Unverified Commit bd74c877 authored by Karl Leswing's avatar Karl Leswing Committed by GitHub
Browse files

Merge pull request #1181 from mlgill/issue1138_mg_remaining_tensor_graph 

Issue 1138: Fix DTNNTensorGraph, DAGTensorGraph, PetroskiSuchTensorGraph, MPNNTensorGraph, and TextCNNTensorGraph
parents 305aea94 b9694079

deepchem/models/__init__.py

+4 −3
Original line number Diff line number Diff line
@@ -17,15 +17,16 @@ from deepchem.models.tensorgraph.IRV import TensorflowMultiTaskIRVClassifier 
from deepchem.models.tensorgraph.robust_multitask import RobustMultitaskClassifier

from deepchem.models.tensorgraph.robust_multitask import RobustMultitaskRegressor

from deepchem.models.tensorgraph.progressive_multitask import ProgressiveMultitaskRegressor, ProgressiveMultitaskClassifier

from deepchem.models.tensorgraph.models.graph_models import WeaveModel, DTNNTensorGraph, DAGTensorGraph, GraphConvModel, MPNNTensorGraph

from deepchem.models.tensorgraph.models.graph_models import WeaveModel, DTNNModel, DAGModel, GraphConvModel, MPNNModel

from deepchem.models.tensorgraph.models.symmetry_function_regression import BPSymmetryFunctionRegression, ANIRegression



from deepchem.models.tensorgraph.models.seqtoseq import SeqToSeq

from deepchem.models.tensorgraph.models.gan import GAN, WGAN

from deepchem.models.tensorgraph.models.text_cnn import TextCNNTensorGraph

from deepchem.models.tensorgraph.models.text_cnn import TextCNNModel

from deepchem.models.tensorgraph.sequential import Sequential

from deepchem.models.tensorgraph.models.sequence_dnn import SequenceDNN



#################### Compatibility imports for renamed TensorGraph models. Remove below with DeepChem 3.0. ####################



from deepchem.models.tensorgraph.models.graph_models import WeaveTensorGraph, GraphConvTensorGraph
 
 No newline at end of file
 
from deepchem.models.tensorgraph.models.text_cnn import TextCNNTensorGraph

from deepchem.models.tensorgraph.models.graph_models import WeaveTensorGraph, DTNNTensorGraph, DAGTensorGraph, GraphConvTensorGraph, MPNNTensorGraph

deepchem/models/tensorgraph/models/graph_models.py

+54 −10
Original line number Diff line number Diff line
@@ -194,7 +194,7 @@ class WeaveModel(TensorGraph): 
    return out





class DTNNTensorGraph(TensorGraph):

class DTNNModel(TensorGraph):



  def __init__(self,

               n_tasks,
@@ -237,7 +237,7 @@ class DTNNTensorGraph(TensorGraph): 
    self.steps = np.expand_dims(self.steps, 0)

    self.output_activation = output_activation

    self.mode = mode

    super(DTNNTensorGraph, self).__init__(**kwargs)

    super(DTNNModel, self).__init__(**kwargs)

    assert self.mode == "regression"

    self.build_graph()
@@ -351,14 +351,14 @@ class DTNNTensorGraph(TensorGraph): 
    if transformers != [] and not isinstance(outputs, collections.Sequence):

      raise ValueError(

          "DTNN does not support single tensor output with transformers")

    retval = super(DTNNTensorGraph, self).predict(dataset, outputs=outputs)

    retval = super(DTNNModel, self).predict(dataset, outputs=outputs)

    if not isinstance(outputs, collections.Sequence):

      return retval

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

    return undo_transforms(retval, transformers)





class DAGTensorGraph(TensorGraph):

class DAGModel(TensorGraph):



  def __init__(self,

               n_tasks,
@@ -390,7 +390,7 @@ class DAGTensorGraph(TensorGraph): 
    self.n_graph_feat = n_graph_feat

    self.n_outputs = n_outputs

    self.mode = mode

    super(DAGTensorGraph, self).__init__(**kwargs)

    super(DAGModel, self).__init__(**kwargs)

    self.build_graph()



  def build_graph(self):
@@ -508,7 +508,7 @@ class DAGTensorGraph(TensorGraph): 
        yield feed_dict



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

    out = super(DAGTensorGraph, self).predict_on_generator(

    out = super(DAGModel, self).predict_on_generator(

        generator, transformers=[], outputs=outputs)

    if outputs is None:

      outputs = self.outputs
@@ -519,7 +519,7 @@ class DAGTensorGraph(TensorGraph): 
    return out





class PetroskiSuchTensorGraph(TensorGraph):

class PetroskiSuchModel(TensorGraph):

  """

      Model from Robust Spatial Filtering with Graph Convolutional Neural Networks

      https://arxiv.org/abs/1703.00792
@@ -545,7 +545,7 @@ class PetroskiSuchTensorGraph(TensorGraph): 
    self.error_bars = True if 'error_bars' in kwargs and kwargs['error_bars'] else False

    self.dropout = dropout

    kwargs['use_queue'] = False

    super(PetroskiSuchTensorGraph, self).__init__(**kwargs)

    super(PetroskiSuchModel, self).__init__(**kwargs)

    self.build_graph()



  def build_graph(self):
@@ -952,7 +952,7 @@ class GraphConvModel(TensorGraph): 
    return y_





class MPNNTensorGraph(TensorGraph):

class MPNNModel(TensorGraph):

  """ Message Passing Neural Network,

      default structures built according to https://arxiv.org/abs/1511.06391 """
@@ -987,7 +987,7 @@ class MPNNTensorGraph(TensorGraph): 
    self.T = T

    self.M = M

    self.mode = mode

    super(MPNNTensorGraph, self).__init__(**kwargs)

    super(MPNNModel, self).__init__(**kwargs)

    self.build_graph()



  def build_graph(self):
@@ -1172,3 +1172,47 @@ class WeaveTensorGraph(WeaveModel): 
  def __init__(self, *args, **kwargs):



    super(WeaveModel, self).__init__(*args, **kwargs)





class DTNNTensorGraph(DTNNModel):



  warnings.warn(

      TENSORGRAPH_DEPRECATION.format("DTNNTensorGraph", "DTNNModel"),

      FutureWarning)



  def __init__(self, *args, **kwargs):



    super(DTNNModel, self).__init__(*args, **kwargs)





class DAGTensorGraph(DAGModel):



  warnings.warn(

      TENSORGRAPH_DEPRECATION.format("DAGTensorGraph", "DAGModel"),

      FutureWarning)



  def __init__(self, *args, **kwargs):



    super(DAGModel, self).__init__(*args, **kwargs)





class PetroskiSuchTensorGraph(PetroskiSuchModel):



  warnings.warn(

      TENSORGRAPH_DEPRECATION.format("PetroskiSuchTensorGraph",

                                     "PetroskiSuchModel"), FutureWarning)



  def __init__(self, *args, **kwargs):



    super(PetroskiSuchModel, self).__init__(*args, **kwargs)





class MPNNTensorGraph(MPNNModel):



  warnings.warn(

      TENSORGRAPH_DEPRECATION.format("MPNNTensorGraph", "MPNNModel"),

      FutureWarning)



  def __init__(self, *args, **kwargs):



    super(MPNNModel, self).__init__(*args, **kwargs)

deepchem/models/tensorgraph/models/text_cnn.py

+21 −3
Original line number Diff line number Diff line
@@ -55,7 +55,7 @@ default_dict = { 
}





class TextCNNTensorGraph(TensorGraph):

class TextCNNModel(TensorGraph):

  """ A Convolutional neural network on smiles strings

  Reimplementation of the discriminator module in ORGAN: https://arxiv.org/abs/1705.10843

  Originated from: http://emnlp2014.org/papers/pdf/EMNLP2014181.pdf
@@ -117,7 +117,7 @@ class TextCNNTensorGraph(TensorGraph): 
    self.num_filters = num_filters

    self.dropout = dropout

    self.mode = mode

    super(TextCNNTensorGraph, self).__init__(**kwargs)

    super(TextCNNModel, self).__init__(**kwargs)

    self.build_graph()



  @staticmethod
@@ -271,7 +271,7 @@ class TextCNNTensorGraph(TensorGraph): 
    return np.array(seq)



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

    out = super(TextCNNTensorGraph, self).predict_on_generator(

    out = super(TextCNNModel, self).predict_on_generator(

        generator, transformers=[], outputs=outputs)

    if outputs is None:

      outputs = self.outputs
@@ -280,3 +280,21 @@ class TextCNNTensorGraph(TensorGraph): 


    out = undo_transforms(out, transformers)

    return out





#################### Deprecation warnings for renamed TensorGraph models ####################



import warnings



TENSORGRAPH_DEPRECATION = "{} is deprecated and has been renamed to {} and will be removed in DeepChem 3.0."





class TextCNNTensorGraph(TextCNNModel):



  warnings.warn(

      TENSORGRAPH_DEPRECATION.format("TextCNNTensorGraph", "TextCNNModel"),

      FutureWarning)



  def __init__(self, *args, **kwargs):



    super(TextCNNTensorGraph, self).__init__(*args, **kwargs)

deepchem/models/tests/test_overfit.py

+7 −7
Original line number Diff line number Diff line
@@ -540,7 +540,7 @@ class TestOverfit(test_util.TensorFlowTestCase): 
    n_tasks = y.shape[1]

    batch_size = 10



    model = dc.models.DTNNTensorGraph(

    model = dc.models.DTNNModel(

        n_tasks,

        n_embedding=20,

        n_distance=100,
@@ -661,7 +661,7 @@ class TestOverfit(test_util.TensorFlowTestCase): 
    transformer = dc.trans.DAGTransformer(max_atoms=50)

    dataset = transformer.transform(dataset)



    model = dc.models.DAGTensorGraph(

    model = dc.models.DAGModel(

        n_tasks,

        max_atoms=50,

        n_atom_feat=n_feat,
@@ -777,7 +777,7 @@ class TestOverfit(test_util.TensorFlowTestCase): 
    n_atom_feat = 75

    n_pair_feat = 14

    batch_size = 10

    model = dc.models.MPNNTensorGraph(

    model = dc.models.MPNNModel(

        n_tasks,

        n_atom_feat=n_atom_feat,

        n_pair_feat=n_pair_feat,
@@ -811,10 +811,10 @@ class TestOverfit(test_util.TensorFlowTestCase): 


    classification_metric = dc.metrics.Metric(dc.metrics.accuracy_score)



    char_dict, length = dc.models.TextCNNTensorGraph.build_char_dict(dataset)

    char_dict, length = dc.models.TextCNNModel.build_char_dict(dataset)

    batch_size = 10



    model = dc.models.TextCNNTensorGraph(

    model = dc.models.TextCNNModel(

        n_tasks,

        char_dict,

        seq_length=length,
@@ -849,10 +849,10 @@ class TestOverfit(test_util.TensorFlowTestCase): 
    regression_metric = dc.metrics.Metric(

        dc.metrics.pearson_r2_score, task_averager=np.mean)



    char_dict, length = dc.models.TextCNNTensorGraph.build_char_dict(dataset)

    char_dict, length = dc.models.TextCNNModel.build_char_dict(dataset)

    batch_size = 10



    model = dc.models.TextCNNTensorGraph(

    model = dc.models.TextCNNModel(

        n_tasks,

        char_dict,

        seq_length=length,

deepchem/molnet/run_benchmark_models.py

+9 −10
Original line number Diff line number Diff line
@@ -223,7 +223,7 @@ def benchmark_classification(train_dataset, 
      test_dataset.reshard(reshard_size)

      test_dataset = transformer.transform(test_dataset)



    model = deepchem.models.DAGTensorGraph(

    model = deepchem.models.DAGModel(

        len(tasks),

        max_atoms=max_atoms,

        n_atom_feat=n_features,
@@ -264,10 +264,9 @@ def benchmark_classification(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)

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



    model = deepchem.models.TextCNNTensorGraph(

    model = deepchem.models.TextCNNModel(

        len(tasks),

        char_dict,

        seq_length=length,
@@ -287,7 +286,7 @@ def benchmark_classification(train_dataset, 
    T = hyper_parameters['T']

    M = hyper_parameters['M']



    model = deepchem.models.MPNNTensorGraph(

    model = deepchem.models.MPNNModel(

        len(tasks),

        n_atom_feat=n_features[0],

        n_pair_feat=n_features[1],
@@ -524,7 +523,7 @@ def benchmark_regression(train_dataset, 
    n_distance = hyper_parameters['n_distance']

    assert len(n_features) == 2, 'DTNN is only applicable to qm datasets'



    model = deepchem.models.DTNNTensorGraph(

    model = deepchem.models.DTNNModel(

        len(tasks),

        n_embedding=n_embedding,

        n_distance=n_distance,
@@ -558,7 +557,7 @@ def benchmark_regression(train_dataset, 
      test_dataset.reshard(reshard_size)

      test_dataset = transformer.transform(test_dataset)



    model = deepchem.models.DAGTensorGraph(

    model = deepchem.models.DAGModel(

        len(tasks),

        max_atoms=max_atoms,

        n_atom_feat=n_features,
@@ -597,10 +596,10 @@ def benchmark_regression(train_dataset, 
    filter_sizes = hyper_parameters['filter_sizes']

    num_filters = hyper_parameters['num_filters']



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

    char_dict, length = deepchem.models.TextCNNModel.build_char_dict(

        train_dataset)



    model = deepchem.models.TextCNNTensorGraph(

    model = deepchem.models.TextCNNModel(

        len(tasks),

        char_dict,

        seq_length=length,
@@ -660,7 +659,7 @@ def benchmark_regression(train_dataset, 
    T = hyper_parameters['T']

    M = hyper_parameters['M']



    model = deepchem.models.MPNNTensorGraph(

    model = deepchem.models.MPNNModel(

        len(tasks),

        n_atom_feat=n_features[0],

        n_pair_feat=n_features[1],

CRA Git | Maintained and supported by SUSTech CRA and CCSE