Commit cde0a8dc authored by miaecle's avatar miaecle
Browse files

update

parent f7b1e5e1

deepchem/models/tensorgraph/models/graph_models.py

+18 −5
Original line number Diff line number Diff line
@@ -139,7 +139,7 @@ class WeaveTensorGraph(TensorGraph): 
          pad_batches=pad_batches):



        feed_dict = dict()

        if y_b is not None and not predict:

        if y_b is not None:

          for index, label in enumerate(self.labels_fd):

            if self.mode == "classification":

              feed_dict[label] = to_one_hot(y_b[:, index])
@@ -182,6 +182,12 @@ class WeaveTensorGraph(TensorGraph): 
        feed_dict[self.atom_to_pair] = np.concatenate(atom_to_pair, axis=0)

        yield feed_dict



  def predict_on_generator(self, generator, transformers=[], outputs=None):

      outputs = super(WeaveTensorGraph, self).predict_on_generator(

          generator, 

          transformers=transformers, 

          outputs=outputs)

      return np.stack(outputs, axis=1)



class DTNNTensorGraph(TensorGraph):
@@ -294,7 +300,7 @@ class DTNNTensorGraph(TensorGraph): 
          pad_batches=pad_batches):



        feed_dict = dict()

        if y_b is not None and not predict:

        if y_b is not None:

          for index, label in enumerate(self.labels_fd):

            feed_dict[label] = y_b[:, index:index + 1]

        if w_b is not None:
@@ -334,6 +340,7 @@ class DTNNTensorGraph(TensorGraph): 


        yield feed_dict



  '''

  def predict(self, dataset, transformers=[], outputs=None):

    if outputs is None:

      outputs = self.outputs
@@ -345,7 +352,7 @@ class DTNNTensorGraph(TensorGraph): 
      return retval

    retval = np.concatenate(retval, axis=-1)

    return undo_transforms(retval, transformers)



  '''



class DAGTensorGraph(TensorGraph):
@@ -456,7 +463,7 @@ class DAGTensorGraph(TensorGraph): 
          pad_batches=pad_batches):



        feed_dict = dict()

        if y_b is not None and not predict:

        if y_b is not None:

          for index, label in enumerate(self.labels_fd):

            if self.mode == "classification":

              feed_dict[label] = to_one_hot(y_b[:, index])
@@ -496,6 +503,12 @@ class DAGTensorGraph(TensorGraph): 
        feed_dict[self.n_atoms] = n_atoms

        yield feed_dict



  def predict_on_generator(self, generator, transformers=[], outputs=None):

      outputs = super(DAGTensorGraph, self).predict_on_generator(

          generator, 

          transformers=transformers, 

          outputs=outputs)

      return np.stack(outputs, axis=1)



class PetroskiSuchTensorGraph(TensorGraph):

  """
@@ -1034,7 +1047,7 @@ class MPNNTensorGraph(TensorGraph): 
          pad_batches=pad_batches):



        feed_dict = dict()

        if y_b is not None and not predict:

        if y_b is not None:

          for index, label in enumerate(self.labels_fd):

            if self.mode == "classification":

              feed_dict[label] = to_one_hot(y_b[:, index])

deepchem/models/tensorgraph/models/text_cnn.py

+7 −0
Original line number Diff line number Diff line
@@ -270,3 +270,10 @@ class TextCNNTensorGraph(TensorGraph): 
      # Padding with '_'

      seq.append(self.char_dict['_'])

    return np.array(seq)



  def predict_on_generator(self, generator, transformers=[], outputs=None):

      outputs = super(TextCNNTensorGraph, self).predict_on_generator(

          generator, 

          transformers=transformers, 

          outputs=outputs)

      return np.stack(outputs, axis=1)
 
 No newline at end of file

deepchem/molnet/preset_hyper_parameters.py

+17 −17
Original line number Diff line number Diff line
@@ -51,8 +51,8 @@ hps['irv'] = { 
    'n_K': 10

}

hps['graphconv'] = {

    'batch_size': 50,

    'nb_epoch': 50,

    'batch_size': 64,

    'nb_epoch': 100,

    'learning_rate': 0.0005,

    'n_filters': 64,

    'n_fully_connected_nodes': 128,
@@ -61,23 +61,23 @@ hps['graphconv'] = { 
hps['dag'] = {

    'batch_size': 64,

    'nb_epoch': 100,

    'learning_rate': 0.001,

    'learning_rate': 0.0005,

    'n_graph_feat': 30,

    'default_max_atoms': 60,

    'seed': 123

}

hps['weave'] = {

    'batch_size': 64,

    'nb_epoch': 50,

    'learning_rate': 0.001,

    'nb_epoch': 100,

    'learning_rate': 0.0005,

    'n_graph_feat': 128,

    'n_pair_feat': 14,

    'seed': 123

}

hps['textcnn'] = {

    'batch_size': 64,

    'nb_epoch': 50,

    'learning_rate': 0.001,

    'nb_epoch': 100,

    'learning_rate': 0.0005,

    'n_embedding': 75,

    'filter_sizes': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20],

    'num_filters': [100, 200, 200, 200, 200, 100, 100, 100, 100, 100, 160, 160],
@@ -131,7 +131,7 @@ hps['krr'] = {'alpha': 1e-3} 
hps['krr_ft'] = {'alpha': 1e-3}

hps['graphconvreg'] = {

    'batch_size': 128,

    'nb_epoch': 50,

    'nb_epoch': 100,

    'learning_rate': 0.0005,

    'n_filters': 128,

    'n_fully_connected_nodes': 256,
@@ -139,32 +139,32 @@ hps['graphconvreg'] = { 
}

hps['dtnn'] = {

    'batch_size': 64,

    'nb_epoch': 50,

    'learning_rate': 0.001,

    'n_embedding': 30,

    'n_distance': 100,

    'nb_epoch': 100,

    'learning_rate': 0.0005,

    'n_embedding': 50,

    'n_distance': 170,

    'seed': 123

}

hps['dag_regression'] = {

    'batch_size': 64,

    'nb_epoch': 100,

    'learning_rate': 0.001,

    'learning_rate': 0.0005,

    'n_graph_feat': 30,

    'default_max_atoms': 60,

    'seed': 123

}

hps['weave_regression'] = {

    'batch_size': 64,

    'nb_epoch': 50,

    'learning_rate': 0.001,

    'nb_epoch': 100,

    'learning_rate': 0.0005,

    'n_graph_feat': 128,

    'n_pair_feat': 14,

    'seed': 123

}

hps['textcnn_regression'] = {

    'batch_size': 64,

    'nb_epoch': 50,

    'learning_rate': 0.001,

    'nb_epoch': 100,

    'learning_rate': 0.0005,

    'n_embedding': 75,

    'filter_sizes': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20],

    'num_filters': [100, 200, 200, 200, 200, 100, 100, 100, 100, 100, 160, 160],

deepchem/molnet/run_benchmark_models.py

+5 −3
Original line number Diff line number Diff line
@@ -262,7 +262,8 @@ def benchmark_classification(train_dataset, 
    filter_sizes = hyper_parameters['filter_sizes']

    num_filters = hyper_parameters['num_filters']



    char_dict, length = deepchem.models.TextCNNTensorGraph.build_char_dict(train_dataset)

    all_data = deepchem.data.DiskDataset.merge([train_dataset, valid_dataset, test_dataset])

    char_dict, length = deepchem.models.TextCNNTensorGraph.build_char_dict(all_data)

    

    model = deepchem.models.TextCNNTensorGraph(

        len(tasks),
@@ -510,9 +511,10 @@ def benchmark_regression(train_dataset, 
        batch_size=batch_size,

        learning_rate=learning_rate,

        random_seed=seed,

        output_activation=False,

        use_queue=False,

	mode='regression')





  elif model_name == 'dag_regression':

    batch_size = hyper_parameters['batch_size']

    nb_epoch = hyper_parameters['nb_epoch']

CRA Git | Maintained and supported by SUSTech CRA and CCSE