Merge pull request #2407 from peastman/tests (6cf77a92) · Commits · 钟慕尧 / deepchem

deepchem/models/tests/test_graph_models.py

+6 −16

Original line number	Diff line number	Diff line
		@@ -165,7 +165,6 @@ def test_dag_model():
		tasks, dataset, transformers, metric = get_dataset('classification',
		'GraphConv')

		batch_size = 10
		max_atoms = max([mol.get_num_atoms() for mol in dataset.X])
		transformer = dc.trans.DAGTransformer(max_atoms=max_atoms)
		dataset = transformer.transform(dataset)
		@@ -174,11 +173,9 @@ def test_dag_model():
		len(tasks),
		max_atoms=max_atoms,
		mode='classification',
		learning_rate=0.03,
		batch_size=batch_size,
		use_queue=False)
		learning_rate=0.001)

		model.fit(dataset, nb_epoch=40)
		model.fit(dataset, nb_epoch=30)
		scores = model.evaluate(dataset, [metric], transformers)
		assert scores['mean-roc_auc_score'] >= 0.9

		@@ -190,20 +187,14 @@ def test_dag_regression_model():
		tf.random.set_seed(1234)
		tasks, dataset, transformers, metric = get_dataset('regression', 'GraphConv')

		batch_size = 10
		max_atoms = max([mol.get_num_atoms() for mol in dataset.X])
		transformer = dc.trans.DAGTransformer(max_atoms=max_atoms)
		dataset = transformer.transform(dataset)

		model = DAGModel(
		len(tasks),
		max_atoms=max_atoms,
		mode='regression',
		learning_rate=0.03,
		batch_size=batch_size,
		use_queue=False)
		len(tasks), max_atoms=max_atoms, mode='regression', learning_rate=0.003)

		model.fit(dataset, nb_epoch=1200)
		model.fit(dataset, nb_epoch=100)
		scores = model.evaluate(dataset, [metric], transformers)
		assert scores['mean_absolute_error'] < 0.15

		@@ -250,7 +241,6 @@ def test_dag_regression_uncertainty():
		def test_mpnn_model():
		tasks, dataset, transformers, metric = get_dataset('classification', 'Weave')

		batch_size = 10
		model = MPNNModel(
		len(tasks),
		mode='classification',
		@@ -259,9 +249,9 @@ def test_mpnn_model():
		n_pair_feat=14,
		T=1,
		M=1,
		batch_size=batch_size)
		learning_rate=0.0005)

		model.fit(dataset, nb_epoch=40)
		model.fit(dataset, nb_epoch=150)
		scores = model.evaluate(dataset, [metric], transformers)
		assert scores['mean-roc_auc_score'] >= 0.9

deepchem/models/tests/test_mpnn.py

+2 −5

Original line number	Diff line number	Diff line
		@@ -27,7 +27,7 @@ def test_mpnn_regression():

		# initialize models
		n_tasks = len(tasks)
		model = MPNNModel(mode='regression', n_tasks=n_tasks, batch_size=10)
		model = MPNNModel(mode='regression', n_tasks=n_tasks, learning_rate=0.0005)

		# overfit test
		model.fit(dataset, nb_epoch=400)
		@@ -61,10 +61,7 @@ def test_mpnn_classification():
		# initialize models
		n_tasks = len(tasks)
		model = MPNNModel(
		mode='classification',
		n_tasks=n_tasks,
		batch_size=10,
		learning_rate=0.001)
		mode='classification', n_tasks=n_tasks, learning_rate=0.0005)

		# overfit test
		model.fit(dataset, nb_epoch=200)

deepchem/models/tests/test_pretrained_torch.py

+3 −3

Original line number	Diff line number	Diff line
		@@ -58,7 +58,7 @@ class TestPretrainedTorch(unittest.TestCase):
		for idx, dest_var in enumerate(dest_vars):
		source_var = source_vars[idx]
		assignment_map[source_var] = dest_var
		value_map[source_var] = source_var.detach().numpy()
		value_map[source_var] = source_var.detach().cpu().numpy()

		dest_model.load_from_pretrained(
		source_model=source_model,
		@@ -66,8 +66,8 @@ class TestPretrainedTorch(unittest.TestCase):
		value_map=value_map)

		for source_var, dest_var in assignment_map.items():
		source_val = source_var.detach().numpy()
		dest_val = dest_var.detach().numpy()
		source_val = source_var.detach().cpu().numpy()
		dest_val = dest_var.detach().cpu().numpy()
		np.testing.assert_array_almost_equal(source_val, dest_val)

		def test_restore_equivalency(self):

deepchem/models/tests/test_weave_models.py

+6 −7

Original line number	Diff line number	Diff line
		@@ -58,7 +58,7 @@ def test_compute_features_on_infinity_distance():
		"trainable": True,
		"renorm": True
		},
		learning_rage=0.0005)
		learning_rate=0.0005)
		atom_feat, pair_feat, pair_split, atom_split, atom_to_pair = model.compute_features_on_batch(
		X)

		@@ -95,7 +95,7 @@ def test_compute_features_on_distance_1():
		"trainable": True,
		"renorm": True
		},
		learning_rage=0.0005)
		learning_rate=0.0005)
		atom_feat, pair_feat, pair_split, atom_split, atom_to_pair = model.compute_features_on_batch(
		X)

		@@ -131,10 +131,9 @@ def test_weave_model():
		len(tasks),
		batch_size=batch_size,
		mode='classification',
		final_conv_activation_fn=None,
		dropouts=0,
		learning_rage=0.0003)
		model.fit(dataset, nb_epoch=100)
		learning_rate=0.002)
		model.fit(dataset, nb_epoch=250)
		scores = model.evaluate(dataset, [metric], transformers)
		assert scores['mean-roc_auc_score'] >= 0.9

		@@ -178,7 +177,7 @@ def test_weave_regression_model():
		# "trainable": True,
		# "renorm": True
		# },
		# learning_rage=0.0005)
		# learning_rate=0.0005)
		# model.fit(dataset, nb_epoch=200)
		# transformers = []
		# metric = dc.metrics.Metric(
		@@ -205,7 +204,7 @@ def test_weave_fit_simple_distance_1():
		"trainable": True,
		"renorm": True
		},
		learning_rage=0.0005)
		learning_rate=0.0005)
		model.fit(dataset, nb_epoch=200)
		transformers = []
		metric = dc.metrics.Metric(

deepchem/rl/tests/test_a2c.py

+4 −2

Original line number	Diff line number	Diff line
		@@ -77,12 +77,14 @@ class TestA2C(unittest.TestCase):
		a2c.fit(100000)

		# It should have learned that the expected value is very close to zero, and that the best
		# action is to walk away.
		# action is to walk away. (To keep the test fast, we allow that to be either of the two
		# top actions).

		action_prob, value = a2c.predict([[0]])
		assert -0.5 < value[0] < 0.5
		assert action_prob.argmax() == 37
		assert a2c.select_action([[0]], deterministic=True) == 37
		assert 37 in np.argsort(action_prob.flatten())[-2:]
		assert a2c.select_action([[0]], deterministic=True) == action_prob.argmax()

		# Verify that we can create a new A2C object, reload the parameters from the first one, and
		# get the same result.

Admin message