def handle_argument(argument):
if argument == "-all":
test_all()
elif argument == "-example":
if len(argv) != 3:
print("-example takes as input the name of the example to run. Please run with 'python main.py -example NAME_OF_EXAMPLE.graph'")
return
run_test(argv[2])
elif argument == "-file":
if len(argv) != 3:
print("-file takes as input the name of the example to run. Please run with 'python main.py -file PATH/TO/FILE.graph'")
return
run_file(argv[2])
else:
print("Unrecognized command.")
def test_reformatting(out_test_dir):
out_test_dir = os.path.expanduser(out_test_dir)
test_dir = out_test_dir + '/tmp'
if (os.path.isdir(out_test_dir)
): # Warn if output directory already exists
log.disable(log.NOTSET) # flip logging back on
log.warning(
'Your output directory already exists. Your output directory may already contain output from Fasta-O-Matic in %(out_test_dir)s.'
% locals())
log.disable(log.ERROR) # disable most log output again
else:
assert general.mk_out_sub_directory(
out_test_dir
), 'Failed to create output directory. Check that your output directory path.'
assert general.mk_out_sub_directory(
test_dir
), 'Failed to create output sub-directory for Unit testing. Check that your output directory exists and can be written to.'
assert (test.test_all(test_dir)
), 'Failed to reformat when all three steps were called'
assert test.test_newline(
test_dir
), 'Failed to reformat when only newline and header reformatting was used'
assert test.test_wrapping(
test_dir
), 'Failed to reformat when only wrapping and header reformatting was used'
assert test.test_unique(
test_dir
), 'Failed to reformat or die when only testing uniqueness validation and/or reformatting for headers'
os.rmdir(test_dir)
def test_reformatting(out_test_dir):
out_test_dir = os.path.expanduser(out_test_dir)
test_dir = out_test_dir + '/tmp'
if (os.path.isdir(out_test_dir)): # Warn if output directory already exists
log.disable(log.NOTSET) # flip logging back on
log.warning('Your output directory already exists. Your output directory may already contain output from Fasta-O-Matic in %(out_test_dir)s.' % locals())
log.disable(log.ERROR) # disable most log output again
else:
assert general.mk_out_sub_directory(out_test_dir), 'Failed to create output directory. Check that your output directory path.'
assert general.mk_out_sub_directory(test_dir), 'Failed to create output sub-directory for Unit testing. Check that your output directory exists and can be written to.'
assert (test.test_all(test_dir)), 'Failed to reformat when all three steps were called'
assert test.test_newline(test_dir), 'Failed to reformat when only newline and header reformatting was used'
assert test.test_wrapping(test_dir), 'Failed to reformat when only wrapping and header reformatting was used'
assert test.test_unique(test_dir), 'Failed to reformat or die when only testing uniqueness validation and/or reformatting for headers'
os.rmdir(test_dir)
def quick_sort(input_list):
if input_list not in [None, []]:
lows = []
equal = []
highs = []
if len(input_list) > 1:
pivot = input_list[0]
for x in input_list:
if x < pivot:
lows.append(x)
elif x == pivot:
equal.append(x)
elif x > pivot:
highs.append(x)
return quick_sort(lows) + equal + quick_sort(highs)
return input_list
sort_methods = {
'bubble sort': bubble_sort,
'count sort': count_sort,
'radix sort (base64)': lambda l: radix_sort(l, base=64),
'radix sort (base256)': lambda l: radix_sort(l, base=256),
'merge sort': merge_sort,
'quick_sort': quick_sort
}
test_all(sort_methods)
from test import test_all, run_test, run_file
from sys import argv
def handle_argument(argument):
if argument == "-all":
test_all()
elif argument == "-example":
if len(argv) != 3:
print("-example takes as input the name of the example to run. Please run with 'python main.py -example NAME_OF_EXAMPLE.graph'")
return
run_test(argv[2])
elif argument == "-file":
if len(argv) != 3:
print("-file takes as input the name of the example to run. Please run with 'python main.py -file PATH/TO/FILE.graph'")
return
run_file(argv[2])
else:
print("Unrecognized command.")
if len(argv) == 1:
print("Did not receive a commandline argument. Running all tests by default.")
test_all()
handle_argument(argv[1])
x_train = x_train[:args.train_num]
y_train = y_train[:args.train_num]
# define model
if args.dataset != 4:
model, eval_model, manipulate_model = CapsNet(
input_shape=x_train.shape[1:],
n_class=len(np.unique(np.argmax(y_train, 1))),
routings=args.routings,
l1=args.l1)
else:
model, eval_model, manipulate_model = CapsNet_for_big(
input_shape=x_train.shape[1:],
n_class=len(np.unique(np.argmax(y_train, 1))),
routings=args.routings,
l1=args.l1)
model.summary()
# train or test
if args.weights is not None: # init the model weights with provided one
model.load_weights(args.weights)
if not args.testing:
train(model=model, data=(x_train, y_train), args=args)
else: # as long as weights are given, will run testing
if args.weights is None:
#print('No weights are provided. Will test using random initialized weights.')
test_all(eval_model, (x_test, y_test), args.dataset)
#manipulate_latent(manipulate_model, (x_test, y_test), args)
#save_for_gif(manipulate_model, (x_test, y_test), args)
#test(model=eval_model, data=(x_test, y_test), args=args)
def test():
test_all()
def train(self):
if self.config['is_train']:
data_dir_ir = 'data/ino_24_17/Train_ir_aug'
data_dir_vi = 'data/ino_24_17/Train_vi_aug'
else:
data_dir_ir = 'data/ino_24_17/Test_ir'
data_dir_vi = 'data/ino_24_17/Test_vi'
# image_size = 132, label_size = 120, stride = 14
# get all patches and labels (37710, 132, 132, 1) (37710, 120, 120, 1)
train_data_ir, train_label_ir = preprocessing.get_images2(
data_dir_ir, self.config['image_size'], self.config['label_size'],
self.config['stride'])
train_data_vi, train_label_vi = preprocessing.get_images2(
data_dir_vi, self.config['image_size'], self.config['label_size'],
self.config['stride'])
random_index = torch.randperm(
len(train_data_ir)) # a list of random indices
# shuffle randomly
train_data_vi = train_data_vi[random_index]
train_data_ir = train_data_ir[random_index]
train_label_vi = train_label_vi[random_index]
train_label_ir = train_label_ir[random_index]
batch_size = self.config['batch_size']
print('get patch done')
if self.config['is_train']:
with SummaryWriter(self.config['summary_dir']) as writer:
batch_steps = len(train_data_ir) // batch_size
epochs = self.config['epoch']
# for each epoch
for epoch in range(1, 1 + epochs):
d_loss_mean = 0
g_loss_mean = 0
content_loss_mean = 0
# for each step
for step in range(1, 1 + batch_steps):
print('Training. Epoch:%d/%d Step:%d/%d' %
(epoch, epochs, step, batch_steps))
start_idx = (step - 1) * batch_size
# (32, 132, 132, 1) -> (32, 1, 132, 132)
inf_x = train_data_ir[start_idx:start_idx +
batch_size].transpose(
[0, 3, 1, 2])
inf_y = train_label_ir[start_idx:start_idx +
batch_size].transpose(
[0, 3, 1, 2])
vis_x = train_data_vi[start_idx:start_idx +
batch_size].transpose(
[0, 3, 1, 2])
vis_y = train_label_vi[start_idx:start_idx +
batch_size].transpose(
[0, 3, 1, 2])
inf_x = torch.tensor(inf_x).float().to(device)
inf_y = torch.tensor(inf_y).float().to(device)
vis_x = torch.tensor(vis_x).float().to(device)
vis_y = torch.tensor(vis_y).float().to(device)
d_loss, g_loss, v_gan_loss, content_loss = self.train_step(
vis_x, inf_x, vis_y, inf_y, 2)
d_loss_mean += d_loss
g_loss_mean += g_loss
content_loss_mean += content_loss
print(
'Epoch {}/{}, Step {}/{}, gen loss = {:.4f}, v_gan_loss = {:.4f}, '
'content_loss {:.4f}, dis loss = {:.4f}'.format(
epoch, epochs, step, batch_steps, g_loss,
v_gan_loss, content_loss, d_loss))
test_all(
self.gen,
os.path.join(self.config['output'],
'test{}'.format(epoch)))
d_loss_mean /= batch_steps
g_loss_mean /= batch_steps
content_loss_mean /= batch_steps
writer.add_scalar('scalar/gen_loss', g_loss_mean, epoch)
writer.add_scalar('scalar/dis_loss', d_loss_mean, epoch)
writer.add_scalar('scalar/content_loss', content_loss_mean,
epoch)
# for name, param in self.gen.named_parameters():
# if 'bn' not in name:
# writer.add_histogram('gen/'+name, param, epoch)
#
# for name, param in self.dis.named_parameters():
# if 'bn' not in name:
# writer.add_histogram('dis/'+name, param, epoch)
print('Saving model......')
torch.save(self.gen.state_dict(),
'%s/final_generator.pth' % (self.config['output']))
print("Training Finished, Total EPOCH = %d" % self.config['epoch'])