def test_ignores_exceptions() -> None:
with unittest.mock.patch(
"kerasltisubmission.LTIProvider.submit", autospec=True
) as mocked_submit:
mocked_submit.side_effect = ValueError("Something went wrong")
model = unittest.mock.MagicMock()
Submission(user_token="12", assignment_id=2, model=model).submit()
def test_construction() -> None:
with unittest.mock.patch("kerasltisubmission.LTIProvider",
autospec=True) as mocked_provider:
mock_provider = unittest.mock.MagicMock()
mock_submit = unittest.mock.MagicMock()
mock_provider.submit = mock_submit
mocked_provider.return_value = mock_provider
model = unittest.mock.MagicMock()
Submission(user_token="12", assignment_id=2, model=model).submit()
mocked_provider.assert_called_with(
input_api_endpoint="http://localhost:8080",
submission_api_endpoint="http://localhost:8080/submit",
user_token="12",
)
assert len(mock_submit.call_args_list) == 1
args, kwargs = mock_submit.call_args_list[0]
assert args[0].assignment_id == 2
assert args[0].model == model
from tensorflow import keras
from deeplearning2020 import Submission
data = keras.datasets.mnist
(train_images, train_labels), (test_images, test_labels) = data.load_data()
train_images = train_images / 255.0
test_images = test_images / 255.0
total_classes = 10
train_vec_labels = keras.utils.to_categorical(train_labels, total_classes)
test_vec_labels = keras.utils.to_categorical(test_labels, total_classes)
model = keras.Sequential([
keras.layers.Flatten(input_shape=(28, 28)),
keras.layers.Dense(128, activation="sigmoid"),
keras.layers.Dense(10, activation="sigmoid"),
])
model.compile(optimizer="sgd", loss="mean_squared_error", metrics=["accuracy"])
model.fit(train_images, train_vec_labels, epochs=2, verbose=True)
eval_loss, eval_accuracy = model.evaluate(test_images,
test_vec_labels,
verbose=False)
print("Model accuracy: %.2f" % eval_accuracy)
Submission(user_token="<your-token>", assignment_id=2, model=model).submit()
# 9 Ankle boot
(train_images, train_label), (test_images,
test_label) = fashion_mnist.load_data()
train_images = train_images / 255
test_images = test_images / 255
total_classes = 10
train_label_vectorized = keras.utils.to_categorical(train_label, total_classes)
test_label_vectorized = keras.utils.to_categorical(test_label, total_classes)
model = keras.Sequential([
keras.layers.Flatten(input_shape=(28, 28)), # input
keras.layers.Dense(128, activation='relu'), # hidden
keras.layers.Dense(128, activation='relu'), # hidden
keras.layers.Dense(total_classes, activation='sigmoid') # output
])
model.add(LeakyReLU(alpha=0.1))
model.compile(optimizer='sgd', loss='mean_squared_error', metrics=['accuracy'])
model.fit(train_images, train_label_vectorized, epochs=40)
eval_loss, eval_accuracy = model.evaluate(test_images, test_label_vectorized)
print(eval_loss, eval_accuracy)
Submission('a70a2614a4468a25eb66a1113c846e31', '2', model).submit()
from tensorflow import keras
from deeplearning2020 import Submission
data = keras.datasets.mnist
(train_images, train_labels), (test_images, test_labels) = data.load_data()
train_images = train_images / 255.0
test_images = test_images / 255.0
total_classes = 10
train_vec_labels = keras.utils.to_categorical(train_labels, total_classes)
test_vec_labels = keras.utils.to_categorical(test_labels, total_classes)
model = keras.Sequential([
keras.layers.Flatten(input_shape=(28, 28)),
keras.layers.Dense(128, activation="sigmoid"),
keras.layers.Dense(10, activation="sigmoid"),
])
model.compile(optimizer="sgd", loss="mean_squared_error", metrics=["accuracy"])
model.fit(train_images, train_vec_labels, epochs=2, verbose=True)
eval_loss, eval_accuracy = model.evaluate(test_images,
test_vec_labels,
verbose=False)
print("Model accuracy: %.2f" % eval_accuracy)
Submission(user_token="637a0109-1d88-4afb-a759-8747ab991c2d",
assignment_id=2,
model=model).submit()