StackOverflow allows users to tag their queries with appropriate tag names. This project is about predicting tags automatically for given questio and description.
Embedding for the text are generated using BERT tokenizer using transformers library.
bert-base-uncased model along with fully connected layer is used as predicting model.
f1 score: 70%
precision: 70%
recall: 71%
f1 score: 55%
precision: 55%
recall: 54%
API has been built using FastAPI
To start API: run command uvicorn api:app
url: http://localhost:8000/
check documentation for FastAPI
UI has been designed using streamlit
To run UI: run command streamlit run src/app.py
url: http://localhost:8501/
