DeduplicateHashedInitializersPass

class onnx_ir.passes.common.DeduplicateHashedInitializersPass(size_limit=4294967296)

Remove duplicated initializer tensors (using a hashed method) from the graph.

This pass detects initializers with identical shape, dtype, and hashed content, and replaces all duplicate references with a canonical one.

This pass should have a lower peak memory usage than DeduplicateInitializersPass as it does not store the full tensor data in memory, but instead uses a hash of the tensor data.

Added in version 0.1.7.

Parameters:

size_limit (int)

call(model)

The main entry point for the pass.

Parameters:

model (Model)

Return type:

PassResult