This repository shows how to get a baseline model for test classification task. We experimented with following models:
- Multinomial Naive Bayes
- Bernoulli Naive Bayes
- Logistic Regression
- SVM
Here you can find a notebook with code to seek for a baseline model.
This repository also provides an example how to wrap your model in simple web API interface. For this purpose we use Flask and docker container.
You can either use public image acheshkov/text-clf
from docker hub or build your own. To build own put model file model.joblib
to model
folder and execute:
$ docker build -t IMAGE_TAG_NAME .
To run container specify port you need:
$ docker run -d -p PORT:80 IMAGE_TAG_NAME
or use existing image:
$ docker run -d -p PORT:80 acheshkov/text-clf
Since Web API deployed you can call it
$ curl -d '{"text": "any text to classify"}' https://YOUR_BACKEND:PORT/classify