Skip to main content
Version: Latest

DeploymentSettings_ONNXRuntime_ONNX​

CLASS - DeploymentSettings_ONNXRuntime_ONNX(

)

Ancestors - (BaseDeploymentSettings)

Use this if you wish to deploy to ONNXRuntime in Settings.deployment_settings.

Class Variables​

  • graph_author (Optional[str]) - Name of the Author of the Graph to attach to the final Model File.
  • graph_description (Optional[str]) - Any description that should be attached to the final Model file
  • input_shapes_for_deployment (Union[list, tuple, List[list], List[tuple], Tuple[tuple], Tuple[list], None]) - Shapes to use for the deployment of the final Model file. Spatial Dimensions can be None if your model does not depend on it. For example, OK: Conv2d(...) AdaptiveAvgPool(output_size=(1,1)) Flatten() Linear() or Fully Convolutional model NOT OK: Conv2d(...) Flatten() Linear()
  • utilize_full_int_range (bool) - Set to True if you're planning to run the model on a CPU that supports AVX512-VNNI.

See Also

 https://onnxruntime.ai/docs/performance/model-optimizations/quantization.html#when-and-why-do-i-need-to-try-u8u8

DeploymentSettings_TFLite​

CLASS - DeploymentSettings_TFLite(

)

Ancestors - (BaseDeploymentSettings)

Use this if you wish to deploy to TFLite in Settings.deployment_settings.

Class Variables​

  • graph_author (Optional[str]) - Name of the Author of the Graph to attach to the final Model File.
  • graph_description (Optional[str]) - Any description that should be attached to the final Model file
  • input_shapes_for_deployment (Union[list, tuple, List[list], List[tuple], Tuple[tuple], Tuple[list], None]) - Shapes to use for the deployment of the final Model file. Spatial Dimensions can be None if your model does not depend on it. For example, OK: Conv2d(...) AdaptiveAvgPool(output_size=(1,1)) Flatten() Linear() or Fully Convolutional model NOT OK: Conv2d(...) Flatten() Linear()

DeploymentSettings_TensorRT_ONNX​

CLASS - DeploymentSettings_TensorRT_ONNX(

)

Ancestors - (BaseDeploymentSettings)

Use this if you wish to deploy to TensorRT in Settings.deployment_settings.

Class Variables​

  • graph_author (Optional[str]) - Name of the Author of the Graph to attach to the final Model File.
  • graph_description (Optional[str]) - Any description that should be attached to the final Model file
  • input_shapes_for_deployment (Union[list, tuple, List[list], List[tuple], Tuple[tuple], Tuple[list], None]) - Shapes to use for the deployment of the final Model file. Spatial Dimensions can be None if your model does not depend on it. For example, OK: Conv2d(...) AdaptiveAvgPool(output_size=(1,1)) Flatten() Linear() or Fully Convolutional model NOT OK: Conv2d(...) Flatten() Linear()