Analyser App

This app offers a fullt trained Pegasus Large and a Transformer to allow classification of a given text ans its reason. The detection can be made using an API endpoint.

Installation

Local PC:

• git clone https://github.com/augsaksham/AnalyserApp
• cd dev_app/project
• pip install -r requirements.txt
• python -m spacy download en_core_web_sm
• python -m spacy download en
• Goto "saved_mode" dir and donload the pretrained model from readme
• Open console and input : uvicorn main:app
• This will start a local server at localhost:8000

Goto localhost:8000/docs to see the API Documentation

Docker

• git clone https://github.com/augsaksham/AnalyserApp
• cd dev_app/project
• docker build . -t dev_nlp
• docker run -d --name dev_curr -p 80:80 dev_nlp This will start a local server at localhost:80

Goto localhost:80/docs to see the API Documentation

Features

• Robust classification
• Documented API for easier usage
• Mulitple Models

Traning and Evaluation :

To see the results of training and evalaution heat to comparison.ipynb file.