highwind 0.4.0

Python SDK for Highwind (https://highwind.ai/)

Highwind

A Python SDK for Highwind.

Installation

The package is made available on the Python Package Index (PyPI).

pip install highwind

Usage and Current Features

1. Run Use Case inference

from typing import Dict

from highwind import UseCase

use_case: UseCase = UseCase(id="a1...88") # You'll need to get this from the Highwind Web UI

use_case.name # 'IRIS Classifier'

inference_payload: Dict = {
    "inputs": [
        {
            "name": "input-0",
            "shape": [1],
            "datatype": "BYTES",
            "parameters": None,
            "data": [[6.1, 2.8, 4.7, 1.2]],
        }
    ]
}

result: Dict = use_case.run_inference(inference_payload)

print(result)
# {
#     "id": "c1958d66-27d0-403d-9ae4-45fb9f7dac8b",
#     "model_name": "c8uj6o4d",
#     "model_version": None,
#     "outputs": [
#         {
#             "data": [1],
#             "datatype": "INT64",
#             "name": "output-0",
#             "parameters": None,
#             "shape": [1],
#         }
#     ],
#     "parameters": None,
# }

Full feature set

• Create an Asset and upload its files from your local machine
• Create a UseCase
• Search for all your Assets and get their UUIDs
• Link Assets to a UseCase
• Search for all your UseCases and get their UUIDs
• Deploy a UseCase
• Run inference on a deployed UseCase
• Many more future features as Highwind develops

Contributing

1. Fork the repository
2. Make your desired changes
3. Ensure adequate test coverage
4. Open a Pull Request

Installing dependencies

We use Poetry, so install that first by following the official docs.

Then run:

poetry install

We also use pre-commit, so install that too by following the official docs.

Then run:

pre-commit install

Running tests

Simply run:

pytest

Building and Publishing the SDK

To build the package, simply bump the version number in pyproject.toml. Then run:

poetry build

This creates a new (gitignored) folder called dist/ in the root of the project.

To publish the package to PyPI, simply run:

poetry publish