module onnxrt.optim.onnx_optimisation_redundant¶
Short summary¶
module mlprodict.onnxrt.optim.onnx_optimisation_redundant
Optimisation of ONNX graphs.
Functions¶
function |
truncated documentation |
|---|---|
Hash the content of an object. |
|
Removes redundant part of the graph. A redundant part is a set of nodes which takes the same inputs and produces … |
Documentation¶
Optimisation of ONNX graphs.
-
mlprodict.onnxrt.optim.onnx_optimisation_redundant._hash_obj_content(obj, max_size=1000)¶ Hash the content of an object.
-
mlprodict.onnxrt.optim.onnx_optimisation_redundant.onnx_remove_node_redundant(onnx_model, recursive=True, debug_info=None, max_hash_size=1000, **options)¶ Removes redundant part of the graph. A redundant part is a set of nodes which takes the same inputs and produces the same outputs. It first starts by looking into duplicated initializers, then looks into nodes taking the same inputs and sharing the same type and parameters.
- Parameters
onnx_model – onnx model
recursive – looks into subgraphs
debug_info – debug information (private)
max_hash_size – limit the size of a hash used to detect identical subgraphs
options – additional options (unused)
- Returns
new onnx _model
