Progressive debugging (0ad58e3d) · Commits · 钟慕尧 / deepchem

deepchem/metrics/init.py

+16 −0

Original line number	Diff line number	Diff line
		@@ -181,6 +181,22 @@ class Metric(object):
		metric_value = self.compute_singletask_metric(
		y_task, y_pred_task, w_task)
		computed_metrics.append(metric_value)
		##################################################################### DEBUG
		# print("n_tasks")
		# print(n_tasks)
		# print("len(computed_metrics)")
		# print(len(computed_metrics))
		# print("computed_metrics")
		# print(computed_metrics)
		# print("len(computed_metrics)")
		# print(len(computed_metrics))
		# print("computed_metrics")
		# print(computed_metrics)
		# print("len(computed_metrics)")
		# print(len(computed_metrics))
		# print("y_true.shape, y_pred.shape")
		# print(y_true.shape, y_pred.shape)
		##################################################################### DEBUG
		log("computed_metrics: %s" % str(computed_metrics), self.verbosity)
		if n_tasks == 1:
		computed_metrics = computed_metrics[0]

+8 −0

Original line number	Diff line number	Diff line
		@@ -177,6 +177,14 @@ class Model(object):
		# The iterbatches does padding with zero-weight examples on the last batch.
		# Remove padded examples.
		n_samples = len(dataset)
		############################################################ DEBUG
		#print("y_pred[:10]")
		#print(y_pred[:10])
		#print("dataset.y[:10]")
		#print(dataset.y[:10])
		#print("y_pred.shape")
		#print(y_pred.shape)
		############################################################ DEBUG
		y_pred = np.reshape(y_pred, (n_samples, n_tasks))
		# Special case to handle singletasks.
		if n_tasks == 1:

+2 −2

Original line number	Diff line number	Diff line
		@@ -120,7 +120,7 @@ def dropout(tensor, dropout_prob, training=True, training_only=True):
		return tensor


		def fully_connected_layer(tensor, size, weight_init=None, bias_init=None,
		def fully_connected_layer(tensor, size=None, weight_init=None, bias_init=None,
		name=None):
		"""Fully connected layer.

		@@ -129,7 +129,7 @@ def fully_connected_layer(tensor, size, weight_init=None, bias_init=None,
		tensor: tf.Tensor
		Input tensor.
		size: int
		Number of nodes in this layer.
		Number of output nodes for this layer.
		weight_init: float
		Weight initializer.
		bias_init: float

+641 −287

File changed.

Preview size limit exceeded, changes collapsed.

+16 −6

Original line number	Diff line number	Diff line
		@@ -757,22 +757,32 @@ class TestOverfit(test_util.TensorFlowTestCase):
		np.random.seed(123)
		ids = np.arange(n_samples)
		X = np.random.rand(n_samples, n_features)
		y = np.zeros((n_samples, n_tasks))
		y = np.ones((n_samples, n_tasks))
		w = np.ones((n_samples, n_tasks))

		dataset = dc.data.NumpyDataset(X, y, w, ids)

		regression_metric = dc.metrics.Metric(
		dc.metrics.mean_squared_error, task_averager=np.mean)
		metric = dc.metrics.Metric(dc.metrics.rms_score, task_averager=np.mean)
		model = dc.models.ProgressiveMultitaskRegressor(
		n_tasks, n_features, layer_sizes=[50], bypass_layer_sizes=[10],
		dropouts=[0.], learning_rate=0.003, weight_init_stddevs=[.1], seed=123,
		alpha_init_stddevs=[.02], batch_size=n_samples, verbosity="high")

		# Fit trained model
		model.fit(dataset, nb_epoch=25)
		model.fit(dataset, nb_epoch=20)
		model.save()

		# Eval model on train
		scores = model.evaluate(dataset, [regression_metric])
		assert scores[regression_metric.name] < .2
		scores = model.evaluate(dataset, [metric])
		print("scores")
		print(scores)
		y_pred = model.predict(dataset)
		print("y")
		print(y)
		print("y_pred")
		print(y_pred)
		print("w")
		print(w)
		print("metric.compute_metric(y, y_pred, w)")
		print(metric.compute_metric(y, y_pred, w))
		assert scores[metric.name] < .2