Covers: theory of Model Packaging
Estimated time needed to finish: 30 minutes
Questions this item addresses:
  • How can trained models be packaged?
How to use this item?

This folder contain files that will help you:

  • Common Serialization Methods using Iris model
  • MLflow hands on BERT model and Improvements
  • Convert Model in ONNX format for getting cross-platform ML interoperability
  • Inference Model serving overview
  • Flask based Model Serving
Author(s) / creator(s) / reference(s)
Brendan McGivern
Programming Languages: Python
0 comment

Packaging Trained Models for Production

Total time needed: ~2 hours
This recipe helps you understand the steps necessary to prepare a transportable version of your trained model to go from development to test and production environments
Potential Use Cases
Transporting trained models to different environments
Who is This For ?
INTERMEDIATEData Scientists, Machine Learning Engineers
Click on each of the following annotated items to see details.
Resource Asset6/6
VIDEO 1. Model Packaging Overview
  • How does model packaging work?
  • Why is model packaging important?
4 minutes
VIDEO 2. Common Serialization Approaches
  • How to serialize models in PyTorch?
9 minutes
VIDEO 3. MLflow Hands On
  • How can MLFlow be used for packaging model and environment?
20 minutes
VIDEO 4. Onnx Overview
  • How can ONNX be used for portable ML models?
5 minutes
VIDEO 5. Onnx Hands On
  • How can i serialize a PyTorch model in ONNX?
13 minutes
REPO 6. Hands-on Model Packaging
  • How can trained models be packaged?
30 minutes

Concepts Covered

0 comment