train['labels'] = encode(train["labels"])
dev['labels'] = encode(dev["labels"])
dev_sentences = dev['text'].tolist()
dev_preds = np.zeros((len(dev), args["n_fold"]))
if args["evaluate_during_training"]:
for i in range(args["n_fold"]):
if os.path.exists(args['output_dir']) and os.path.isdir(
args['output_dir']):
shutil.rmtree(args['output_dir'])
print("Started Fold {}".format(i))
model = ClassificationModel(
MODEL_TYPE,
MODEL_NAME,
args=args,
use_cuda=torch.cuda.is_available()
) # You can set class weights by using the optional weight argument
train_df, eval_df = train_test_split(train,
test_size=0.1,
random_state=SEED * i)
model.train_model(train_df,
eval_df=eval_df,
macro_f1=macro_f1,
weighted_f1=weighted_f1,
accuracy=sklearn.metrics.accuracy_score)
model = ClassificationModel(MODEL_TYPE,
args["best_model_dir"],
args=args,
use_cuda=torch.cuda.is_available())
train['labels'] = encode(train["labels"])
test['labels'] = encode(test["labels"])
test_sentences = test['text'].tolist()
test_preds = np.zeros((len(test), args["n_fold"]))
if args["evaluate_during_training"]:
for i in range(args["n_fold"]):
if os.path.exists(args['output_dir']) and os.path.isdir(
args['output_dir']):
shutil.rmtree(args['output_dir'])
print("Started Fold {}".format(i))
model = ClassificationModel(
MODEL_TYPE,
MODEL_NAME,
args=args,
use_cuda=torch.cuda.is_available()
) # You can set class weights by using the optional weight argument
train_df, eval_df = train_test_split(train,
test_size=0.1,
random_state=SEED * i)
model.train_model(train_df,
eval_df=eval_df,
macro_f1=macro_f1,
weighted_f1=weighted_f1,
accuracy=sklearn.metrics.accuracy_score)
model = ClassificationModel(MODEL_TYPE,
args["best_model_dir"],
args=args,
use_cuda=torch.cuda.is_available())