onnx_ir.passes.common¶

Built-in passes provided by the ONNX IR

See Writing transformation passes for composition and custom pass authoring, Subgraphs and functions for inlining, and Model I/O for validation workflows.

Built-in passes¶

`AddDefaultAttributesPass`	Add default values for optional attributes that are not present in nodes.
`AddInitializersToInputsPass`	Add initializers to inputs.
`CheckerPass`	Run onnx checker on the model.
`ClearMetadataAndDocStringPass`	Clear all metadata and docstring from the model, graphs, nodes, and functions.
`CommonSubexpressionEliminationPass`	Eliminate common subexpression in ONNX graphs.
`DeduplicateHashedInitializersPass`	Remove duplicated initializer tensors (using a hashed method) from the graph.
`DeduplicateInitializersPass`	Remove duplicated initializer tensors from the main graph and all subgraphs.
`IdentityEliminationPass`	Pass for eliminating redundant Identity nodes.
`InlinePass`	Inline model local functions to the main graph and functions and remove unused functions.
`LiftConstantsToInitializersPass`	Lift constants to initializers.
`LiftSubgraphInitializersToMainGraphPass`	Lift subgraph initializers to main graph.
`NameFixPass`	Pass for fixing names to ensure all values and nodes have unique names.
`OutputFixPass`	Pass for adding Identity nodes to fix invalid output configurations.
`RemoveInitializersFromInputsPass`	Remove initializers from inputs.
`RemoveUnusedFunctionsPass`
`RemoveUnusedNodesPass`	Pass for removing unused nodes and initializers (dead code elimination).
`RemoveUnusedOpsetsPass`	Remove unused opset imports from the model and functions.
`ShapeInferencePass`	This pass performs shape inference on the graph.
`TopologicalSortPass`	Topologically sort graphs and functions in a model.