Examples — Guide to Core ML Tools (original) (raw)

Contents

• For a Quick Start
• ML Program with Typed Execution
• TensorFlow 2
• TensorFlow 1
• PyTorch
• Model Intermediate Language (MIL)
• Conversion Options
  • Image Inputs
  • Classifiers
  • Flexible Input Shapes
  • Composite and Custom Operators
• Optimization
• Trees and Linear Models
• MLModel
  • MLModel Overview
  • Model Prediction
  • Xcode Model Preview Types
  • MLModel Utilities
• Updatable Models

Examples#

The following are code example snippets and full examples of using Core ML Tools to convert models.

For a Quick Start#

Full example:

• Getting Started: Demonstrates how to convert an image classifier model trained using the TensorFlow Keras API to the Core ML format.

ML Program with Typed Execution#

Full example:

• Typed Execution Workflow Example: Demonstrates a workflow for checking accuracy using ML Programs with Typed Execution.

TensorFlow 2#

• Load and Convert Model Workflow
• TensorFlow 2 Workflow
• Convert a Pre-trained Model
• Convert a User-defined Model

Full examples:

• Getting Started: Demonstrates how to convert an image classifier model trained using the TensorFlow Keras API to the Core ML format.
• Converting TensorFlow 2 BERT Transformer Models: Converts an object of the tf.keras.Model class and a SavedModel in the TensorFlow 2 format.

TensorFlow 1#

• Convert From TensorFlow 1.
• Export as Frozen Graph and Convert.
• Convert a Pre-trained Model.

Full examples:

• Converting a TensorFlow 1 Image Classifier: Demonstrates the importance of setting the image preprocessing parameters correctly during conversion to get the right results.
• Converting a TensorFlow 1 DeepSpeech Model: Demonstrates automatic handling of flexible shapes using automatic speech recognition.

PyTorch#

• Convert from PyTorch.
• PyTorch Conversion Workflow.
• Model Tracing.
• Model Exporting.

Full examples:

• Converting a torchvision Model from PyTorch: Traces / Exports a torchvision MobileNetV2 model, adds preprocessing for image input, and then converts it to Core ML.
• Converting a PyTorch Segmentation Model: Converts a PyTorch segmentation model that takes an image and outputs a class prediction for each pixel of the image.
• Converting an Open Efficient Language Model: Converts a PyTorch Open Efficient Language Model to Core ML

Model Intermediate Language (MIL)#

Full example:

• Model Intermediate Language: Construct a MIL program using the Python builder.”

Conversion Options#

Image Inputs#

• Use an MLMultiArray.
• Use an ImageType.
• Add Image Preprocessing Options.

Classifiers#

• Produce a Classifier Model.

Flexible Input Shapes#

• Select from Predetermined Shapes.
• Set the Range for Each Dimension.
• Update a Core ML Model to Use Flexible Input shapes

Composite and Custom Operators#

• Composite Operators: Defining a composite operation by decomposing it into MIL operations.

Full example:

• Custom Operators: Augment Core ML with your own operators and implement them in Swift.

Optimization#

Full examples:

• Training-Time Compression Examples: Use magnitude pruning, linear quantization, or palettization while training your model, or start from a pre-trained model and fine-tune it with training data.
• Compressing Neural Network Weights: Reduce the size of a neural network by reducing the number of bits that represent a number.

Trees and Linear Models#

• LibSVM
• Scikit-learn
• XGBoost

MLModel#

MLModel Overview#

• Load and save the MLModel.
• Use the MLModel for Prediction.
• Work with the Spec.
• Update the Metadata and Input/output Descriptions.

Model Prediction#

• Make Predictions
• Multi-array Prediction
• Image Prediction
• Image Prediction for a Multi-array Model
• Predict From the Compiled Model

Full example:

• Compiled Model Timing Example: Demonstrates timing differences with calling a large model.

Xcode Model Preview Types#

Full examples:

• Segmentation Example
• Body Pose Example

MLModel Utilities#

• Rename a Feature.
• Convert All Double Multi-array Feature Descriptions to Float.
• Evaluate Classifier, Regressor, and Transformer models.

Updatable Models#

Full examples:

• Nearest Neighbor Classifier: Create an updatable empty k-nearest neighbor.
• Neural Network Classifier: Create a simple convolutional model with Keras, convert the model to Core ML, and make the model updatable.
• Pipeline Classifier: Use a pipeline composed of a drawing-embedding model and a nearest neighbor classifier to create a model for training a sketch classifier. If you have a code example you’d like to submit, see Contributing.