Add accuracy by changing outputs on the model (0b77c989) · Commits · 钟慕尧 / deepchem

deepchem/models/tensorgraph/models/scscore.py

+2 −2

Original line number	Diff line number	Diff line
		@@ -80,9 +80,9 @@ class ScScoreModel(TensorGraph):
		self.add_output(Sigmoid(readout_m1) * 4 + 1)
		self.add_output(Sigmoid(readout_m2) * 4 + 1)

		difference = readout_m1 - readout_m2
		self.difference = readout_m1 - readout_m2
		label = Label(shape=(None, 1))
		loss = HingeLoss(in_layers=[label, difference])
		loss = HingeLoss(in_layers=[label, self.difference])
		self.my_task_weights = Weights(shape=(None, 1))
		loss = WeightedError(in_layers=[loss, self.my_task_weights])
		self.set_loss(loss)

+11 −2

Original line number	Diff line number	Diff line
		@@ -301,6 +301,12 @@ class TestEstimators(unittest.TestCase):
		# Create a TensorGraph model.

		model = dc.models.ScScoreModel(n_features, dropouts=0)
		model.outputs.clear()
		model.outputs.append(model.difference)

		def accuracy(labels, predictions, weights):
		predictions = tf.nn.relu(tf.sign(predictions))
		return tf.metrics.accuracy(labels, predictions, weights)

		# Create an estimator from it.

		@@ -309,7 +315,9 @@ class TestEstimators(unittest.TestCase):
		weight_col = tf.feature_column.numeric_column('weights', shape=(1,))

		estimator = model.make_estimator(
		feature_columns=[x_col1, x_col2], metrics={}, weight_column=weight_col)
		feature_columns=[x_col1, x_col2],
		metrics={'accuracy': accuracy},
		weight_column=weight_col)

		# Train the model.

		@@ -318,4 +326,5 @@ class TestEstimators(unittest.TestCase):
		# Evaluate the model.

		results = estimator.evaluate(input_fn=lambda: input_fn(1))
		assert results['loss'] < 1e-4
		assert results['loss'] < 0.5
		assert results['accuracy'] > 0.6