Skip to content

A Quora Insincere Question Classification project that uses natural language processing (NLP) to detect and classify toxic, misleading, or insincere questions, helping to maintain a healthy and respectful online discussion environment.

License

Notifications You must be signed in to change notification settings

menacingly-coded/Quora-Insincere-Question-Classification

Repository files navigation

Quora Insincere Questions Classification

CircleCI

4th Place Solution for Quora Insincere Questions Classification

Solution overview is as below:

overview

Requirement

  • docker >= 17.12.0
  • docker-compose >= 1.19.0
  • nvidia-docker2 >= 2.0.2

Getting started

🔰 Setup

Setup kaggle API credentials

Download kaggle.json and place in the location: ~/.kaggle/kaggle.json.
See details: https://github.com/Kaggle/kaggle-api

Build Docker image

docker-compose build

Download and unzip competition datasets

docker-compose run cpu kaggle competitions download quora-insincere-questions-classification -p input
unzip "input/*.zip" -d input

🚀 Train model

Train with GPU

docker-compose run gpu python exec/train.py -m python exec/train.py -m models/submit/submit1_embed_smpl_400.py -g <GPU_ID>

Train with CPU

docker-compose run cpu python exec/train.py -m python exec/train.py -m models/submit/submit1_embed_smpl_400.py

Contribution

Below command will run both flake8 and pytest:

$ docker-compose run test

Coding Guidelines

We use PEP8 syntax conventions, so please check your python changes:

$ docker-compose run cpu flake8

Testing

Before sending your PR, please make sure all tests are passing:

$ docker-compose run cpu nosetests

About

A Quora Insincere Question Classification project that uses natural language processing (NLP) to detect and classify toxic, misleading, or insincere questions, helping to maintain a healthy and respectful online discussion environment.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published