def test_ni_sit_single_dataset_fixed_assignment(self):
mnist_train = MNIST('./data/mnist', train=True, download=True)
mnist_test = MNIST('./data/mnist', train=False, download=True)
ni_scenario_reference = ni_scenario(mnist_train,
mnist_test,
5,
shuffle=True,
seed=1234)
reference_assignment = ni_scenario_reference.\
train_exps_patterns_assignment
my_ni_scenario = ni_scenario(mnist_train,
mnist_test,
5,
shuffle=True,
seed=4321,
fixed_exp_assignment=reference_assignment)
self.assertEqual(ni_scenario_reference.n_experiences,
my_ni_scenario.n_experiences)
self.assertEqual(ni_scenario_reference.train_exps_patterns_assignment,
my_ni_scenario.train_exps_patterns_assignment)
self.assertEqual(ni_scenario_reference.exp_structure,
my_ni_scenario.exp_structure)
def test_ni_sit_single_dataset_reproducibility_data(self):
mnist_train = MNIST(root=expanduser("~") + "/.avalanche/data/mnist/",
train=True,
download=True)
mnist_test = MNIST(root=expanduser("~") + "/.avalanche/data/mnist/",
train=False,
download=True)
ni_scenario_reference = ni_scenario(mnist_train,
mnist_test,
5,
shuffle=True,
seed=1234)
rep_data = ni_scenario_reference.get_reproducibility_data()
my_ni_scenario = ni_scenario(mnist_train,
mnist_test,
0,
reproducibility_data=rep_data)
self.assertEqual(ni_scenario_reference.n_experiences,
my_ni_scenario.n_experiences)
self.assertEqual(ni_scenario_reference.train_exps_patterns_assignment,
my_ni_scenario.train_exps_patterns_assignment)
self.assertEqual(ni_scenario_reference.exp_structure,
my_ni_scenario.exp_structure)
def test_ni_sit_multi_dataset_merge(self):
split_mapping = [0, 1, 2, 3, 4, 0, 1, 2, 3, 4]
mnist_train = MNIST('./data/mnist', train=True, download=True)
mnist_test = MNIST('./data/mnist', train=False, download=True)
train_part1 = make_nc_transformation_subset(mnist_train, None, None,
range(5))
train_part2 = make_nc_transformation_subset(mnist_train, None, None,
range(5, 10))
train_part2 = AvalancheSubset(train_part2, class_mapping=split_mapping)
test_part1 = make_nc_transformation_subset(mnist_test, None, None,
range(5))
test_part2 = make_nc_transformation_subset(mnist_test, None, None,
range(5, 10))
test_part2 = AvalancheSubset(test_part2, class_mapping=split_mapping)
my_ni_scenario = ni_scenario([train_part1, train_part2],
[test_part1, test_part2],
5,
shuffle=True,
seed=1234,
balance_experiences=True)
self.assertEqual(5, my_ni_scenario.n_experiences)
self.assertEqual(10, my_ni_scenario.n_classes)
for batch_id in range(5):
self.assertEqual(
10, len(my_ni_scenario.classes_in_experience[batch_id]))
all_classes = set()
for batch_id in range(5):
all_classes.update(my_ni_scenario.classes_in_experience[batch_id])
self.assertEqual(10, len(all_classes))
def test_ni_sit_single_dataset(self):
mnist_train = MNIST(root=expanduser("~") + "/.avalanche/data/mnist/",
train=True,
download=True)
mnist_test = MNIST(root=expanduser("~") + "/.avalanche/data/mnist/",
train=False,
download=True)
my_ni_scenario = ni_scenario(mnist_train,
mnist_test,
5,
shuffle=True,
seed=1234,
balance_experiences=True)
self.assertEqual(5, my_ni_scenario.n_experiences)
self.assertEqual(10, my_ni_scenario.n_classes)
for batch_id in range(5):
self.assertEqual(
10, len(my_ni_scenario.classes_in_experience[batch_id]))
_, unique_count = torch.unique(torch.as_tensor(mnist_train.targets),
return_counts=True)
min_batch_size = torch.sum(unique_count //
my_ni_scenario.n_experiences).item()
max_batch_size = min_batch_size + my_ni_scenario.n_classes
pattern_count = 0
batch_info: NIExperience
for batch_id, batch_info in enumerate(my_ni_scenario.train_stream):
cur_train_set = batch_info.dataset
t = batch_info.task_label
self.assertEqual(0, t)
self.assertEqual(batch_id, batch_info.current_experience)
self.assertGreaterEqual(len(cur_train_set), min_batch_size)
self.assertLessEqual(len(cur_train_set), max_batch_size)
pattern_count += len(cur_train_set)
self.assertEqual(len(mnist_train), pattern_count)
self.assertEqual(1, len(my_ni_scenario.test_stream))
pattern_count = 0
for batch_id, batch_info in enumerate(my_ni_scenario.test_stream):
cur_test_set = batch_info.dataset
t = batch_info.task_label
self.assertEqual(0, t)
self.assertEqual(batch_id, batch_info.current_experience)
pattern_count += len(cur_test_set)
self.assertEqual(len(mnist_test), pattern_count)
def test_ni_sit_slicing(self):
mnist_train = MNIST(root=expanduser("~") + "/.avalanche/data/mnist/",
train=True,
download=True)
mnist_test = MNIST(root=expanduser("~") + "/.avalanche/data/mnist/",
train=False,
download=True)
my_ni_scenario = ni_scenario(mnist_train,
mnist_test,
5,
shuffle=True,
seed=1234)
experience: NIExperience
for batch_id, experience in enumerate(my_ni_scenario.train_stream):
self.assertEqual(batch_id, experience.current_experience)
self.assertIsInstance(experience, NIExperience)
self.assertEqual(1, len(my_ni_scenario.test_stream))
for batch_id, experience in enumerate(my_ni_scenario.test_stream):
self.assertEqual(batch_id, experience.current_experience)
self.assertIsInstance(experience, NIExperience)
iterable_slice = [3, 4, 1]
sliced_stream = my_ni_scenario.train_stream[iterable_slice]
self.assertIsInstance(sliced_stream, GenericScenarioStream)
self.assertEqual(len(iterable_slice), len(sliced_stream))
self.assertEqual('train', sliced_stream.name)
for batch_id, experience in enumerate(sliced_stream):
self.assertEqual(iterable_slice[batch_id],
experience.current_experience)
self.assertIsInstance(experience, NIExperience)
with self.assertRaises(IndexError):
# The test stream only has one element (the complete test set)
sliced_stream = my_ni_scenario.test_stream[iterable_slice]
iterable_slice = [0, 0, 0]
sliced_stream = my_ni_scenario.test_stream[iterable_slice]
self.assertIsInstance(sliced_stream, GenericScenarioStream)
self.assertEqual(len(iterable_slice), len(sliced_stream))
self.assertEqual('test', sliced_stream.name)
for batch_id, experience in enumerate(sliced_stream):
self.assertEqual(iterable_slice[batch_id],
experience.current_experience)
self.assertIsInstance(experience, NIExperience)
