[ ] ~(optional) is_idiom custom attribute~ better to assign it with an Entity value.
[x] refactoring loaders: loaders just get the path directly from config. 조삼모사.
[x] project root settings - setup.py should be in the project root. not the library. slide folder should go back to merge-idioms directory. merge_Idioms directory is what gets distributed. So...we need the vocabulary. learn from the project structure of spaCy.
[x] new script: save target idioms.
[x] save vocabulary separately -> something to think about? You shouldn't be building it every single time.. 아, 이 문제는, 그냥 하나의 모델을 shipping 하면 끝남. 아닌가? vocabulary가 달라질 수 있다는 것이 문제. 매번 동일한 것을.. 더하고 있으니. 아니 근데... 사실 큰 문제가 아닐수도? 큰 문제 맞음. 그냥 hash함수를 씀. 다른 환경으로 옮겨 졌을 때, 각 idiom이 어떤 값을 얻을지 모른다. 그게 문제.
To-do's
is_idiom
custom attribute~ better to assign it with an Entity value.setup.py
should be in the project root. not the library.slide
folder should go back tomerge-idioms
directory.merge_Idioms
directory is what gets distributed. So...we need the vocabulary. learn from the project structure ofspaCy
.