jiesutd / yato Goto Github PK

I got the training type error as follow while training my Sequence Tagging model.

Training model...
Training model...
Training model...
Training model...
Training model...

TypeError Traceback (most recent call last)
Cell In[23], line 3
1 from yato import YATO
2 model = YATO(configuration_file_myPOS_MyanBerta)
----> 3 model.train()

File /opt/conda/lib/python3.10/site-packages/yato/yato.py:38, in YATO.train(self, log, metric)
36 self.data.build_pretrain_emb()
37 self.data.summary()
---> 38 train(self.data, log, metric)

File /opt/conda/lib/python3.10/site-packages/yato/main.py:441, in train(data, log, metric)
439 logger.info("Training model...")
440 save_data_name = data.dset_dir
--> 441 data.save(save_data_name)
442 best_test = [{"acc": {"best test": 0, "best dev": 0, "epoch num": 0}},
443 {"f": {"best test": 0, "best dev": 0, "epoch num": 0}}]
444 metric_seq = ["acc", 'f']

File /opt/conda/lib/python3.10/site-packages/yato/utils/data.py:596, in Data.save(self, save_file)
595 def save(self, save_file):
--> 596 f = open(save_file, 'wb')
597 pickle.dump(self.dict, f, 2)
598 f.close()

TypeError: expected str, bytes or os.PathLike object, not NoneType

The following is my Config file

use # to comment out the configure item

I/O

train_dir=/kaggle/input/myposv3-ncrfpp/train.bmes
dev_dir=/kaggle/input/myposv3-ncrfpp/train.bmes
test_dir=/kaggle/input/myposv3-ncrfpp/train.bmes
model_dir=/myPOS_MyanBerta/YATO_MODEL/myPOS_MyanBerta_model

#dset_dir=/myPOS_MyanBerta/YATO_MODEL/myPOS_MyanBerta_c03.dset

norm_word_emb=False
norm_char_emb=False
number_normalized=False
seg=True
word_emb_dim=50
char_emb_dim=30

###NetworkConfiguration###
use_crf=False
use_char=False
char_seq_feature=None
use_word_seq=False
use_word_emb=False
word_seq_feature=None
low_level_transformer=None
low_level_transformer_finetune=False
high_level_transformer=UCSYNLP/MyanBERTa
high_level_transformer_finetune=True

###TrainingSetting###
status=train
optimizer=AdamW
iteration=10
batch_size=16
ave_batch_loss=False

###Hyperparameters###
cnn_layer=4
char_hidden_dim=50
hidden_dim=768
dropout=0.3
lstm_layer=2
bilstm=True
learning_rate=3e-5
gpu=True
device=cuda:0
scheduler=get_cosine_schedule_with_warmup
warmup_step_rate=0.05