PassManager

class onnx_ir.passes.PassManager(passes, steps=1, early_stop=True)

Pass manager for the IR.

The PassManager is a Pass that runs a sequence of passes on a model.

Example:

import onnx_ir as ir
import onnx_ir.passes.common as common_passes

model = ir.load("model.onnx")
passes = ir.passes.PassManager(
    [
        # Pass managers can be nested
        ir.passes.PassManager(
            [
                common_passes.DeduplicateHashedInitializersPass(size_limit=1024 * 1024),
                common_passes.CommonSubexpressionEliminationPass(),
            ],
            steps=2,
            early_stop=True,
        ),
        common_passes.ClearMetadataAndDocStringPass(),
    ],
    steps=2,
    early_stop=False,
)

# Apply the passes to the model
result = passes(model)

Parameters:

• passes (Sequence[PassBase])

• steps (int)

• early_stop (bool)

passes

The passes to run.

steps

The number of times to run the passes.

early_stop

Whether to stop running the passes if the graph stops changing.

call(model)

Run the set of passes steps number of times or until the graph stops changing.

Parameters:

model (Model)

Return type:

PassResult