def fit_better():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0:pick_size, :, :, :]
valid_labels = valid_labels[0:pick_size, :]
test_dataset = test_dataset[0:pick_size, :, :, :]
test_labels = test_labels[0:pick_size, :]
better_hps_list = etc_hp()
for better_hps in better_hps_list:
basic_hypers = {
'batch_size': int(float(better_hps[0])),
'patch_size': int(float(better_hps[4])),
'depth': int(float(better_hps[1])),
'num_hidden': int(float(better_hps[2])),
'layer_sum': int(float(better_hps[3])),
'starter_learning_rate': 0.1
}
# if basic_hypers['batch_size'] < 10:
# basic_hypers['batch_size'] = 10
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1]
for _ in range(basic_hypers['layer_sum'])]
conv_train(train_dataset, train_labels, valid_dataset, valid_labels,
test_dataset, test_labels, image_size, num_labels,
basic_hypers, stride_params)
else:
print('can not find better hypers')
def hp_train():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0:pick_size, :, :, :]
valid_labels = valid_labels[0:pick_size, :]
test_dataset = test_dataset[0:pick_size, :, :, :]
test_labels = test_labels[0:pick_size, :]
basic_hypers = {
'batch_size': 32,
'patch_size': 5,
'depth': 16,
'num_hidden': 64,
'layer_sum': 2
}
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
conv_train(train_dataset,
train_labels,
valid_dataset,
valid_labels,
test_dataset,
test_labels,
image_size,
num_labels,
basic_hypers,
stride_params,
drop=True,
lrd=False)
def fit_better():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0: pick_size, :, :, :]
valid_labels = valid_labels[0: pick_size, :]
test_dataset = test_dataset[0: pick_size, :, :, :]
test_labels = test_labels[0: pick_size, :]
better_hps_list = etc_hp()
for better_hps in better_hps_list:
basic_hypers = {
'batch_size': int(float(better_hps[0])),
'patch_size': int(float(better_hps[4])),
'depth': int(float(better_hps[1])),
'num_hidden': int(float(better_hps[2])),
'layer_sum': int(float(better_hps[3])),
'starter_learning_rate': 0.1
}
# if basic_hypers['batch_size'] < 10:
# basic_hypers['batch_size'] = 10
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
conv_train(
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels,
image_size, num_labels, basic_hypers, stride_params)
else:
print('can not find better hypers')
def fit_predict():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0: pick_size, :, :, :]
valid_labels = valid_labels[0: pick_size, :]
test_dataset = test_dataset[0: pick_size, :, :, :]
test_labels = test_labels[0: pick_size, :]
basic_hypers = {
'batch_size': 10,
'patch_size': 10,
'depth': 10,
'num_hidden': 10,
'layer_sum': 3
}
while(True):
basic_hypers['batch_size'] = random.randint(basic_hypers['batch_size'] / 2, basic_hypers['batch_size'] * 2)
basic_hypers['patch_size'] = random.randint(basic_hypers['patch_size'] / 2, basic_hypers['patch_size'] * 2)
basic_hypers['depth'] = random.randint(basic_hypers['depth'] / 2, basic_hypers['depth'] * 2)
basic_hypers['num_hidden'] = random.randint(basic_hypers['num_hidden'] / 2, basic_hypers['num_hidden'] * 2)
basic_hypers['layer_sum'] = random.randint(basic_hypers['layer_sum'] / 2, basic_hypers['layer_sum'] * 2)
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
conv_train(train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels,
image_size, num_labels, basic_hypers, stride_params, lrd=True)
def fit_better():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0:pick_size, :, :, :]
valid_labels = valid_labels[0:pick_size, :]
test_dataset = test_dataset[0:pick_size, :, :, :]
test_labels = test_labels[0:pick_size, :]
basic_hypers = {
'batch_size': 10,
'patch_size': 10,
'depth': 10,
'num_hidden': 10,
'layer_sum': 3,
'starter_learning_rate': 0.1
}
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
if basic_hypers['layer_sum'] > 3:
basic_hypers['layer_sum'] = 3
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
better_hps = conv_train(train_dataset, train_labels, valid_dataset,
valid_labels, test_dataset, test_labels,
image_size, num_labels, basic_hypers,
stride_params)
while valid_hp(better_hps):
basic_hypers = {
'batch_size': better_hps[0],
'patch_size': better_hps[4],
'depth': better_hps[1],
'num_hidden': better_hps[2],
'layer_sum': better_hps[3],
'starter_learning_rate': better_hps[5]
}
# if basic_hypers['batch_size'] < 10:
# basic_hypers['batch_size'] = 10
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
if basic_hypers['layer_sum'] > 3:
basic_hypers['layer_sum'] = 3
if basic_hypers['patch_size'] <= basic_hypers['layer_sum']:
basic_hypers['patch_size'] = basic_hypers['layer_sum']
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1]
for _ in range(basic_hypers['layer_sum'])]
better_hps = conv_train(train_dataset, train_labels, valid_dataset,
valid_labels, test_dataset, test_labels,
image_size, num_labels, basic_hypers,
stride_params)
else:
print('can not find better hypers')
def random_hp():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0:pick_size, :, :, :]
valid_labels = valid_labels[0:pick_size, :]
test_dataset = test_dataset[0:pick_size, :, :, :]
test_labels = test_labels[0:pick_size, :]
basic_hypers = {
'batch_size': 10,
'patch_size': 10,
'depth': 10,
'num_hidden': 10,
'layer_sum': 3,
'starter_learning_rate': 0.1
}
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
for _ in range(200):
random_hypers = {
'batch_size':
basic_hypers['batch_size'] + random.randint(-9, 30),
'patch_size':
basic_hypers['patch_size'] + random.randint(-5, 15),
'depth':
basic_hypers['depth'] + random.randint(-9, 10),
'num_hidden':
basic_hypers['num_hidden'] * random.randint(1, 10),
'layer_sum':
basic_hypers['layer_sum'] + random.randint(-1, 2),
'starter_learning_rate':
basic_hypers['starter_learning_rate'] * random.uniform(0, 5)
}
# if basic_hypers['batch_size'] < 10:
# basic_hypers['batch_size'] = 10
if random_hypers['patch_size'] > 28:
random_hypers['patch_size'] = 28
print('=' * 80)
print(random_hypers)
stride_params = [[1, 2, 2, 1]
for _ in range(random_hypers['layer_sum'])]
hps, loss_es = conv_train(train_dataset, train_labels, valid_dataset,
valid_labels, test_dataset, test_labels,
image_size, num_labels, random_hypers,
stride_params)
file_helper.write('hps', str(hps))
file_helper.write('losses', str(loss_es))
def fit_better():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0: pick_size, :, :, :]
valid_labels = valid_labels[0: pick_size, :]
test_dataset = test_dataset[0: pick_size, :, :, :]
test_labels = test_labels[0: pick_size, :]
basic_hypers = {
'batch_size': 10,
'patch_size': 10,
'depth': 10,
'num_hidden': 10,
'layer_sum': 3,
'starter_learning_rate': 0.1
}
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
if basic_hypers['layer_sum'] > 3:
basic_hypers['layer_sum'] = 3
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
better_hps = conv_train(train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels,
image_size, num_labels, basic_hypers, stride_params)
while valid_hp(better_hps):
basic_hypers = {
'batch_size': better_hps[0],
'patch_size': better_hps[4],
'depth': better_hps[1],
'num_hidden': better_hps[2],
'layer_sum': better_hps[3],
'starter_learning_rate': better_hps[5]
}
# if basic_hypers['batch_size'] < 10:
# basic_hypers['batch_size'] = 10
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
if basic_hypers['layer_sum'] > 3:
basic_hypers['layer_sum'] = 3
if basic_hypers['patch_size'] <= basic_hypers['layer_sum']:
basic_hypers['patch_size'] = basic_hypers['layer_sum']
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
better_hps = conv_train(
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels,
image_size, num_labels, basic_hypers, stride_params)
else:
print('can not find better hypers')
def fit_predict():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0:pick_size, :, :, :]
valid_labels = valid_labels[0:pick_size, :]
test_dataset = test_dataset[0:pick_size, :, :, :]
test_labels = test_labels[0:pick_size, :]
basic_hypers = {
'batch_size': 10,
'patch_size': 10,
'depth': 10,
'num_hidden': 10,
'layer_sum': 3
}
while (True):
basic_hypers['batch_size'] = random.randint(
basic_hypers['batch_size'] / 2, basic_hypers['batch_size'] * 2)
basic_hypers['patch_size'] = random.randint(
basic_hypers['patch_size'] / 2, basic_hypers['patch_size'] * 2)
basic_hypers['depth'] = random.randint(basic_hypers['depth'] / 2,
basic_hypers['depth'] * 2)
basic_hypers['num_hidden'] = random.randint(
basic_hypers['num_hidden'] / 2, basic_hypers['num_hidden'] * 2)
basic_hypers['layer_sum'] = random.randint(
basic_hypers['layer_sum'] / 2, basic_hypers['layer_sum'] * 2)
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1]
for _ in range(basic_hypers['layer_sum'])]
conv_train(train_dataset,
train_labels,
valid_dataset,
valid_labels,
test_dataset,
test_labels,
image_size,
num_labels,
basic_hypers,
stride_params,
lrd=True)
def fit_better():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0: pick_size, :, :, :]
valid_labels = valid_labels[0: pick_size, :]
test_dataset = test_dataset[0: pick_size, :, :, :]
test_labels = test_labels[0: pick_size, :]
basic_hypers = {
'batch_size': 10,
'patch_size': 10,
'depth': 10,
'num_hidden': 10,
'layer_sum': 3
}
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
ret, better_hps = conv_train(train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels,
image_size, num_labels, basic_hypers, stride_params, lrd=True)
while ret and valid_hp(better_hps):
basic_hypers = {
'batch_size': better_hps[0],
'patch_size': better_hps[4],
'depth': better_hps[1],
'num_hidden': better_hps[2],
'layer_sum': better_hps[3]
}
# if basic_hypers['batch_size'] < 10:
# basic_hypers['batch_size'] = 10
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
ret, better_hps = conv_train(
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels,
image_size, num_labels, basic_hypers, stride_params, lrd=True)
else:
print('can not find better hypers')
def hp_train():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0: pick_size, :, :, :]
valid_labels = valid_labels[0: pick_size, :]
test_dataset = test_dataset[0: pick_size, :, :, :]
test_labels = test_labels[0: pick_size, :]
basic_hypers = {
'batch_size': 32,
'patch_size': 5,
'depth': 16,
'num_hidden': 64,
'layer_sum': 2
}
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
conv_train(train_dataset, train_labels, valid_dataset, valid_labels, test_dataset,
test_labels,
image_size, num_labels, basic_hypers, stride_params, drop=True, lrd=False)
def fit_better():
image_size = 28
num_labels = 10
train_dataset, train_labels, valid_dataset, valid_labels, test_dataset, test_labels = \
format_mnist()
pick_size = 2048
valid_dataset = valid_dataset[0:pick_size, :, :, :]
valid_labels = valid_labels[0:pick_size, :]
test_dataset = test_dataset[0:pick_size, :, :, :]
test_labels = test_labels[0:pick_size, :]
basic_hypers = {
'batch_size': 10,
'patch_size': 10,
'depth': 10,
'num_hidden': 10,
'layer_sum': 3
}
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1] for _ in range(basic_hypers['layer_sum'])]
ret, better_hps = conv_train(train_dataset,
train_labels,
valid_dataset,
valid_labels,
test_dataset,
test_labels,
image_size,
num_labels,
basic_hypers,
stride_params,
lrd=True)
while ret and valid_hp(better_hps):
basic_hypers = {
'batch_size': better_hps[0],
'patch_size': better_hps[4],
'depth': better_hps[1],
'num_hidden': better_hps[2],
'layer_sum': better_hps[3]
}
# if basic_hypers['batch_size'] < 10:
# basic_hypers['batch_size'] = 10
if basic_hypers['patch_size'] > 28:
basic_hypers['patch_size'] = 28
print('=' * 80)
print(basic_hypers)
stride_params = [[1, 2, 2, 1]
for _ in range(basic_hypers['layer_sum'])]
ret, better_hps = conv_train(train_dataset,
train_labels,
valid_dataset,
valid_labels,
test_dataset,
test_labels,
image_size,
num_labels,
basic_hypers,
stride_params,
lrd=True)
else:
print('can not find better hypers')