ort_patching
This module contains all the patched functions from ORT.
Functions
Load model while performing symbolic shape infer and ONNX shape inference. |
|
Patches the ORT modules. |
|
Save the original model and load model with shape inferred. |
- load_model_with_shape_infer(model_path)
Load model while performing symbolic shape infer and ONNX shape inference.
- Parameters:
model_path (Path)
- Return type:
ModelProto
- patch_ort_modules(calibrate_per_node=False)
Patches the ORT modules.
- Parameters:
calibrate_per_node (bool)
- save_and_reload_model_with_shape_infer(model)
Save the original model and load model with shape inferred.
- Parameters:
model (ModelProto)
- Return type:
ModelProto