/ML-python-tensorflow-mnist-gpu-service

Apache License 2.0Apache-2.0

ML-python-tensorflow-mnist-gpu-service

Quickstart project for executing a MNIST classifier using TensorFlow on a GPU.

  • Running pip install requirements.txt and then python app.py will start the app on localhost where the user can send GET requests to perform inference.

  • NOTE: you will need to be running this service on a GPU for the model to load.

  • eg .../mnist/{index} where index is in the range [0:9999]

  • NOTE: this quickstart has no onnx model included. This means the tests will fail on the first build.

  • This quickstart has been written using the standard http.server python library as there are currently issues with using gunicorn with onnx models on a GPU.