def main(args):
results_type = args.results_type
saving_dir = os.path.join(".", "models_evaluation", "results",
results_type)
os.makedirs(saving_dir, exist_ok=True)
address_parser = AddressParser(model_type=args.model_type, device=0)
directory_path = args.test_directory
test_files = os.listdir(directory_path)
training_test_results = {}
zero_shot_test_results = {}
for idx, test_file in enumerate(test_files):
results, country = test_on_country_data(address_parser, test_file,
directory_path, args)
print(f"{idx} file done of {len(test_files)}.")
if train_country_file(test_file):
training_test_results.update({country: results["test_accuracy"]})
elif zero_shot_eval_country_file(test_file):
zero_shot_test_results.update({country: results["test_accuracy"]})
else:
print(
f"Error with the identification of test file type {test_file}."
)
training_base_string = "training_test_results"
training_incomplete_base_string = "training_incomplete_test_results"
zero_shot_base_string = "zero_shot_test_results"
with open(
os.path.join(saving_dir,
f"{training_base_string}_{args.model_type}.json"),
"w",
encoding="utf-8",
) as file:
json.dump(training_test_results, file, ensure_ascii=False)
with open(
os.path.join(saving_dir,
f"{zero_shot_base_string}_{args.model_type}.json"),
"w",
encoding="utf-8",
) as file:
json.dump(zero_shot_test_results, file, ensure_ascii=False)
incomplete_test_directory = args.incomplete_test_directory
incomplete_test_files = os.listdir(incomplete_test_directory)
incomplete_training_test_results = {}
for idx, incomplete_test_file in enumerate(incomplete_test_files):
results, country = test_on_country_data(address_parser,
incomplete_test_file,
incomplete_test_directory,
args)
print(f"{idx} file done of {len(incomplete_test_files)}.")
if train_country_file(incomplete_test_file):
incomplete_training_test_results.update(
{country: results["test_accuracy"]})
else:
print(
f"Error with the identification of test file type {incomplete_test_file}."
)
with open(
os.path.join(
saving_dir,
f"{training_incomplete_base_string}_{args.model_type}.json"),
"w",
encoding="utf-8",
) as file:
json.dump(incomplete_training_test_results, file, ensure_ascii=False)
def test_givenATrainCountryFile_whenIsTrainCountryFile_thenReturnTrue(self):
for train_test_file in self.some_train_test_files:
self.assertTrue(train_country_file(train_test_file))
def test_givenANonZeroShotCountryFile_whenIsZeroShotCountryFile_thenReturnFalse(
self,
):
for zero_shot_test_file in self.some_zero_shot_test_files:
self.assertFalse(train_country_file(zero_shot_test_file))
def main(args):
saving_dir = os.path.join(".", "models_evaluation", "results")
os.makedirs(saving_dir, exist_ok=True)
address_parser = AddressParser(model_type=args.model_type, device=0)
directory_path = args.test_directory
test_files = os.listdir(directory_path)
training_test_results = {}
zero_shot_test_results = {}
for idx, test_file in enumerate(test_files):
results, country = test_on_country_data(address_parser, test_file,
directory_path, args)
print(f"{idx} file done of {len(test_files)}.")
if train_country_file(test_file):
training_test_results.update({country: results['test_accuracy']})
elif zero_shot_eval_country_file(test_file):
zero_shot_test_results.update({country: results['test_accuracy']})
else:
print(
f"Error with the identification of test file type {test_file}."
)
if args.fine_tuning:
training_base_string = "training_fine_tuned_test_results"
training_noisy_base_string = "training_noisy_fine_tuned_test_results"
zero_shot_base_string = "zero_shot_fine_tuned_test_results"
else:
training_base_string = "training_test_results"
training_noisy_base_string = "training_noisy_test_results"
zero_shot_base_string = "zero_shot_test_results"
json.dump(training_test_results,
open(os.path.join(
saving_dir,
f"{training_base_string}_{args.model_type}.json"),
"w",
encoding="utf-8"),
ensure_ascii=False)
json.dump(zero_shot_test_results,
open(os.path.join(
saving_dir,
f"{zero_shot_base_string}_{args.model_type}.json"),
"w",
encoding="utf-8"),
ensure_ascii=False)
noisy_test_directory = args.noisy_test_directory
noisy_test_files = os.listdir(noisy_test_directory)
noisy_training_test_results = {}
for idx, noisy_test_file in enumerate(noisy_test_files):
results, country = test_on_country_data(address_parser,
noisy_test_file,
noisy_test_directory, args)
print(f"{idx} file done of {len(noisy_test_files)}.")
if train_country_file(noisy_test_file):
noisy_training_test_results.update(
{country: results['test_accuracy']})
else:
print(
f"Error with the identification of test file type {noisy_test_file}."
)
json.dump(noisy_training_test_results,
open(os.path.join(
saving_dir,
f"{training_noisy_base_string}_{args.model_type}.json"),
"w",
encoding="utf-8"),
ensure_ascii=False)
