Loading deepchem/models/tensorgraph/layers.py +66 −48 Original line number Diff line number Diff line Loading @@ -1352,6 +1352,7 @@ class GraphPool(Layer): self.out_tensor = out_tensor return out_tensor class GraphGather(Layer): def __init__(self, batch_size, activation_fn=None, **kwargs): Loading Loading @@ -1399,7 +1400,6 @@ class GraphGather(Layer): return out_tensor class LSTMStep(Layer): """ LSTM whose call is a single step in the LSTM. Loading Loading @@ -1444,15 +1444,18 @@ class LSTMStep(Layer): def none_tensors(self): W, U, b, out_tensor = self.W, self.U, self.b, self.out_tensor h, c = self.h, self.c trainable_weights = self.trainable_weights self.W, self.U, self.b, self.out_tensor = None, None, None, None self.h, self.c = None, None self.trainable_weights = [] return W, U, b, out_tensor, trainable_weights return W, U, b, h, c, out_tensor, trainable_weights def set_tensors(self, tensor): self.W, self.U, self.b, self.out_tensor, self.trainable_weights = tensor (self.W, self.U, self.b, self.h, self.c, self.out_tensor, self.trainable_weights) = tensor def create_tensor(self, in_layers=None, **kwargs): def create_tensor(self, in_layers=None, set_tensors=True, **kwargs): """Execute this layer on input tensors. Parameters Loading Loading @@ -1489,10 +1492,13 @@ class LSTMStep(Layer): h = o * activation(c) # TODO(rbharath): This is wrong!! if set_tensors: self.h = h self.c = c self.out_tensor = h return h, [h, c] def cos(x, y): """Computes the inner preduct (cosine distance) between two tensors. Loading @@ -1508,6 +1514,7 @@ def cos(x, y): + model_ops.epsilon()) return model_ops.dot(x, tf.transpose(y)) / denom class AttnLSTMEmbedding(Layer): """Implements AttnLSTM as in matching networks paper. Loading @@ -1519,12 +1526,7 @@ class AttnLSTMEmbedding(Layer): https://arxiv.org/abs/1511.06391 """ def __init__(self, n_test, n_support, n_feat, max_depth, **kwargs): def __init__(self, n_test, n_support, n_feat, max_depth, **kwargs): """ Parameters ---------- Loading @@ -1544,7 +1546,7 @@ class AttnLSTMEmbedding(Layer): self.n_support = n_support self.n_feat = n_feat def create_tensor(self, in_layers=None, **kwargs): def create_tensor(self, in_layers=None, set_tensors=True, **kwargs): """Execute this layer on input tensors. Parameters Loading Loading @@ -1594,21 +1596,26 @@ class AttnLSTMEmbedding(Layer): y = model_ops.concatenate([q, r], axis=1) q, states = lstm(y, *states) # TODO(rbharath): This is wrong!! if set_tensors: self.out_tensor = xp self.xq = x + q self.xp = xp return [x + q, xp] def none_tensors(self): q_init, r_init, states_init = self.q_init, self.r_init, self.states_init xq, xp = self.xq, self.xp out_tensor = self.out_tensor trainable_weights = self.trainable_weights self.q_init, self.r_init, self.states_init = None, None, None self.xq, self.xp = None, None self.out_tensor = None self.trainable_weights = [] return q_init, r_init, states_init, out_tensor, trainable_weights return q_init, r_init, states_init, xq, xp, out_tensor, trainable_weights def set_tensors(self, tensor): self.q_init, self.r_init, self.states_init, self.out_tensor, self.trainable_weights = tensor (self.q_init, self.r_init, self.states_init, self.xq, self.xp, self.out_tensor, self.trainable_weights) = tensor class IterRefLSTMEmbedding(Layer): Loading Loading @@ -1644,33 +1651,16 @@ class IterRefLSTMEmbedding(Layer): """ super(IterRefLSTMEmbedding, self).__init__(**kwargs) self.init = initializations.get(init) # Set weight initialization self.activation = activations.get(activation) # Get activations self.max_depth = max_depth self.n_test = n_test self.n_support = n_support self.n_feat = n_feat def build(self): """Builds this layer. """ n_feat = self.n_feat # Support set lstm self.support_lstm = LSTMStep(n_feat, 2 * n_feat) self.q_init = model_ops.zeros([self.n_support, n_feat]) self.support_states_init = self.support_lstm.get_initial_states( [self.n_support, n_feat]) # Test lstm self.test_lstm = LSTMStep(n_feat, 2 * n_feat) self.p_init = model_ops.zeros([self.n_test, n_feat]) self.test_states_init = self.test_lstm.get_initial_states( [self.n_test, n_feat]) self.trainable_weights = [] #def build(self): # """Builds this layer. # """ def create_tensor(self, in_layers=None, **kwargs): def create_tensor(self, in_layers=None, set_tensors=True, **kwargs): """Execute this layer on input tensors. Parameters Loading @@ -1687,10 +1677,26 @@ class IterRefLSTMEmbedding(Layer): Returns two tensors of same shape as input. Namely the output shape will be [(n_test, n_feat), (n_support, n_feat)] """ self.build() n_feat = self.n_feat # Support set lstm support_lstm = LSTMStep(n_feat, 2 * n_feat) self.q_init = model_ops.zeros([self.n_support, n_feat]) self.support_states_init = support_lstm.get_initial_states( [self.n_support, n_feat]) # Test lstm test_lstm = LSTMStep(n_feat, 2 * n_feat) self.p_init = model_ops.zeros([self.n_test, n_feat]) self.test_states_init = test_lstm.get_initial_states([self.n_test, n_feat]) self.trainable_weights = [] #self.build() inputs = self._get_input_tensors(in_layers) if len(inputs) != 2: raise ValueError("AttnLSTMEmbedding layer must have exactly two parents") raise ValueError( "IterRefLSTMEmbedding layer must have exactly two parents") x, xp = inputs # Get initializations Loading @@ -1715,28 +1721,40 @@ class IterRefLSTMEmbedding(Layer): # Generate new support attention states qr = model_ops.concatenate([q, r], axis=1) q, states = self.support_lstm(qr, *states) q, states = support_lstm(qr, *states) # Generate new test attention states ps = model_ops.concatenate([p, s], axis=1) p, x_states = self.test_lstm(ps, *x_states) p, x_states = test_lstm(ps, *x_states) # Redefine z = r if set_tensors: self.xp = x + p self.xpq = xp + q self.out_tensor = self.xp return [x + p, xp + q] def none_tensors(self): q_init, r_init, states_init = self.q_init, self.r_init, self.states_init p_init, q_init = self.p_init, self.q_init, support_states_init, test_states_init = (self.support_states_init, self.test_states_init) xp, xpq = self.xp, self.xpq out_tensor = self.out_tensor trainable_weights = self.trainable_weights self.q_init, self.r_init, self.states_init = None, None, None (self.p_init, self.q_init, self.support_states_init, self.test_states_init) = (None, None, None, None) self.xp, self.xpq = None, None self.out_tensor = None self.trainable_weights = [] return q_init, r_init, states_init, out_tensor, trainable_weights return (p_init, q_init, support_states_init, test_states_init, xp, xpq, out_tensor, trainable_weights) def set_tensors(self, tensor): self.q_init, self.r_init, self.states_init, self.out_tensor, self.trainable_weights = tensor (self.p_init, self.q_init, self.support_states_init, self.test_states_init, self.xp, self.xpq, self.out_tensor, self.trainable_weights) = tensor class BatchNorm(Layer): Loading deepchem/models/tensorgraph/tensor_graph.py +1 −26 Original line number Diff line number Diff line Loading @@ -389,10 +389,6 @@ class TensorGraph(Model): for node in order: with tf.name_scope(node): node_layer = self.layers[node] ############################################################ DEBUG #print("node_layer") #print(node_layer) ############################################################ DEBUG node_layer.create_tensor(training=self._training_placeholder) self.rnn_initial_states += node_layer.rnn_initial_states self.rnn_final_states += node_layer.rnn_final_states Loading Loading @@ -510,28 +506,7 @@ class TensorGraph(Model): # Pickle itself pickle_name = os.path.join(self.model_dir, "model.pickle") ########################################################## DEBUG #print("self") #print(self) #def debug_pickle(instance): # attribute = None # for k, v in instance.__dict__.iteritems(): # print("checking: %s" % k) # print("type(v)") # print(type(v)) # try: # pickle.dumps(v) # except: # attribute = k # break # return attribute #debug_pickle(self) ##pickle.dumps(self) ########################################################## DEBUG with open(pickle_name, 'wb') as fout: try: pickle.dump(self, fout) Loading deepchem/models/tensorgraph/tests/test_layers.py +7 −8 Original line number Diff line number Diff line Loading @@ -416,13 +416,12 @@ class TestLayers(test_util.TensorFlowTestCase): lstm = LSTMStep(n_feat, 2 * n_feat) out_tensor = lstm(y, state_zero, state_one) sess.run(tf.global_variables_initializer()) h_out, h_copy_out, c_out = ( out_tensor[0].eval(), out_tensor[1][0].eval(), out_tensor[1][1].eval()) h_out, h_copy_out, c_out = (out_tensor[0].eval(), out_tensor[1][0].eval(), out_tensor[1][1].eval()) assert h_out.shape == (n_test, n_feat) assert h_copy_out.shape == (n_test, n_feat) assert c_out.shape == (n_test, n_feat) def test_attn_lstm_embedding(self): """Test that attention LSTM computation works properly.""" max_depth = 5 Loading Loading @@ -457,8 +456,8 @@ class TestLayers(test_util.TensorFlowTestCase): test = tf.convert_to_tensor(test, dtype=tf.float32) support = tf.convert_to_tensor(support, dtype=tf.float32) iter_ref_embedding_layer = IterRefLSTMEmbedding( n_test, n_support, n_feat, max_depth) iter_ref_embedding_layer = IterRefLSTMEmbedding(n_test, n_support, n_feat, max_depth) out_tensor = iter_ref_embedding_layer(test, support) sess.run(tf.global_variables_initializer()) test_out, support_out = out_tensor[0].eval(), out_tensor[1].eval() Loading deepchem/models/tensorgraph/tests/test_layers_pickle.py +22 −5 Original line number Diff line number Diff line Loading @@ -5,7 +5,7 @@ from deepchem.models.tensorgraph.layers import Feature, Conv1D, Dense, Flatten, CombineMeanStd, Repeat, GRU, L2Loss, Concat, SoftMax, Constant, Variable, Add, Multiply, InteratomicL2Distances, \ SoftMaxCrossEntropy, ReduceMean, ToFloat, ReduceSquareDifference, Conv2D, MaxPool, ReduceSum, GraphConv, GraphPool, \ GraphGather, BatchNorm, WeightedError, \ LSTMStep, AttnLSTMEmbedding LSTMStep, AttnLSTMEmbedding, IterRefLSTMEmbedding from deepchem.models.tensorgraph.graph_layers import Combine_AP, Separate_AP, \ WeaveLayer, WeaveGather, DTNNEmbedding, DTNNGather, DTNNStep, \ DTNNExtract, DAGLayer, DAGGather, MessagePassing, SetGather Loading Loading @@ -452,8 +452,8 @@ def test_AttnLSTM_pickle(): tg = TensorGraph(batch_size=n_test) test = Feature(shape=(None, n_feat)) support = Feature(shape=(None, n_feat)) out = AttnLSTMEmbedding(n_test, n_support, n_feat, max_depth, in_layers=[test, support]) out = AttnLSTMEmbedding( n_test, n_support, n_feat, max_depth, in_layers=[test, support]) tg.add_output(out) tg.set_loss(out) tg.build() Loading @@ -474,6 +474,23 @@ def test_LSTMStep_pickle(): tg.save() def test_IterRefLSTM_pickle(): """Tests that IterRefLSTM can be pickled.""" n_feat = 10 max_depth = 5 n_test = 5 n_support = 5 tg = TensorGraph() test = Feature(shape=(None, n_feat)) support = Feature(shape=(None, n_feat)) lstm = IterRefLSTMEmbedding( n_test, n_support, n_feat, max_depth, in_layers=[test, support]) tg.add_output(lstm) tg.set_loss(lstm) tg.build() tg.save() def test_SetGather_pickle(): tg = TensorGraph() atom_feature = Feature(shape=(None, 100)) Loading deepchem/nn/layers.py +5 −4 Original line number Diff line number Diff line Loading @@ -355,7 +355,8 @@ class GraphGather(Layer): batch_size: int Number of elements in batch of data. """ warnings.warn("The dc.nn.GraphGather is " warnings.warn( "The dc.nn.GraphGather is " "deprecated. Will be removed in DeepChem 1.4. " "Will be replaced by dc.models.tensorgraph.layers.GraphGather", DeprecationWarning) Loading Loading
deepchem/models/tensorgraph/layers.py +66 −48 Original line number Diff line number Diff line Loading @@ -1352,6 +1352,7 @@ class GraphPool(Layer): self.out_tensor = out_tensor return out_tensor class GraphGather(Layer): def __init__(self, batch_size, activation_fn=None, **kwargs): Loading Loading @@ -1399,7 +1400,6 @@ class GraphGather(Layer): return out_tensor class LSTMStep(Layer): """ LSTM whose call is a single step in the LSTM. Loading Loading @@ -1444,15 +1444,18 @@ class LSTMStep(Layer): def none_tensors(self): W, U, b, out_tensor = self.W, self.U, self.b, self.out_tensor h, c = self.h, self.c trainable_weights = self.trainable_weights self.W, self.U, self.b, self.out_tensor = None, None, None, None self.h, self.c = None, None self.trainable_weights = [] return W, U, b, out_tensor, trainable_weights return W, U, b, h, c, out_tensor, trainable_weights def set_tensors(self, tensor): self.W, self.U, self.b, self.out_tensor, self.trainable_weights = tensor (self.W, self.U, self.b, self.h, self.c, self.out_tensor, self.trainable_weights) = tensor def create_tensor(self, in_layers=None, **kwargs): def create_tensor(self, in_layers=None, set_tensors=True, **kwargs): """Execute this layer on input tensors. Parameters Loading Loading @@ -1489,10 +1492,13 @@ class LSTMStep(Layer): h = o * activation(c) # TODO(rbharath): This is wrong!! if set_tensors: self.h = h self.c = c self.out_tensor = h return h, [h, c] def cos(x, y): """Computes the inner preduct (cosine distance) between two tensors. Loading @@ -1508,6 +1514,7 @@ def cos(x, y): + model_ops.epsilon()) return model_ops.dot(x, tf.transpose(y)) / denom class AttnLSTMEmbedding(Layer): """Implements AttnLSTM as in matching networks paper. Loading @@ -1519,12 +1526,7 @@ class AttnLSTMEmbedding(Layer): https://arxiv.org/abs/1511.06391 """ def __init__(self, n_test, n_support, n_feat, max_depth, **kwargs): def __init__(self, n_test, n_support, n_feat, max_depth, **kwargs): """ Parameters ---------- Loading @@ -1544,7 +1546,7 @@ class AttnLSTMEmbedding(Layer): self.n_support = n_support self.n_feat = n_feat def create_tensor(self, in_layers=None, **kwargs): def create_tensor(self, in_layers=None, set_tensors=True, **kwargs): """Execute this layer on input tensors. Parameters Loading Loading @@ -1594,21 +1596,26 @@ class AttnLSTMEmbedding(Layer): y = model_ops.concatenate([q, r], axis=1) q, states = lstm(y, *states) # TODO(rbharath): This is wrong!! if set_tensors: self.out_tensor = xp self.xq = x + q self.xp = xp return [x + q, xp] def none_tensors(self): q_init, r_init, states_init = self.q_init, self.r_init, self.states_init xq, xp = self.xq, self.xp out_tensor = self.out_tensor trainable_weights = self.trainable_weights self.q_init, self.r_init, self.states_init = None, None, None self.xq, self.xp = None, None self.out_tensor = None self.trainable_weights = [] return q_init, r_init, states_init, out_tensor, trainable_weights return q_init, r_init, states_init, xq, xp, out_tensor, trainable_weights def set_tensors(self, tensor): self.q_init, self.r_init, self.states_init, self.out_tensor, self.trainable_weights = tensor (self.q_init, self.r_init, self.states_init, self.xq, self.xp, self.out_tensor, self.trainable_weights) = tensor class IterRefLSTMEmbedding(Layer): Loading Loading @@ -1644,33 +1651,16 @@ class IterRefLSTMEmbedding(Layer): """ super(IterRefLSTMEmbedding, self).__init__(**kwargs) self.init = initializations.get(init) # Set weight initialization self.activation = activations.get(activation) # Get activations self.max_depth = max_depth self.n_test = n_test self.n_support = n_support self.n_feat = n_feat def build(self): """Builds this layer. """ n_feat = self.n_feat # Support set lstm self.support_lstm = LSTMStep(n_feat, 2 * n_feat) self.q_init = model_ops.zeros([self.n_support, n_feat]) self.support_states_init = self.support_lstm.get_initial_states( [self.n_support, n_feat]) # Test lstm self.test_lstm = LSTMStep(n_feat, 2 * n_feat) self.p_init = model_ops.zeros([self.n_test, n_feat]) self.test_states_init = self.test_lstm.get_initial_states( [self.n_test, n_feat]) self.trainable_weights = [] #def build(self): # """Builds this layer. # """ def create_tensor(self, in_layers=None, **kwargs): def create_tensor(self, in_layers=None, set_tensors=True, **kwargs): """Execute this layer on input tensors. Parameters Loading @@ -1687,10 +1677,26 @@ class IterRefLSTMEmbedding(Layer): Returns two tensors of same shape as input. Namely the output shape will be [(n_test, n_feat), (n_support, n_feat)] """ self.build() n_feat = self.n_feat # Support set lstm support_lstm = LSTMStep(n_feat, 2 * n_feat) self.q_init = model_ops.zeros([self.n_support, n_feat]) self.support_states_init = support_lstm.get_initial_states( [self.n_support, n_feat]) # Test lstm test_lstm = LSTMStep(n_feat, 2 * n_feat) self.p_init = model_ops.zeros([self.n_test, n_feat]) self.test_states_init = test_lstm.get_initial_states([self.n_test, n_feat]) self.trainable_weights = [] #self.build() inputs = self._get_input_tensors(in_layers) if len(inputs) != 2: raise ValueError("AttnLSTMEmbedding layer must have exactly two parents") raise ValueError( "IterRefLSTMEmbedding layer must have exactly two parents") x, xp = inputs # Get initializations Loading @@ -1715,28 +1721,40 @@ class IterRefLSTMEmbedding(Layer): # Generate new support attention states qr = model_ops.concatenate([q, r], axis=1) q, states = self.support_lstm(qr, *states) q, states = support_lstm(qr, *states) # Generate new test attention states ps = model_ops.concatenate([p, s], axis=1) p, x_states = self.test_lstm(ps, *x_states) p, x_states = test_lstm(ps, *x_states) # Redefine z = r if set_tensors: self.xp = x + p self.xpq = xp + q self.out_tensor = self.xp return [x + p, xp + q] def none_tensors(self): q_init, r_init, states_init = self.q_init, self.r_init, self.states_init p_init, q_init = self.p_init, self.q_init, support_states_init, test_states_init = (self.support_states_init, self.test_states_init) xp, xpq = self.xp, self.xpq out_tensor = self.out_tensor trainable_weights = self.trainable_weights self.q_init, self.r_init, self.states_init = None, None, None (self.p_init, self.q_init, self.support_states_init, self.test_states_init) = (None, None, None, None) self.xp, self.xpq = None, None self.out_tensor = None self.trainable_weights = [] return q_init, r_init, states_init, out_tensor, trainable_weights return (p_init, q_init, support_states_init, test_states_init, xp, xpq, out_tensor, trainable_weights) def set_tensors(self, tensor): self.q_init, self.r_init, self.states_init, self.out_tensor, self.trainable_weights = tensor (self.p_init, self.q_init, self.support_states_init, self.test_states_init, self.xp, self.xpq, self.out_tensor, self.trainable_weights) = tensor class BatchNorm(Layer): Loading
deepchem/models/tensorgraph/tensor_graph.py +1 −26 Original line number Diff line number Diff line Loading @@ -389,10 +389,6 @@ class TensorGraph(Model): for node in order: with tf.name_scope(node): node_layer = self.layers[node] ############################################################ DEBUG #print("node_layer") #print(node_layer) ############################################################ DEBUG node_layer.create_tensor(training=self._training_placeholder) self.rnn_initial_states += node_layer.rnn_initial_states self.rnn_final_states += node_layer.rnn_final_states Loading Loading @@ -510,28 +506,7 @@ class TensorGraph(Model): # Pickle itself pickle_name = os.path.join(self.model_dir, "model.pickle") ########################################################## DEBUG #print("self") #print(self) #def debug_pickle(instance): # attribute = None # for k, v in instance.__dict__.iteritems(): # print("checking: %s" % k) # print("type(v)") # print(type(v)) # try: # pickle.dumps(v) # except: # attribute = k # break # return attribute #debug_pickle(self) ##pickle.dumps(self) ########################################################## DEBUG with open(pickle_name, 'wb') as fout: try: pickle.dump(self, fout) Loading
deepchem/models/tensorgraph/tests/test_layers.py +7 −8 Original line number Diff line number Diff line Loading @@ -416,13 +416,12 @@ class TestLayers(test_util.TensorFlowTestCase): lstm = LSTMStep(n_feat, 2 * n_feat) out_tensor = lstm(y, state_zero, state_one) sess.run(tf.global_variables_initializer()) h_out, h_copy_out, c_out = ( out_tensor[0].eval(), out_tensor[1][0].eval(), out_tensor[1][1].eval()) h_out, h_copy_out, c_out = (out_tensor[0].eval(), out_tensor[1][0].eval(), out_tensor[1][1].eval()) assert h_out.shape == (n_test, n_feat) assert h_copy_out.shape == (n_test, n_feat) assert c_out.shape == (n_test, n_feat) def test_attn_lstm_embedding(self): """Test that attention LSTM computation works properly.""" max_depth = 5 Loading Loading @@ -457,8 +456,8 @@ class TestLayers(test_util.TensorFlowTestCase): test = tf.convert_to_tensor(test, dtype=tf.float32) support = tf.convert_to_tensor(support, dtype=tf.float32) iter_ref_embedding_layer = IterRefLSTMEmbedding( n_test, n_support, n_feat, max_depth) iter_ref_embedding_layer = IterRefLSTMEmbedding(n_test, n_support, n_feat, max_depth) out_tensor = iter_ref_embedding_layer(test, support) sess.run(tf.global_variables_initializer()) test_out, support_out = out_tensor[0].eval(), out_tensor[1].eval() Loading
deepchem/models/tensorgraph/tests/test_layers_pickle.py +22 −5 Original line number Diff line number Diff line Loading @@ -5,7 +5,7 @@ from deepchem.models.tensorgraph.layers import Feature, Conv1D, Dense, Flatten, CombineMeanStd, Repeat, GRU, L2Loss, Concat, SoftMax, Constant, Variable, Add, Multiply, InteratomicL2Distances, \ SoftMaxCrossEntropy, ReduceMean, ToFloat, ReduceSquareDifference, Conv2D, MaxPool, ReduceSum, GraphConv, GraphPool, \ GraphGather, BatchNorm, WeightedError, \ LSTMStep, AttnLSTMEmbedding LSTMStep, AttnLSTMEmbedding, IterRefLSTMEmbedding from deepchem.models.tensorgraph.graph_layers import Combine_AP, Separate_AP, \ WeaveLayer, WeaveGather, DTNNEmbedding, DTNNGather, DTNNStep, \ DTNNExtract, DAGLayer, DAGGather, MessagePassing, SetGather Loading Loading @@ -452,8 +452,8 @@ def test_AttnLSTM_pickle(): tg = TensorGraph(batch_size=n_test) test = Feature(shape=(None, n_feat)) support = Feature(shape=(None, n_feat)) out = AttnLSTMEmbedding(n_test, n_support, n_feat, max_depth, in_layers=[test, support]) out = AttnLSTMEmbedding( n_test, n_support, n_feat, max_depth, in_layers=[test, support]) tg.add_output(out) tg.set_loss(out) tg.build() Loading @@ -474,6 +474,23 @@ def test_LSTMStep_pickle(): tg.save() def test_IterRefLSTM_pickle(): """Tests that IterRefLSTM can be pickled.""" n_feat = 10 max_depth = 5 n_test = 5 n_support = 5 tg = TensorGraph() test = Feature(shape=(None, n_feat)) support = Feature(shape=(None, n_feat)) lstm = IterRefLSTMEmbedding( n_test, n_support, n_feat, max_depth, in_layers=[test, support]) tg.add_output(lstm) tg.set_loss(lstm) tg.build() tg.save() def test_SetGather_pickle(): tg = TensorGraph() atom_feature = Feature(shape=(None, 100)) Loading
deepchem/nn/layers.py +5 −4 Original line number Diff line number Diff line Loading @@ -355,7 +355,8 @@ class GraphGather(Layer): batch_size: int Number of elements in batch of data. """ warnings.warn("The dc.nn.GraphGather is " warnings.warn( "The dc.nn.GraphGather is " "deprecated. Will be removed in DeepChem 1.4. " "Will be replaced by dc.models.tensorgraph.layers.GraphGather", DeprecationWarning) Loading
