@ def test_compute_neighbor_cells(self):
     # TODO(rbharath): The test below only checks that shapes work out.
     # Need to do a correctness implementation vs. a simple CPU impl.
 
-    with self.test_session() as sess:
+    with self.session() as sess:
       cells = get_cells(start, stop, nbr_cutoff, ndim=ndim)
       nbr_cells = compute_neighbor_cells(cells, ndim, n_cells)
       nbr_cells = nbr_cells.eval()
@@ -136,7 +136,7 @@ def test_compute_closest_neighbors(self):
     # TODO(rbharath): The test below only checks that shapes work out.
     # Need to do a correctness implementation vs. a simple CPU impl.
 
-    with self.test_session() as sess:
+    with self.session() as sess:
       cells = get_cells(start, stop, nbr_cutoff, ndim=ndim)
       nbr_cells = compute_neighbor_cells(cells, ndim, n_cells)
       coords = np.random.rand(N, ndim)
@@ -158,7 +158,7 @@ def test_get_cells_for_atoms(self):
     # TODO(rbharath): The test below only checks that shapes work out.
     # Need to do a correctness implementation vs. a simple CPU impl.
 
-    with self.test_session() as sess:
+    with self.session() as sess:
       cells = get_cells(start, stop, nbr_cutoff, ndim=ndim)
       coords = np.random.rand(N, ndim)
       cells_for_atoms = get_cells_for_atoms(coords, cells, N, n_cells, ndim)