def get_cifar10_date_batches(batch_size_train = 512, batch_size_test = 512, transforms = [Crop(32, 32), FlipLR(), Cutout(8, 8)]):
CIAFR10_DATA_DIR = './data'
dataset_10 = cifar10(root = CIAFR10_DATA_DIR)
train_set_10 = list(zip(transpose(normalise(pad(dataset_10['train']['data'], 4))), dataset_10['train']['labels']))
test_set_10 = list(zip(transpose(normalise(dataset_10['test']['data'])), dataset_10['test']['labels']))
train_batches_cifar10 = Batches(Transform(train_set_10, transforms), batch_size_train, shuffle=True,
set_random_choices=True, drop_last=True)
test_batches_cifar10 = Batches(test_set_10, batch_size_test, shuffle=False, drop_last=False)
return train_batches_cifar10, test_batches_cifar10
def get_cifar100_date_batches(batch_size = 512, transforms = [Crop(32, 32), FlipLR(), Cutout(8, 8)]):
CIAFR100_DATA_DIR = './data'
dataset_100 = cifar100(root = CIAFR100_DATA_DIR)
train_set_100 = list(zip(transpose(normalise(pad(dataset_100['train']['data'], 4), mean=[0.4914, 0.4822, 0.4465], std=[0.2675, 0.2565, 0.2761])), dataset_100['train']['labels']))
test_set_100 = list(zip(transpose(normalise(dataset_100['test']['data'], mean=[0.4914, 0.4822, 0.4465], std=[0.2675, 0.2565, 0.2761])), dataset_100['test']['labels']))
train_batches_cifar100 = Batches(Transform(train_set_100, transforms), batch_size, shuffle=True,
set_random_choices=True, drop_last=True)
test_batches_cifar100 = Batches(test_set_100, batch_size, shuffle=False, drop_last=False)
return train_batches_cifar100, test_batches_cifar100
def preprocess(experiment, dataset):
with timed("Preprocessing training data"):
train_set = list(
zip(transpose(normalise(pad(dataset['train']['data'], 4))),
dataset['train']['labels']))
with timed("Preprocessing test data"):
test_set = list(
zip(transpose(normalise(dataset['test']['data'])),
dataset['test']['labels']))
return train_set, test_set
def normal_to_world(self, normal):
normal = multiply_tuple(transpose(self.__inverse_transform), normal)
normal[3] = 0
normal = normalize(normal)
if self.parent:
normal = self.parent.normal_to_world(normal)
return normal
def main():
print("What happens when you invert the identity matrix?")
a = identity_matrix()
print(inverse(a))
print()
print("What do you get when you multiply a matrix by its inverse?")
a = identity_matrix()
a[1][0] = 2
a[0][3] = -5
b = multiply_matrix(a, inverse(a))
print(b)
print()
print(
"Is there any difference between the inverse of the transpose of a matrix, and the transpose of the inverse?"
)
c = inverse(transpose(a))
d = transpose(inverse(a))
print(c)
print(d)
print()
print(
"Remember how multiplying the identity matrix by a tuple gives you the tuple, unchanged?"
)
t = tuple_4d(5, 4, 3, 1)
i = identity_matrix()
c = multiply_tuple(i, t)
print(c)
print()
print(
"Now, try changing any single element of the identity matrix to a different number, and then multiplying it by a tuple. What happens to the tuple?"
)
i[1][1] = 2
i[2][2] = 3
c = multiply_tuple(i, t)
print(c)
def step_impl(context):
expected = matrix(4, 4)
for r in range(4):
for c in range(4):
expected[r][c] = float(context.table[r][c])
assert transpose(context.a) == expected
def step_impl(context):
context.a = transpose(identity_matrix())