Skip to content

Latest commit

 

History

History
69 lines (47 loc) · 2.25 KB

README.md

File metadata and controls

69 lines (47 loc) · 2.25 KB
Raw

AI Platform Prediction

You can host your trained machine learning models on AI Platform for predictions (inference) from new examples.

Currently, AI Platform can serve models generated by scikit-learn, xgboost, Pytorch and TensorFlow.

Steps to Deploy

To deploy a model on AI Platform, there are two general steps that need be taken:

  1. Create a model resource: A model resource is like a folder for all different versions of a model.
gcloud ai-platform models create MODEL_NAME \
  --regions REGION
  1. Create a model version: Once a model resource is created, the model object is uploaded to AI Platform, along with a version.
gcloud ai-platform versions create $VERSION_NAME \
  --model $MODEL_NAME \
  --origin $MODEL_DIR \
  --runtime-version=1.15 \
  --framework $FRAMEWORK \
  --python-version=3.7
  1. Custom Prediction Routine
gcloud components install beta

gcloud beta ai-platform versions create $VERSION_NAME \
  --model $MODEL_NAME \
  --origin $MODEL_DIR \
  --runtime-version=1.15 \
  --python-version=3.7
  --package-uris=$CUSTOM_CODE_PATH \
  --prediction-class=$PREDICTOR_CLASS

More information here

Make Prediction

After a model version is created, it can be used to make predictions through a simple Python API or using gcloud command. We will show the details of all the required steps in the samples in this directory.

gcloud ai-platform predict --model $MODEL_NAME  \
                   --version $VERSION_NAME \
                   --json-instances $INPUT_DATA_FILE

More information here

Tools

  • Locust: Easily run a load test on models deployed on Cloud AI Platform from GCE, GKE, or local.

  • Model warmup: Learn how to do Model warmup in AI Platform Prediction

Further Information

For further information on how the prediction on AI Platform works, please click here.