Fixed examples using MultitaskFitTranformRegressor (501a71bd) · Commits · 钟慕尧 / deepchem

deepchem/models/fcnet.py

+10 −11

Original line number	Diff line number	Diff line
		@@ -299,7 +299,7 @@ class MultitaskFitTransformRegressor(MultitaskRegressor):
		>>> fit_transformers = [dc.trans.CoulombFitTransformer(dataset)]
		>>> model = dc.models.MultitaskFitTransformRegressor(n_tasks, [n_features, n_features],
		... dropouts=[0.], learning_rate=0.003, weight_init_stddevs=[np.sqrt(6)/np.sqrt(1000)],
		... batch_size=n_samples, fit_transformers=fit_transformers, n_evals=1)
		... batch_size=n_samples, fit_transformers=fit_transformers)
		>>> model.n_features
		12
		"""
		@@ -308,7 +308,6 @@ class MultitaskFitTransformRegressor(MultitaskRegressor):
		n_tasks,
		n_features,
		fit_transformers=[],
		n_evals=1,
		batch_size=50,
		**kwargs):
		"""Create a MultitaskFitTransformRegressor.
		@@ -324,11 +323,8 @@ class MultitaskFitTransformRegressor(MultitaskRegressor):
		number of features
		fit_transformers: list
		List of dc.trans.FitTransformer objects
		n_evals: int
		Number of evalations per example at predict time
		"""
		self.fit_transformers = fit_transformers
		self.n_evals = n_evals

		# Run fit transformers on dummy dataset to determine n_features after transformation

		@@ -362,17 +358,20 @@ class MultitaskFitTransformRegressor(MultitaskRegressor):
		if mode == 'fit':
		for transformer in self.fit_transformers:
		X_b = transformer.X_transform(X_b)
		yield ([X_b], [y_b], [w_b])
		if mode == 'predict':
		dropout = np.array(0.0)
		else:
		dropout = np.array(1.0)
		yield ([X_b, dropout], [y_b], [w_b])

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

		def transform_generator():
		for inputs, labels, weights in generator:
		for i in range(self.n_evals):
		X_t = inputs[0]
		for transformer in self.fit_transformers:
		X_t = transformer.X_transform(X_t)
		yield ([X_t], labels, weights)
		yield ([X_t] + inputs[1:], labels, weights)

		return super(MultitaskFitTransformRegressor, self).predict_on_generator(
		transform_generator(), transformers, outputs)

deepchem/models/tests/test_overfit.py

+1 −1

Original line number	Diff line number	Diff line
		@@ -195,7 +195,7 @@ class TestOverfit(test_util.TensorFlowTestCase):
		regression_metric = dc.metrics.Metric(dc.metrics.mean_squared_error)
		model = dc.models.MultitaskFitTransformRegressor(
		n_tasks, [n_features, n_features],
		dropouts=[0.],
		dropouts=[0.01],
		weight_init_stddevs=[np.sqrt(6) / np.sqrt(1000)],
		batch_size=n_samples,
		fit_transformers=fit_transformers,

examples/qm7/qm7_tf_model.py

+0 −1

Original line number	Diff line number	Diff line
		@@ -33,7 +33,6 @@ model = dc.models.MultitaskFitTransformRegressor(
		layer_sizes=[400, 100, 100],
		dropouts=[0.01, 0.01, 0.01],
		fit_transformers=fit_transformers,
		n_evals=10,
		seed=123)

		# Fit trained model

examples/qm7/qm7b_tf_model.py

+0 −2

Original line number	Diff line number	Diff line
		@@ -29,12 +29,10 @@ model = dc.models.MultitaskFitTransformRegressor(
		layer_sizes=[400, 100, 100],
		dropouts=[0.01, 0.01, 0.01],
		fit_transformers=fit_transformers,
		n_evals=10,
		seed=123)

		# Fit trained model
		model.fit(train_dataset, nb_epoch=50)
		model.save()

		train_scores = model.evaluate(train_dataset, regression_metric, transformers)
		print("Train scores [kcal/mol]")

examples/qm8/qm8_tf_model.py

+0 −2

Original line number	Diff line number	Diff line
		@@ -29,12 +29,10 @@ model = dc.models.MultitaskFitTransformRegressor(
		layer_sizes=[400, 100, 100],
		dropouts=[0.01, 0.01, 0.01],
		fit_transformers=fit_transformers,
		n_evals=10,
		seed=123)

		# Fit trained model
		model.fit(train_dataset, nb_epoch=50)
		model.save()

		train_scores = model.evaluate(train_dataset, regression_metric, transformers)
		print("Train scores [kcal/mol]")

Admin message