Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[BUG] Grammar 모델 negative score #67

Closed
Tracked by #12
9ooDa opened this issue Mar 31, 2024 · 3 comments
Closed
Tracked by #12

[BUG] Grammar 모델 negative score #67

9ooDa opened this issue Mar 31, 2024 · 3 comments
Assignees
Labels
🐞 Bug 버그 발생 💻 Model 모델 관련

Comments

@9ooDa
Copy link
Contributor

9ooDa commented Mar 31, 2024

Describe the bug

image
Grammar 모델단에서 Error Tag가 중복으로 나오는 오류 때문에 최종 Grammar score가 negative로 나오는 것을 확인했습니다.
현재는 AI hub 데이터로 돌린 1162 개의 데이터 중 12 개의 데이터에 negative score가 있는 것을 보아 나오는 확률이 높지는 않지만, 서비스를 위해 버그를 잡아야 할 거 같습니다.
메트릭을 변경하거나 Error Tag 후처리로 진행할 예정입니다.
추후 업데이트 사항은 코멘트로 남겨놓겠습니다.

Additional context

related issue: #12

@9ooDa 9ooDa self-assigned this Mar 31, 2024
@9ooDa 9ooDa added 🐞 Bug 버그 발생 💻 Model 모델 관련 labels Mar 31, 2024
@9ooDa
Copy link
Contributor Author

9ooDa commented Apr 1, 2024

확인해보니 모델이 맞는 문장도 틀리게 잡는 거 같네요. 조금 더 뜯어본 다음에 업데이트 남기겠습니다..

{
   "phase_2":{
      "score":-18.18,
      "original_passage":" I like to watch movie like in a week sometimes.",
      "corrected_passage":" do a a sometimes.",
      "tag_grammar_info":[
         {
            "sid":0,
            "sentence":" I like to watch movie like in a week sometimes.",
            "corrected_sentence":" do a a sometimes.",
            "edited":true,
            "ref_word":[
               "to",
               "movie",
               "like",
               "to",
               "a",
               "once",
               "week",
               "like",
               "it",
               "watch",
               "in",
               "I",
               "it"
            ],
            "category":[
               "어휘",
               "어휘",
               "어휘",
               "어휘",
               "어휘",
               "어휘",
               "어휘",
               "어휘",
               "동사 형태",
               "어휘",
               "어휘",
               "어휘",
               "어휘"
            ],
            "tag":[
               "$APPEND_a",
               "$DELETE",
               "$REPLACE_once",
               "$DELETE",
               "$APPEND_it",
               "$DELETE",
               "$DELETE",
               "$DELETE",
               "$TRANSFORM_VERB_VB_VBG",
               "$DELETE",
               "$DELETE",
               "$REPLACE_do",
               "$DELETE"
            ],
            "grammar_description":[
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "PRESENT TO PRESENT PARTICIPLE",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY",
               "WRONG USE OF VOCABULARY"
            ]
         }
      ]
   }
}

@9ooDa
Copy link
Contributor Author

9ooDa commented Apr 3, 2024

grammar_api.py에 nltk로 sentence tokenization한 부분 주석처리하고 따로 처리했더니 해결됐습니다!

9ooDa added a commit that referenced this issue Apr 3, 2024
@9ooDa 9ooDa closed this as completed Apr 3, 2024
@C7C4FF
Copy link
Contributor

C7C4FF commented Apr 3, 2024

확인했습니다 ^-^)9 서버도 껐다 켰어요

@9ooDa 9ooDa mentioned this issue Jul 16, 2024
17 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
🐞 Bug 버그 발생 💻 Model 모델 관련
Projects
None yet
Development

No branches or pull requests

2 participants