Loading deepchem/models/layers.py +33 −50 Original line number Diff line number Diff line Loading @@ -2344,7 +2344,13 @@ class WeaveLayer(tf.keras.layers.Layer): input_shape: tuple Ignored since we don't need the input shape to create internal weights. """ init = initializers.get(self.init) # Set weight initialization def init(input_shape): return self.add_weight( name='kernel', shape=(input_shape[0], input_shape[1]), initializer=self.init, trainable=True) self.W_AA = init([self.n_atom_input_feat, self.n_hidden_AA]) self.b_AA = backend.zeros(shape=[ Loading Loading @@ -2566,7 +2572,14 @@ class WeaveGather(tf.keras.layers.Layer): def build(self, input_shape): if self.compress_post_gaussian_expansion: init = initializers.get(self.init) def init(input_shape): return self.add_weight( name='kernel', shape=(input_shape[0], input_shape[1]), initializer=self.init, trainable=True) self.W = init([self.n_input * 11, self.n_input]) self.b = backend.zeros(shape=[self.n_input]) self.built = True Loading Loading @@ -2935,37 +2948,22 @@ class DAGLayer(tf.keras.layers.Layer): self.W_list = [] self.b_list = [] self.dropouts = [] init = initializers.get(self.init) prev_layer_size = self.n_inputs for layer_size in self.layer_sizes: self.W_list.append( self.add_weight( name='kernel', shape=(prev_layer_size, layer_size), initializer='glorot_uniform', trainable=True)) self.b_list.append( self.add_weight( name='bias', shape=(layer_size,), initializer='zeros', trainable=True)) self.W_list.append(init([prev_layer_size, layer_size])) self.b_list.append(backend.zeros(shape=[ layer_size, ])) if self.dropout is not None and self.dropout > 0.0: self.dropouts.append(Dropout(rate=self.dropout)) else: self.dropouts.append(None) prev_layer_size = layer_size self.W_list.append( self.add_weight( name='kernel', shape=(prev_layer_size, self.n_outputs), initializer=self.init, trainable=True)) self.b_list.append( self.add_weight( name='bias', shape=(self.n_outputs,), initializer='zeros', trainable=True)) self.W_list.append(init([prev_layer_size, self.n_outputs])) self.b_list.append(backend.zeros(shape=[ self.n_outputs, ])) if self.dropout is not None and self.dropout > 0.0: self.dropouts.append(Dropout(rate=self.dropout)) else: Loading Loading @@ -3083,37 +3081,22 @@ class DAGGather(tf.keras.layers.Layer): self.W_list = [] self.b_list = [] self.dropouts = [] init = initializers.get(self.init) prev_layer_size = self.n_graph_feat for layer_size in self.layer_sizes: self.W_list.append( self.add_weight( name='kernel', shape=(prev_layer_size, layer_size), initializer='glorot_uniform', trainable=True)) self.b_list.append( self.add_weight( name='bias', shape=(layer_size,), initializer='zeros', trainable=True)) self.W_list.append(init([prev_layer_size, layer_size])) self.b_list.append(backend.zeros(shape=[ layer_size, ])) if self.dropout is not None and self.dropout > 0.0: self.dropouts.append(Dropout(rate=self.dropout)) else: self.dropouts.append(None) prev_layer_size = layer_size self.W_list.append( self.add_weight( name='kernel', shape=(prev_layer_size, self.n_outputs), initializer=self.init, trainable=True)) self.b_list.append( self.add_weight( name='bias', shape=(self.n_outputs,), initializer='zeros', trainable=True)) self.W_list.append(init([prev_layer_size, self.n_outputs])) self.b_list.append(backend.zeros(shape=[ self.n_outputs, ])) if self.dropout is not None and self.dropout > 0.0: self.dropouts.append(Dropout(rate=self.dropout)) else: Loading deepchem/models/tests/test_reload.py +27 −98 Original line number Diff line number Diff line Loading @@ -522,141 +522,70 @@ def test_progressivemultitaskregressor_reload(): assert scores[regression_metric.name] < 0.1 def test_DAG_regression_reload(): """Test DAG regressor reloads.""" def test_weave_classification_reload(): """Test weave model can be reloaded.""" np.random.seed(123) tf.random.set_seed(123) n_tasks = 1 #current_dir = os.path.dirname(os.path.abspath(__file__)) # Load mini log-solubility dataset. featurizer = dc.feat.ConvMolFeaturizer() featurizer = dc.feat.WeaveFeaturizer() tasks = ["outcome"] mols = ["CC", "CCO", "CC", "CCC", "CCCCO", "CO", "CC", "CCCCC", "CCC", "CCCO"] mols = ["CC", "CCCCC", "CCCCC", "CCC", "COOO", "COO", "OO"] n_samples = len(mols) X = featurizer(mols) y = np.random.rand(n_samples, n_tasks) y = [1, 1, 1, 1, 0, 0, 0] dataset = dc.data.NumpyDataset(X, y) regression_metric = dc.metrics.Metric( dc.metrics.pearson_r2_score, task_averager=np.mean) classification_metric = dc.metrics.Metric(dc.metrics.roc_auc_score) n_feat = 75 batch_size = 10 transformer = dc.trans.DAGTransformer(max_atoms=50) dataset = transformer.transform(dataset) batch_size = 5 model_dir = tempfile.mkdtemp() model = dc.models.DAGModel( model = dc.models.eaveModel( n_tasks, max_atoms=50, n_atom_feat=n_feat, batch_size=batch_size, learning_rate=0.001, use_queue=False, mode="regression", learning_rate=0.01, mode="classification", dropouts=0.0, model_dir=model_dir) # Fit trained model model.fit(dataset, nb_epoch=100) # Eval model on train scores = model.evaluate(dataset, [regression_metric]) assert scores[regression_metric.name] > .1 scores = model.evaluate(dataset, [classification_metric]) assert scores[classification_metric.name] > .6 # Check predictions match on random sample predmols = ["CCCC", "CCCCCO", "CCCCC"] Xpred = featurizer(predmols) reloaded_model = dc.models.DAGModel( predset = dc.data.NumpyDataset(Xpred) origpred = model.predict(predset) reloaded_model = dc.models.WeaveModel( n_tasks, max_atoms=50, n_atom_feat=n_feat, batch_size=batch_size, learning_rate=0.001, use_queue=False, mode="regression", learning_rate=0.003, mode="classification", dropouts=0.0, model_dir=model_dir) reloaded_model.restore() # Check predictions match on random sample predmols = ["CCCC", "CCCCCO", "CCCCC"] Xpred = featurizer(predmols) predset = dc.data.NumpyDataset(Xpred) predset = transformer.transform(predset) origpred = model.predict(predset) reloadpred = reloaded_model.predict(predset) assert np.all(origpred == reloadpred) #Eval model on train scores = reloaded_model.evaluate(dataset, [regression_metric]) assert scores[regression_metric.name] > .1 scores = reloaded_model.evaluate(dataset, [classification_metric]) assert scores[classification_metric.name] > .6 ## TODO: THIS IS FAILING! #def test_weave_classification_reload_alt(): # """Test weave model can be reloaded.""" # np.random.seed(123) # tf.random.set_seed(123) # n_tasks = 1 # # # Load mini log-solubility dataset. # featurizer = dc.feat.WeaveFeaturizer() # tasks = ["outcome"] # mols = ["C", "CO", "CC"] # n_samples = len(mols) # X = featurizer(mols) # y = np.random.randint(2, size=(n_samples, n_tasks)) # dataset = dc.data.NumpyDataset(X, y) # # classification_metric = dc.metrics.Metric(dc.metrics.roc_auc_score) # # batch_size = 10 # # model_dir = tempfile.mkdtemp() # model = dc.models.WeaveModel( # n_tasks, # batch_size=batch_size, # learning_rate=0.0003, # mode="classification", # dropouts=0.0, # model_dir=model_dir) # # # Fit trained model # model.fit(dataset, nb_epoch=30) # # # Eval model on train # scores = model.evaluate(dataset, [classification_metric]) # assert scores[classification_metric.name] > .9 # # # Custom save # save_dir = tempfile.mkdtemp() # model.model.save(save_dir) # # from tensorflow import keras # reloaded = keras.models.load_model(save_dir) # # reloaded_model = dc.models.WeaveModel( # n_tasks, # batch_size=batch_size, # learning_rate=0.0003, # mode="classification", # dropouts=0.0, # model_dir=model_dir) # #reloaded_model.restore() # reloaded_model.model = reloaded # # # Check predictions match on random sample # predmols = ["CCCC", "CCCCCO", "CCCCC"] # Xpred = featurizer(predmols) # predset = dc.data.NumpyDataset(Xpred) # origpred = model.predict(predset) # reloadpred = reloaded_model.predict(predset) # assert np.all(origpred == reloadpred) # # # Eval model on train # scores = reloaded_model.evaluate(dataset, [classification_metric]) # assert scores[classification_metric.name] > .9 # # ## TODO: THIS IS FAILING! #@pytest.mark.slow #def test_weave_classification_reload(): Loading Loading
deepchem/models/layers.py +33 −50 Original line number Diff line number Diff line Loading @@ -2344,7 +2344,13 @@ class WeaveLayer(tf.keras.layers.Layer): input_shape: tuple Ignored since we don't need the input shape to create internal weights. """ init = initializers.get(self.init) # Set weight initialization def init(input_shape): return self.add_weight( name='kernel', shape=(input_shape[0], input_shape[1]), initializer=self.init, trainable=True) self.W_AA = init([self.n_atom_input_feat, self.n_hidden_AA]) self.b_AA = backend.zeros(shape=[ Loading Loading @@ -2566,7 +2572,14 @@ class WeaveGather(tf.keras.layers.Layer): def build(self, input_shape): if self.compress_post_gaussian_expansion: init = initializers.get(self.init) def init(input_shape): return self.add_weight( name='kernel', shape=(input_shape[0], input_shape[1]), initializer=self.init, trainable=True) self.W = init([self.n_input * 11, self.n_input]) self.b = backend.zeros(shape=[self.n_input]) self.built = True Loading Loading @@ -2935,37 +2948,22 @@ class DAGLayer(tf.keras.layers.Layer): self.W_list = [] self.b_list = [] self.dropouts = [] init = initializers.get(self.init) prev_layer_size = self.n_inputs for layer_size in self.layer_sizes: self.W_list.append( self.add_weight( name='kernel', shape=(prev_layer_size, layer_size), initializer='glorot_uniform', trainable=True)) self.b_list.append( self.add_weight( name='bias', shape=(layer_size,), initializer='zeros', trainable=True)) self.W_list.append(init([prev_layer_size, layer_size])) self.b_list.append(backend.zeros(shape=[ layer_size, ])) if self.dropout is not None and self.dropout > 0.0: self.dropouts.append(Dropout(rate=self.dropout)) else: self.dropouts.append(None) prev_layer_size = layer_size self.W_list.append( self.add_weight( name='kernel', shape=(prev_layer_size, self.n_outputs), initializer=self.init, trainable=True)) self.b_list.append( self.add_weight( name='bias', shape=(self.n_outputs,), initializer='zeros', trainable=True)) self.W_list.append(init([prev_layer_size, self.n_outputs])) self.b_list.append(backend.zeros(shape=[ self.n_outputs, ])) if self.dropout is not None and self.dropout > 0.0: self.dropouts.append(Dropout(rate=self.dropout)) else: Loading Loading @@ -3083,37 +3081,22 @@ class DAGGather(tf.keras.layers.Layer): self.W_list = [] self.b_list = [] self.dropouts = [] init = initializers.get(self.init) prev_layer_size = self.n_graph_feat for layer_size in self.layer_sizes: self.W_list.append( self.add_weight( name='kernel', shape=(prev_layer_size, layer_size), initializer='glorot_uniform', trainable=True)) self.b_list.append( self.add_weight( name='bias', shape=(layer_size,), initializer='zeros', trainable=True)) self.W_list.append(init([prev_layer_size, layer_size])) self.b_list.append(backend.zeros(shape=[ layer_size, ])) if self.dropout is not None and self.dropout > 0.0: self.dropouts.append(Dropout(rate=self.dropout)) else: self.dropouts.append(None) prev_layer_size = layer_size self.W_list.append( self.add_weight( name='kernel', shape=(prev_layer_size, self.n_outputs), initializer=self.init, trainable=True)) self.b_list.append( self.add_weight( name='bias', shape=(self.n_outputs,), initializer='zeros', trainable=True)) self.W_list.append(init([prev_layer_size, self.n_outputs])) self.b_list.append(backend.zeros(shape=[ self.n_outputs, ])) if self.dropout is not None and self.dropout > 0.0: self.dropouts.append(Dropout(rate=self.dropout)) else: Loading
deepchem/models/tests/test_reload.py +27 −98 Original line number Diff line number Diff line Loading @@ -522,141 +522,70 @@ def test_progressivemultitaskregressor_reload(): assert scores[regression_metric.name] < 0.1 def test_DAG_regression_reload(): """Test DAG regressor reloads.""" def test_weave_classification_reload(): """Test weave model can be reloaded.""" np.random.seed(123) tf.random.set_seed(123) n_tasks = 1 #current_dir = os.path.dirname(os.path.abspath(__file__)) # Load mini log-solubility dataset. featurizer = dc.feat.ConvMolFeaturizer() featurizer = dc.feat.WeaveFeaturizer() tasks = ["outcome"] mols = ["CC", "CCO", "CC", "CCC", "CCCCO", "CO", "CC", "CCCCC", "CCC", "CCCO"] mols = ["CC", "CCCCC", "CCCCC", "CCC", "COOO", "COO", "OO"] n_samples = len(mols) X = featurizer(mols) y = np.random.rand(n_samples, n_tasks) y = [1, 1, 1, 1, 0, 0, 0] dataset = dc.data.NumpyDataset(X, y) regression_metric = dc.metrics.Metric( dc.metrics.pearson_r2_score, task_averager=np.mean) classification_metric = dc.metrics.Metric(dc.metrics.roc_auc_score) n_feat = 75 batch_size = 10 transformer = dc.trans.DAGTransformer(max_atoms=50) dataset = transformer.transform(dataset) batch_size = 5 model_dir = tempfile.mkdtemp() model = dc.models.DAGModel( model = dc.models.eaveModel( n_tasks, max_atoms=50, n_atom_feat=n_feat, batch_size=batch_size, learning_rate=0.001, use_queue=False, mode="regression", learning_rate=0.01, mode="classification", dropouts=0.0, model_dir=model_dir) # Fit trained model model.fit(dataset, nb_epoch=100) # Eval model on train scores = model.evaluate(dataset, [regression_metric]) assert scores[regression_metric.name] > .1 scores = model.evaluate(dataset, [classification_metric]) assert scores[classification_metric.name] > .6 # Check predictions match on random sample predmols = ["CCCC", "CCCCCO", "CCCCC"] Xpred = featurizer(predmols) reloaded_model = dc.models.DAGModel( predset = dc.data.NumpyDataset(Xpred) origpred = model.predict(predset) reloaded_model = dc.models.WeaveModel( n_tasks, max_atoms=50, n_atom_feat=n_feat, batch_size=batch_size, learning_rate=0.001, use_queue=False, mode="regression", learning_rate=0.003, mode="classification", dropouts=0.0, model_dir=model_dir) reloaded_model.restore() # Check predictions match on random sample predmols = ["CCCC", "CCCCCO", "CCCCC"] Xpred = featurizer(predmols) predset = dc.data.NumpyDataset(Xpred) predset = transformer.transform(predset) origpred = model.predict(predset) reloadpred = reloaded_model.predict(predset) assert np.all(origpred == reloadpred) #Eval model on train scores = reloaded_model.evaluate(dataset, [regression_metric]) assert scores[regression_metric.name] > .1 scores = reloaded_model.evaluate(dataset, [classification_metric]) assert scores[classification_metric.name] > .6 ## TODO: THIS IS FAILING! #def test_weave_classification_reload_alt(): # """Test weave model can be reloaded.""" # np.random.seed(123) # tf.random.set_seed(123) # n_tasks = 1 # # # Load mini log-solubility dataset. # featurizer = dc.feat.WeaveFeaturizer() # tasks = ["outcome"] # mols = ["C", "CO", "CC"] # n_samples = len(mols) # X = featurizer(mols) # y = np.random.randint(2, size=(n_samples, n_tasks)) # dataset = dc.data.NumpyDataset(X, y) # # classification_metric = dc.metrics.Metric(dc.metrics.roc_auc_score) # # batch_size = 10 # # model_dir = tempfile.mkdtemp() # model = dc.models.WeaveModel( # n_tasks, # batch_size=batch_size, # learning_rate=0.0003, # mode="classification", # dropouts=0.0, # model_dir=model_dir) # # # Fit trained model # model.fit(dataset, nb_epoch=30) # # # Eval model on train # scores = model.evaluate(dataset, [classification_metric]) # assert scores[classification_metric.name] > .9 # # # Custom save # save_dir = tempfile.mkdtemp() # model.model.save(save_dir) # # from tensorflow import keras # reloaded = keras.models.load_model(save_dir) # # reloaded_model = dc.models.WeaveModel( # n_tasks, # batch_size=batch_size, # learning_rate=0.0003, # mode="classification", # dropouts=0.0, # model_dir=model_dir) # #reloaded_model.restore() # reloaded_model.model = reloaded # # # Check predictions match on random sample # predmols = ["CCCC", "CCCCCO", "CCCCC"] # Xpred = featurizer(predmols) # predset = dc.data.NumpyDataset(Xpred) # origpred = model.predict(predset) # reloadpred = reloaded_model.predict(predset) # assert np.all(origpred == reloadpred) # # # Eval model on train # scores = reloaded_model.evaluate(dataset, [classification_metric]) # assert scores[classification_metric.name] > .9 # # ## TODO: THIS IS FAILING! #@pytest.mark.slow #def test_weave_classification_reload(): Loading
