lightmlrestapi

It implements a light machine learning REST API based on falcon. If I were to start again this project, I would probably choose FastAPI. You can test a dummy wsgi server by running:

start_mlrestapi --name=dummy

And then query it with:

import requests
import ujson
features = ujson.dumps({'X': [0.1, 0.2]})
r = requests.post('http://127.0.0.1:8081', data=features)
print(r)
print(r.json())

It should return:

{'Y': [[0.4994216179, 0.4514893599, 0.0490890222]]}

The module was first tried with success in a hackathon in 2018. Participants could upload their model and retrieve their predictions through a REST API to check it was producing the same one as they had. A simple way to put a model into production.

• GitHub/lightmlrestapi

• documentation

• Blog

current - 2021-01-01 - 0.00Mb

• 23: Issue with ujson>=3.0, reject_bytes must be set to False (2020-06-06)

• 21: Fix issue with non-serializable numpy array (ujson does not do it anymore) (2020-03-29)

0.2.151 - 2019-05-17 - 0.12Mb

• 20: fix issue with falcon 2.0 (2019-05-08)

• 17: reload module mapped to subfolder (2019-01-11)

• 16: add full example to start a rest api (2018-12-01)

• 9: add version number to REST API (2018-11-13)

• 6: allow a zip file which contains data and code (2018-11-13)

• 4: add authentification to the rest api (2018-11-13)

• 8: allow clear logs (2018-11-02)

• 5: add a load function (2018-11-02)

• 3: add ip address in the logging (2018-04-15)

• 2: add encrypted logging (2018-04-15)

• 1: fix gallery of examples (style) (2018-01-05)