import bert_classifier
from utils import config
from utils import data_repairer
if config.need_repair:
data_repairer.repair_train_data()
cm = bert_classifier.ClassificationModel(gpu=True, seed=0)
if config.load_frompretrain != "None":
cm.load_model(config.model_state_path, config.model_config_path)
else:
cm.new_model()
# cm.save_model(config.save_path + '/model',config.save_path + '/config')
cm.train(config.epochs, config.batch_size, config.lr, config.plot_path,
config.save_path + '/model', config.save_path + '/config')
cm.create_test_predictions("./pred.csv")
import bert_classifier
from utils import config
config.USE_GPU = False
cm = bert_classifier.ClassificationModel(gpu=False, seed=0)
if config.load_frompretrain == True:
cm.load_model(config.model_state_path, config.model_config_path)
else:
cm.new_model()
# cm.save_model(config.save_path + '/model',config.save_path + '/config')
cm.train(config.epochs, config.batch_size, config.lr, config.plot_path,
config.save_path + '/model', config.save_path + 'config')
cm.create_test_predictions("./pred.csv")
if __name__ == '__main__':
print('running')
config.USE_GPU = False