Skip to content

GalenReich/Deploy-BERT-for-Sentiment-Analysis-with-FastAPI

 
 

Repository files navigation

Deploy BERT for Sentiment Analsysi with FastAPI

Deploy a pre-trained BERT model for Sentiment Analysis as a REST API using FastAPI

Demo

The model is trained to classify sentiment (negative, neutral, and positive) on a custom dataset from app reviews on Google Play. Here's a sample request to the API:

http POST http://127.0.0.1:8000/predict text="Good basic lists, i would like to create more lists, but the annual fee for unlimited lists is too out there"

The response you'll get looks something like this:

{
    "confidence": 0.9999083280563354,
    "probabilities": {
        "negative": 3.563107020454481e-05,
        "neutral": 0.9999083280563354,
        "positive": 5.596495248028077e-05
    },
    "sentiment": "neutral"
}

You can also read the complete tutorial here

Installation

Clone this repo:

git clone [email protected]:curiousily/Deploy-BERT-for-Sentiment-Analysis-with-FastAPI.git
cd Deploy-BERT-for-Sentiment-Analysis-with-FastAPI

Install the dependencies:

pipenv install --dev

Download the pre-trained model:

bin/download_model

Test the setup

Start the HTTP server:

bin/start_server

Send a test request:

bin/test_request

License

MIT

About

Deploy BERT for Sentiment Analysis as REST API using FastAPI, Transformers by Hugging Face and PyTorch

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 95.8%
  • Shell 4.2%