thunlp / FewRel

A Large-Scale Few-Shot Relation Extraction Dataset
https://thunlp.github.io/fewrel.html
MIT License
731 stars 165 forks source link

关于评测时的'test_demo.py' #47

Closed Wangpeiyi9979 closed 4 years ago

Wangpeiyi9979 commented 4 years ago

您好,请问测评时提到的test_demo.py在哪里呢,请问需要自己实现吗

gaotianyu1350 commented 4 years ago

诶应该是文档忘记更新了,请问是哪里提到了test_demo.py?

Wangpeiyi9979 commented 4 years ago

![Uploading image.png…]()

Wangpeiyi9979 commented 4 years ago

https://thunlp.github.io/1/fewrel/codalab.html Step4

Wangpeiyi9979 commented 4 years ago

我把所有程序上传到codalab上,因为个人习惯问题,我一般先从data中预处理数据得到需要的vocab.txt,然后写入到data文件夹,最后模型构建embedder时再使用vocab.txt, 请问这个写入vocab.txt到data文件夹下的操作不被运行吗

Traceback (most recent call last): File "create_tool_data.py", line 37, in with open(vocab_out_path, 'w') as f: OSError: [Errno 30] Read-only file system: './data/vocab.txt'

这里的vocab_out_path = vocab.txt

Wangpeiyi9979 commented 4 years ago

已经可以在网站上对dev进行评测了