else:
argv = sys.argv[1:]
args = parser.parse_args(argv)
args = complete_default_train_parser(args)
logger.info('-' * 100)
logger.info('Input Argument Information')
logger.info('-' * 100)
args_dict = vars(args)
for a in args_dict:
logger.info('%-28s %s' % (a, args_dict[a]))
#########################################################################
# Read Data
##########################################################################
helper = DataHelper(gz=True, config=args)
# Set datasets
train_dataloader = helper.train_loader
dev_example_dict = helper.dev_example_dict
dev_feature_dict = helper.dev_feature_dict
dev_dataloader = helper.dev_loader
#########################################################################
# Initialize Model
##########################################################################
cached_config_file = join(args.exp_name, 'cached_config.bin')
if os.path.exists(cached_config_file):
cached_config = torch.load(cached_config_file)
encoder_path = join(args.exp_name, cached_config['encoder'])
model_path = join(args.exp_name, cached_config['model'])
def prepare_data(self):
helper = DataHelper(gz=True, config=self.args)
self.train_data = helper.train_loader
self.dev_example_dict = helper.dev_example_dict
self.dev_feature_dict = helper.dev_feature_dict
self.dev_data = helper.dev_loader