API to manage deep-learning models and other artifacts. It is used primarily by qute.
$ conda create -n qute-api-env python=3.11 # or 3.10
$ conda activate qute-api-env
$ git clone https://github.com/aarpon/qute_api /path/to/qute_api
$ cd /path/to/qute_api
$ python -m pip install -e .
$ pip install -r dev-requirements.txt
$ cd qute_api/qute_api
- Copy
config.ini.template
toconfig.ini
and set theMODELS_DIR
variable to point to the models collection. The description of the models collection structure will follow soon.
$ export FLASK_APP=qute_api/app.py; flask run
To test, point your browser to http://127.0.0.1:5000.
To get a list of modes: /models
To get a list of versions of a specific model: /models/<model_name>
To download a specific model and version: /models/<model_name>/<model_version>
To download the hyper-parameters YAML file: /models/<model_name>/<model_version>/hparams
Alternatively, you can use the Gunicorn WSGI server:
$ gunicorn -w 4 -b 0.0.0.0:8000 'qute_api.app:app'
See qute_api/qute_client/example.py
.