zhw12 / bertrl Goto Github PK

Hi Hanwen,

I'm very interested in your work and want to know more details using your code implementation.

However, when I run the code by myself, one problem occurs:

Traceback (most recent call last):
File "run_bertrl.py", line 320, in
main()
File "run_bertrl.py", line 135, in main
GlueDataset(data_args, tokenizer=tokenizer, cache_dir=model_args.cache_dir) if training_args.do_train else None
File "/home/zhixiang/miniconda3/envs/py36/lib/python3.6/site-packages/transformers/data/datasets/glue.py", line 121, in init
with FileLock(lock_path):
File "/home/zhixiang/miniconda3/envs/py36/lib/python3.6/site-packages/filelock.py", line 323, in enter
self.acquire()
File "/home/zhixiang/miniconda3/envs/py36/lib/python3.6/site-packages/filelock.py", line 271, in acquire
self._acquire()
File "/home/zhixiang/miniconda3/envs/py36/lib/python3.6/site-packages/filelock.py", line 384, in _acquire
fd = os.open(self._lock_file, open_mode)
FileNotFoundError: [Errno 2] No such file or directory: 'bertrl_data/fb237_hop3_full_neg10_max_inductive/cached_train_BertTokenizerFast_128_mrpc.lock'

This may due to the transformers version. Since version 3.3.1 is not available, I tried the latest version 4.18.0.
Could you please give me some suggestions to fix it?

你好，

1. 请问这篇文章被哪个期刊或会议录用了？

2. 读了这篇论文，不是很理解BERT的输入是什么？

3.你的论文中 unseen relation 指的是什么？不是很理解。

希望您有空的时候，帮我解决上述疑问，谢谢。

Miao

why the code doesn't use the train.tsv in bertrl_data as train dataset but use Glue dataset? I not see any code use train.tsv.