aaroosh-07 / bons.ai

0 stars 0 forks source link

bons.ai

bons.ai is web application, which allows you to perform Question Answering task on pdf files uploaded or text provided. Currently we have developed a simple python script to use hugging face models to perform question answering task.

Getting Started

Setting up Environment

  1. Create python virtual environment using this command
python -m venv virtualenv
  1. Activate the virtual environment
//for windows
virtualenv/Scripts/Activate.ps1 //In Powershell

//for linux/ Mac
source virtualenv/bin/activate
  1. Install all the required modules.
pip install -r requirements.txt

Adding API_TOKEN from hugging face

  1. generate your API_TOKEN from hugging face.
  2. create a .env file in root directory of your project.
  3. add the following content in your .env file
API_TOKEN = "<your api token>"

Running the script

Run the following commad after activating the virtual environment.

cd experiments
python main.py

Running FastAPI server

This command will run fastapi development server on localhost:8000

fastapi dev server.py

API Endpoints

The following API endpoints have been implmented.

Request

GET /

curl -X 'GET' \
  'http://127.0.0.1:8000/' \
  -H 'accept: application/json'

Response body

{
  "Message": "welcome to bons.ai"
}

Request

POST /prompt

curl -X 'POST' \
  'http://127.0.0.1:8000/prompt' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "question": "what is my name",
  "context": "my name is aaroosh",
  "model": "distilbert-base-cased"
}'

In this POST request we need to send request body having the following structure

{
  "question": string,
  "context": string,
  "model": "distilbert-base-cased" or "roberta-base-squad2" or "google-bert" or "electra_large_discriminator"
}

Response body

{
  "score": 0.9413116574287415,
  "start": 11,
  "end": 18,
  "answer": "aaroosh"
}