def get_training_data_and_feature(opt, data_loader, preprocessor):
""" prepare feature and data """
if opt.load_feature:
try:
train_features, train_label2id, train_id2label, train_trans_mat = load_feature(opt.train_path.replace('.json', '.saved.pk'))
dev_features, dev_label2id, dev_id2label, dev_trans_mat = load_feature(opt.dev_path.replace('.json', '.saved.pk'))
except FileNotFoundError:
# Not a saved feature file yet, make it
opt.load_feature = False
opt.save_feature = True
train_features, train_label2id, train_id2label, train_trans_mat, \
dev_features, dev_label2id, dev_id2label, dev_trans_mat =\
get_training_data_and_feature(opt, data_loader, preprocessor)
# restore option
opt.load_feature = True
opt.save_feature = False
else:
train_examples, train_max_len, train_max_support_size, train_trans_mat = data_loader.load_data(path=opt.train_path)
dev_examples, dev_max_len, dev_max_support_size, dev_trans_mat = data_loader.load_data(path=opt.dev_path)
train_label2id, train_id2label = make_dict(train_examples)
dev_label2id, dev_id2label = make_dict(dev_examples)
logger.info(' Finish train dev prepare dict ')
train_features = preprocessor.construct_feature(
train_examples, train_max_support_size, train_label2id, train_id2label)
dev_features = preprocessor.construct_feature(
dev_examples, dev_max_support_size, dev_label2id, dev_id2label)
logger.info(' Finish prepare train dev features ')
if opt.save_feature:
save_feature(opt.train_path.replace('.json', '.saved.pk'),
train_features, train_label2id, train_id2label, train_trans_mat)
save_feature(opt.dev_path.replace('.json', '.saved.pk'), dev_features, dev_label2id, dev_id2label, dev_trans_mat)
return train_features, train_label2id, train_id2label, train_trans_mat, \
dev_features, dev_label2id, dev_id2label, dev_trans_mat
def get_training_data_and_feature(opt, data_loader, preprocessor):
""" prepare feature and data """
if opt.load_feature:
try:
train_features, (train_slot_label2id, train_slot_id2label), \
(train_intent_label2id, train_intent_id2label) = \
load_feature(opt.train_path.replace('.json', '.saved.pk'))
dev_features, (dev_slot_label2id, dev_slot_id2label), \
(dev_intent_label2id, dev_intent_id2label) = load_feature(opt.dev_path.replace('.json', '.saved.pk'))
except FileNotFoundError:
opt.load_feature, opt.save_feature = False, True # Not a saved feature file yet, make it
train_features, (train_slot_label2id, train_slot_id2label), (train_intent_label2id, train_intent_id2label),\
dev_features, (dev_slot_label2id, dev_slot_id2label), (dev_intent_label2id, dev_intent_id2label) = \
get_training_data_and_feature(opt, data_loader, preprocessor)
opt.load_feature, opt.save_feature = True, False # restore option
else:
train_examples, _, train_max_support_size = data_loader.load_data(
path=opt.train_path)
dev_examples, _, dev_max_support_size = data_loader.load_data(
path=opt.dev_path)
(train_slot_label2id, train_slot_id2label), (train_intent_label2id, train_intent_id2label) = \
make_dict(opt, train_examples)
(dev_slot_label2id, dev_slot_id2label), (dev_intent_label2id, dev_intent_id2label) = \
make_dict(opt, dev_examples)
logger.info(' Finish train dev prepare dict ')
train_features = preprocessor.construct_feature(
train_examples, train_max_support_size, train_slot_label2id,
train_slot_id2label, train_intent_label2id, train_intent_id2label)
dev_features = preprocessor.construct_feature(
dev_examples, dev_max_support_size, dev_slot_label2id,
dev_slot_id2label, dev_intent_label2id, dev_intent_id2label)
logger.info(' Finish prepare train dev features ')
if opt.do_debug:
print('train_slot_label2id: {}'.format(train_slot_label2id))
print('train_intent_label2id: {}'.format(train_intent_label2id))
print('dev_slot_label2id: {}'.format(dev_slot_label2id))
print('dev_intent_label2id: {}'.format(dev_intent_label2id))
print('train_examples: {}'.format(len(train_examples),
train_examples))
print('train_features: {}'.format(len(train_features),
train_features))
if opt.save_feature:
save_feature(opt.train_path.replace('.json',
'.saved.pk'), train_features,
train_slot_label2id, train_slot_id2label,
train_intent_label2id, train_intent_id2label)
save_feature(opt.dev_path.replace('.json', '.saved.pk'),
dev_features, dev_slot_label2id, dev_slot_id2label,
dev_intent_label2id, dev_intent_id2label)
return train_features, (train_slot_label2id, train_slot_id2label), (train_intent_label2id, train_intent_id2label), \
dev_features, (dev_slot_label2id, dev_slot_id2label), (dev_intent_label2id, dev_intent_id2label)
def get_testing_data_feature(opt, data_loader, preprocessor):
""" prepare feature and data """
if opt.load_feature:
try:
test_features, test_label2id, test_id2label, test_trans_mat = \
load_feature(opt.test_path.replace('.json', '.saved.pk'))
except FileNotFoundError:
# Not a saved feature file yet, make it
opt.load_feature = False
opt.save_feature = True
test_features, test_label2id, test_id2label, test_trans_mat = \
get_testing_data_feature(opt, data_loader, preprocessor)
# restore option
opt.load_feature = True
opt.save_feature = False
else:
test_examples, test_max_len, test_max_support_size, test_trans_mat = data_loader.load_data(path=opt.test_path)
test_label2id, test_id2label = make_dict(test_examples)
logger.info(' Finish prepare test dict')
test_features = preprocessor.construct_feature(
test_examples, test_max_support_size, test_label2id, test_id2label)
logger.info(' Finish prepare test feature')
if opt.save_feature:
save_feature(opt.test_path.replace('.json', '.saved.pk'),
test_features, test_label2id, test_id2label, test_trans_mat)
return test_features, test_label2id, test_id2label, test_trans_mat
def get_testing_data_feature(opt, data_loader, preprocessor):
""" prepare feature and data """
if opt.load_feature:
try:
test_features, (test_slot_label2id, test_slot_id2label), (test_intent_label2id, test_intent_id2label) = \
load_feature(opt.test_path.replace('.json', '.saved.pk'))
except FileNotFoundError:
opt.load_feature, opt.save_feature = False, True # Not a saved feature file yet, make it
test_features, (test_slot_label2id, test_slot_id2label), (test_intent_label2id, test_intent_id2label) = \
get_testing_data_feature(opt, data_loader, preprocessor)
opt.load_feature, opt.save_feature = True, False # restore option
else:
test_examples, test_max_len, test_max_support_size = data_loader.load_data(
path=opt.test_path)
(test_slot_label2id, test_slot_id2label), (test_intent_label2id, test_intent_id2label) = \
make_dict(opt, test_examples)
logger.info(' Finish prepare test dict')
test_features = preprocessor.construct_feature(
test_examples, test_max_support_size, test_slot_label2id,
test_slot_id2label, test_intent_label2id, test_intent_id2label)
logger.info(' Finish prepare test feature')
if opt.do_debug:
print('test_slot_label2id: {}'.format(test_slot_label2id))
print('test_slot_id2label: {}'.format(test_slot_id2label))
if opt.save_feature:
save_feature(opt.test_path.replace('.json', '.saved.pk'),
test_features, test_slot_label2id, test_slot_id2label,
test_intent_label2id, test_intent_id2label)
return test_features, (test_slot_label2id,
test_slot_id2label), (test_intent_label2id,
test_intent_id2label)
