def build_model(self, height, width, num_channels):
self.config.height = height
self.config.width = width
self.config.num_channels = num_channels
self.graph = tf.Graph()
with self.graph.as_default():
self.model_graph = Factory(self.config)
print(self.model_graph)
self.trainable_count = np.sum([
np.prod(v.get_shape().as_list())
for v in tf.trainable_variables()
])
print('\nNumber of trainable paramters', self.trainable_count)
self.test_graph()
'''
-------------------------------------------------------------------------------
GOOGLE COLAB
-------------------------------------------------------------------------------------
'''
if self.config.colab:
self.push_colab()
self.config.push_colab = self.push_colab
self.config.isBuilt = True
file_utils.save_args(self.config.dict(), self.config.model_name,
self.config.config_dir,
['latent_mean', 'latent_std', 'push_colab'])
def save_model(self):
self.save(self.session, self.saver,
self.model_graph.global_step_tensor.eval(self.session))
self.compute_distribution(self.data_train.x)
file_utils.save_args(self.config.dict(), self.config.model_name,
self.config.config_dir,
['latent_mean', 'latent_std'])
gc.collect()
def __init__(self, **kwrds):
self.config = copy.deepcopy(config())
for key in kwrds.keys():
assert key in self.config.keys(), '{} is not a keyword, \n acceptable keywords: {}'. \
format(key, self.config.keys())
self.config[key] = kwrds[key]
self.latent_data = None
self.experiments_root_dir = 'experiments'
file_utils.create_dirs([self.experiments_root_dir])
self.config.model_name = get_model_name(self.config.graph_type, self.config)
if self.config.colab:
self.google2colab()
self.config.checkpoint_dir = os.path.join(self.experiments_root_dir + '/' + self.config.checkpoint_dir + '/',
self.config.model_name)
self.config.config_dir = os.path.join(self.experiments_root_dir + '/' + self.config.config_dir + '/',
self.config.model_name)
self.config.log_dir = os.path.join(self.experiments_root_dir + '/' + self.config.log_dir + '/',
self.config.model_name)
log_dir_subfolders = ['reconst', 'latent2d', 'latent3d', 'samples', 'total_random', 'pretoss_random', 'interpolate']
config_dir_subfolders = ['extra']
file_utils.create_dirs([self.config.checkpoint_dir, self.config.config_dir, self.config.log_dir])
file_utils.create_dirs([self.config.log_dir + '/' + dir_ + '/' for dir_ in log_dir_subfolders])
file_utils.create_dirs([self.config.config_dir + '/' + dir_ + '/' for dir_ in config_dir_subfolders])
load_config = {}
try:
load_config = file_utils.load_args(self.config.model_name, self.config.config_dir, ['latent_mean', 'latent_std', 'samples', 'y_uniqs'])
self.config.update(load_config)
print('Loading previous configuration ...')
except:
print('Unable to load previous configuration ...')
file_utils.save_args(self.config.dict(), self.config.model_name, self.config.config_dir,
['latent_mean', 'latent_std', 'samples', 'y_uniqs'])
if hasattr(self.config, 'height'):
try:
self.config.restore = True
self.build_model(self.config.height, self.config.width, self.config.num_channels)
except:
self.config.isBuilt = False
else:
self.config.isBuilt = False
