Commit ac89587b authored by Bharath Ramsundar's avatar Bharath Ramsundar
Browse files

Graph topology example broken somewhere. Need to debug

parent b1657c5b

deepchem/models/tests/test_overfit.py

+1 −1
Original line number Diff line number Diff line
@@ -692,7 +692,7 @@ class TestOverfitAPI(test_util.TensorFlowTestCase): 
    n_atoms = 50

    n_feat = 71

    batch_size = 10

    graph_model = SequentialGraphModel(n_atoms, n_feat, batch_size)

    graph_model = SequentialGraphModel(n_atoms, n_feat)

    graph_model.add(GraphConv(64, activation='relu'))

    graph_model.add(BatchNormalization(epsilon=1e-5, mode=1))

    graph_model.add(GraphPool())

deepchem/models/tf_keras_models/graph_models.py

+0 −3
Original line number Diff line number Diff line
@@ -33,9 +33,6 @@ class SequentialGraphModel(object): 
    n_feat: int

      Number of features per atom.

    """

    

    #super(SequentialGraphModel, self).__init__()

    # Create graph topology and x

    self.graph_topology = GraphTopology(n_atoms, n_feat)

    self.output = self.graph_topology.get_atom_features_placeholder()

    # Keep track of the layers

deepchem/models/tf_keras_models/graph_topology.py

+4 −24
Original line number Diff line number Diff line
@@ -55,12 +55,12 @@ class GraphTopology(object): 
    self.min_deg = min_deg



    self.atom_features_placeholder = Input(

        #tensor=K.placeholder(

        #    shape=(None, self.n_feat), dtype='float32',

        #    name=self.name+'_atom_features'))

        tensor=K.placeholder(

            shape=(self.n_atoms, self.n_feat), dtype='float32',

            shape=(None, self.n_feat), dtype='float32',

            name=self.name+'_atom_features'))

        #tensor=K.placeholder(

        #    shape=(self.n_atoms, self.n_feat), dtype='float32',

        #    name=self.name+'_atom_features'))

    self.deg_adj_lists_placeholders = [

        Input(tensor=K.placeholder(

          shape=(None, deg), dtype='int32', name=self.name+'_deg_adj'+str(deg)))
@@ -136,23 +136,3 @@ class GraphTopology(object): 
                  self.deg_slice_placeholder : batch.deg_slice,

                  self.membership_placeholder : batch.membership}

    return merge_dicts([atoms_dict, deg_adj_dict])
 


'''

def extract_topology(x):

  # Extracts the topology tensors from x

  topology = x[1::]



  # Extract parsed topology information

  deg_slice = topology[0]

  membership = topology[1]

  deg_adj_lists = topology[2::]



  return deg_slice, membership, deg_adj_lists



def extract_nodes(x):

  # Extracts the nodes from x (just the first tensor in the list of tensors)

  return x[0]



def extract_membership(x):

  return x[2]

'''

deepchem/models/tf_keras_models/support_classifier.py

+8 −33
Original line number Diff line number Diff line
@@ -216,31 +216,20 @@ class SupportGraphClassifier(Model): 
  def construct_feed_dict(self, test, support, training=True, add_phase=False):

    """Constructs tensorflow feed from test/support sets."""

    # Generate dictionary elements for support 

    support_labels_dict = {self.support_label_placeholder: np.squeeze(support.y)}

    support_topo_dict = (

    feed_dict = (

        self.model.support_graph_topology.batch_to_feed_dict(support.X))

    support_dict = merge_dicts([support_topo_dict, support_labels_dict])

  

    # Generate dictionary elements for test

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

    print("test.y.shape, test.w.shape")

    print(test.y.shape, test.w.shape)

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

    target_dict = {self.label_placeholder: np.squeeze(test.y),

                   self.weight_placeholder: np.squeeze(test.w)}

    # Get graph information for x

    feed_dict[self.support_label_placeholder] = np.squeeze(support.y)

    # Get graph information for test 

    batch_topo_dict = (

        self.model.test_graph_topology.batch_to_feed_dict(test.X))

    test_dict =  merge_dicts([batch_topo_dict, target_dict])



    test_support_dicts = merge_dicts([test_dict, support_dict])

    feed_dict = merge_dicts([batch_topo_dict, feed_dict_dict])

    # Generate dictionary elements for test

    feed_dict[self.label_placeholder] = np.squeeze(test.y)

    feed_dict[self.weight_placeholder] = np.squeeze(test.w)



    # Get information for keras 

    if add_phase:

      keras_dict = {K.learning_phase() : training}

      feed_dict = merge_dicts([test_support_dicts, keras_dict])

    else:

      feed_dict = test_support_dicts

      feed_dict[K.learning_phase()] = training

    return feed_dict



  def fit(self, dataset, n_trials_per_epoch=1000, nb_epoch=10, n_pos=1,
@@ -341,26 +330,12 @@ class SupportGraphClassifier(Model): 
  def predict_on_batch(self, support, test_batch):

    """Make predictions on batch of data."""

    n_samples = len(test_batch)

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

    print("predict_on_batch()")

    print("test_batch.X.shape, test_batch.y.shape, test_batch.w.shape, test_batch.ids.shape")

    print(test_batch.X.shape, test_batch.y.shape, test_batch.w.shape, test_batch.ids.shape)

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

    padded_test_batch = NumpyDataset(*pad_batch(

        self.test_batch_size, test_batch.X, test_batch.y, test_batch.w,

        test_batch.ids))

    feed_dict = self.construct_feed_dict(padded_test_batch, support)

    # Get scores

    scores = self.sess.run(self.scores_op, feed_dict=feed_dict)

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

    print("predict_on_batch()")

    print("scores.shape")

    print(scores.shape)

    print("scores")

    print(scores)

    y_pred_batch = to_one_hot(np.round(scores))

    y_pred_batch = y_pred_batch[:n_samples]

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

    return y_pred_batch

    

  def evaluate(self, dataset, test_tasks, metrics, n_trials=1000):

CRA Git | Maintained and supported by SUSTech CRA and CCSE