def test():
data = """
1-3 a: abcde
1-3 b: cdefg
2-9 c: ccccccccc"""
run_tests(p1, [(data, 2)])
run_tests(p2, [(data, 1)])
def main(reg_lambda,
learning_rate,
loss_function,
regulariser,
niterations=10,
enable_test_set_scoring=False,
**kwargs):
global data
# Type of features to use. This can be set to 'bigram' or 'unigram+bigram' to use
# bigram features instead of or in addition to unigram features.
# Not required for assignment.
feature_type = 'unigram'
# First test the parts to be implemented and warn if something's wrong.
print('=============')
print('SANITY CHECKS')
print('=============')
print()
util.run_tests()
# Load the data.
training_data, val_data, test_data, data = load_data(feature_type)
# Train the classifier.
print('Starting training.')
weights, bias, training_log = train(training_data, val_data, loss_function,
regulariser, reg_lambda, learning_rate,
niterations)
print('Training completed.')
print()
print('=====================')
print('MODEL CHARACTERISTICS')
print('=====================')
print()
# Display some useful statistics about the model and the training process.
title = 'Data set: %s - Regulariser(%s): %g - Learning rate: %g ' \
'- Loss Function: %s' % (
data.name, regulariser, reg_lambda, learning_rate,
loss_function)
print()
# Get final accuracy
val_predictions = predict(weights, bias, val_data)
val_accuracy = accuracy(val_data.labels, val_predictions)
print('Accuracy: %g' % val_accuracy)
util.show_stats(title,
training_log,
weights,
bias,
data.vocabulary,
top_n=1,
write_to_file="results.csv",
configuration={
'reg_lambda': reg_lambda,
'learning_rate': learning_rate,
'loss_function': loss_function,
'regulariser': regulariser,
'niterations': niterations,
'val_accuracy': val_accuracy
})
util.create_plots(title,
training_log,
weights,
log_keys=['training_loss_reg', 'val_loss'])
if enable_test_set_scoring:
# Check the performance on the test set.
test_loss = loss_function.unregularised_loss(weights, bias, test_data)
test_predictions = predict(weights, bias, test_data)
test_accuracy = accuracy(test_data.labels, test_predictions)
print()
print('====================')
print('TEST SET PERFORMANCE')
print('====================')
print()
print('Test loss: %g' % test_loss)
print('Test accuracy: %g' % test_accuracy)
def main():
# TRAINING HYPERPARAMETERS
# Modify the following lines to change the training hyperparameters.
# Regularisation strength
reg_lambda = 0.0
# Learning rate
learning_rate = 0.07
# Number of training iterations
niterations = 100
# Loss function to use (select one and comment out the other)
loss_function = LogisticLoss()
# loss_function = HingeLoss()
# Type of regularisation to use (select one and comment out the other)
# regulariser = L1Regulariser()
regulariser = L2Regulariser()
# This should only be enabled once you've decided on a final set of hyperparameters
enable_test_set_scoring = True
# Controls to use perceptron or not
enable_perceptron = True
# Type of features to use. This can be set to 'bigram' or 'unigram+bigram' to use
# bigram features instead of or in addition to unigram features.
# Not required for assignment.
feature_type = 'unigram'
# END OF HYPERPARAMETERS
# First test the parts to be implemented and warn if something's wrong.
print('=============')
print('SANITY CHECKS')
print('=============')
print()
util.run_tests()
# Load the data.
print()
print('===================')
print('CLASSIFIER TRAINING')
print('===================')
print()
print('Loading data sets...')
data_dir = './poldata/poldata.zip'
data = util.load_movie_data(data_dir)
data.select_feature_type(feature_type)
# Split the data set randomly into training, validation and test sets.
training_data, val_data, test_data = data.train_val_test_split()
# Train the classifier.
print('Starting training.')
if enable_perceptron:
weights, bias, training_log = train_perceptron(training_data, val_data,
niterations)
else:
weights, bias, training_log = train(training_data, val_data,
loss_function, regulariser,
reg_lambda, learning_rate,
niterations)
print('Training completed.')
print()
print('=====================')
print('MODEL CHARACTERISTICS')
print('=====================')
print()
# Display some useful statistics about the model and the training process.
title = 'Data set: %s - Regulariser: %g - Learning rate: %g' % (
data.name, reg_lambda, learning_rate)
print()
util.show_stats(title,
training_log,
weights,
bias,
data.vocabulary,
top_n=20)
util.create_plots(title,
training_log,
weights,
log_keys=['training_loss_reg', 'val_loss'])
if enable_test_set_scoring:
# Check the performance on the test set.
test_loss = loss_function.unregularised_loss(weights, bias, test_data)
test_predictions = predict(weights, bias, test_data)
test_accuracy = accuracy(test_data.labels, test_predictions)
print()
print('====================')
print('TEST SET PERFORMANCE')
print('====================')
print()
print('Test loss: %g' % test_loss)
print('Test accuracy: %g' % test_accuracy)
c.add(s[j])
j += 1
l = max(l, j - i)
else:
c.remove(s[i])
i += 1
return l
def longestSubstringWithoutRepeatingCharacters_slidingWindowOptimized(s):
n = len(s)
map = dict()
i, l = 0, 0
for j in range(n):
if s[j] in map:
i = max(map.get(s[j]), i)
l = max(l, j - i + 1)
map[s[j]] = j + 1
return l
if __name__ == "__main__":
solutions = [
longestSubstringWithoutRepeatingCharacters_bruteForce,
longestSubstringWithoutRepeatingCharacters_slidingWindow,
longestSubstringWithoutRepeatingCharacters_slidingWindowOptimized
]
tests = [(("abcabcbb", ), 3), (("bbbbb", ), 1), (("pwwkew", ), 3),
(("", ), 0)]
util.run_tests(solutions, tests)
#!/usr/bin/env python
from util import run_tests
from test_suite import *
if __name__ == "__main__":
tests = arithmetic_tests + \
mod_tests + \
arithmetic_boolean_tests + \
nary_arithmetic_tests + \
bitwise_tests + \
boolean_tests + \
nary_boolean_tests + \
const_arith_tests + \
const_bool_tests + \
ite_tests + \
div_tests + \
array_tests + \
c_array_tests + \
misc_tests + \
biomatch_tests + \
kmeans_tests
# shift_tests
# TODO: add support for return value - int promotion
# unsigned_arithmetic_tests + \
run_tests('c', tests)
#!/usr/bin/env python
from util import run_tests
from test_suite import *
if __name__ == "__main__":
tests = arithmetic_tests + \
arithmetic_boolean_tests + \
nary_arithmetic_tests + \
bitwise_tests + \
boolean_tests + \
nary_boolean_tests + \
const_arith_tests + \
const_bool_tests + \
loop_tests + \
ite_tests + \
function_tests + \
misc_tests
# shift_tests + \
# arr_tests + \
run_tests('zok', tests)
L -= 1
R += 1
return R - L - 1
def longestPalindromicSubstring_expandAroundCenter(s):
if len(s) == 0:
return s
n = len(s)
start, end = 0, 0
for i in range(n):
l1 = expandAroundCenter(s, i, i)
l2 = expandAroundCenter(s, i, i + 1)
l = max(l1, l2)
if l > end - start:
start = i - (l - 1) // 2
end = i + l // 2
return s[start:end + 1]
if __name__ == "__main__":
solutions = [
longestPalindromicSubstring_bruteForce,
longestPalindromicSubstring_longestCommonSubstring,
longestPalindromicSubstring_dynamicProgramming,
longestPalindromicSubstring_expandAroundCenter
]
tests = [(("babad", ), "bab"), (("cbbd", ), "bb"), (("a", ), "a"),
(("ac", ), "a"), (("bb", ), "bb"), (("aacabdkacaa", ), "aca")]
util.run_tests(solutions, tests, equals=lambda a, b: len(a) == len(b))
elif l2 is None:
l1.val += remainder
remainder = l1.val // 10
l1.val %= 10
if remainder == 0:
break
l1_prev = l1
l1 = l1.next
else:
l1.val += l2.val + remainder
remainder = l1.val // 10
l1.val %= 10
l1_prev = l1
l1 = l1.next
l2 = l2.next
if remainder != 0 and l1_prev is not None:
l1_prev.next = ListNode(remainder)
return root
if __name__ == "__main__":
solutions = [addTwoNumbers]
tests = [(([2, 4, 3], [5, 6, 4]), [7, 0, 8]), (([0], [0]), [0]),
(([9, 9, 9, 9, 9, 9, 9], [9, 9, 9, 9]), [8, 9, 9, 9, 0, 0, 0, 1]),
(([2, 4, 9], [5, 6, 4, 9]), [7, 0, 4, 0, 1])]
for i, (inputs, expected) in enumerate(tests):
inputs = tuple(ListNode.toLinkedList(x) for x in inputs)
expected = ListNode.toLinkedList(expected)
tests[i] = (inputs, expected)
util.run_tests(solutions, tests, equals=ListNode.equals)