NLP-Tasks

Web Hosting

• Access the app deployed on DigitalOcean: http://178.128.39.191:8282

Brief Description

Various NLP tasks using Huggingface and Flask. Right now, the app supports only three tasks:

1. Sentiment Analysis: Identify if the sentence's sentiment is Positive or Negative.

2. Extractive Question Answering: For a given Context paragraph ask a Question. You should get an Answer from the paragraph.

3. Text Generation: Provide a Context (start of a sentence) and let the AI complete your story.

Plan is to include more tasks in future. Hugginface's transformers library makes these things very easy to do.

Contributions are most welcome.

Usage

0. Clone the repo:

   git clone https://github.com/ashishu007/NLP-Tasks.git

With Docker

1. Run with docker-composer

   docker-compose up

Without Docker

1. Navigate into the downloaded repo:

   cd NLP-Tasks/flask

2. Install the required dependencies:

   pip install -r requirements.txt

3. Run the flask-app:

   • Windows Powershell:

     $env:FLASK_APP="app.py"
     flask run

   • Linux:

     export FLASK_APP=app.py
     flask run

Main Screen

A screenshot of main screen

Contributions

1. Add more tasks
2. Improve the user-interface

Acknowledgement

Huggingface's transformers library has revolutionised the way state-of-the-art NLP models are being used in real-world. Without, this library, the app made here would have taken tremendous amount of work (compared to what it took rn). :)