Nyckel

Python package for www.nyckel.com

Docs

Visit our Docs pages

Quickstart

from nyckel import Credentials, TextClassificationFunction

# Get credentials from https://www.nyckel.com/console/keys
credentials = Credentials(client_id=..., client_secret=...)

# Create a new text classification function.
func = TextClassificationFunction.create(credentials=credentials, name="IsToxic")

# Provide a few examples.
func.create_samples([
    ("This is a nice comment", "not toxic"),
    ("Hello friend", "not toxic"),
    ("This is a bad comment", "toxic"),
    ("Who is this? Go away!", "toxic"),
])

# Nyckel trains and deploys your model in a few seconds...
# ...and you can start classifying text right away!

# Classify a new piece of text.
predictions = func.invoke(["New text"])

Contributors

Setup dev environment

Create a python3.11 venv and run:

pip install -r requirements.txt

Building docs locally

Install additional packages

pip install -r docs/requirements.txt

Run

mkdocs build

Testing

Testing locally requires a local server running the API. CI will run tests against a staging environment.

1. Ensure the local server is running
2. Ensure the nyckel package is installed in the venv using pip install -e .
3. Run pytest from the root directory

Deploying

One time

1. pip install hatch

Each time

hatch publish