Understand models in Sentis

Sentis can import and run trained machine learning model files in Open Neural Network Exchange (ONNX) format.

To get a model that's compatible with Sentis, you can do one of the following:

  • Train a model in a framework like TensorFlow, PyTorch or Keras, then export it in ONNX format.
  • Download a trained model file and convert it to ONNX format. Refer to the ONNXXMLTools Python package for more information.
  • Download a trained model that's already in ONNX format, for example from the ONNX Model Zoo.

How Sentis optimizes a model

When you import a model, each ONNX operator in the model graph becomes a Sentis layer.

Open the Model Asset Import Settings to check the list of layers in the imported model, in the order Sentis runs them. Refer to Supported ONNX operators for more information.

Sentis optimizes models to make them smaller and more efficient. For example, Sentis might do the following to an imported model:

  • Remove a layer or subgraph and turn it into a constant.
  • Replace a layer or subgraph with a simpler layer or subgraph that works the same way.
  • Set a layer to run on the CPU, if the data must be read at inference time.

The optimization doesn't affect what the model inputs or outputs.

To deactivate optimization, select your imported model asset then deactivate Optimize Model in the Model Asset Import Settings window.

Model inputs

The input dimensions for a model are either fixed or dynamic.

When you Inspect a model to get its inputs, each dimension in the an input shape displays one of the following:

  • A digit to represent a fixed dimension, for example (1). This means your input tensor should use the same dimension size, because Sentis optimizes the model for that size.
  • A question mark or a string to represent a dynamic dimension, for example (?) or (batch). This means the input tensor shape is symbolic, and your input tensor can use any dimension size.

If a model uses inputs with dynamic shapes, Sentis might not be able to optimize the model as efficiently as a model that uses fixed input dimensions. This might slow down the model.

Additional resources