Implementation:Onnx Onnx Compose Add Prefix

Knowledge Sources	ONNX ONNX Compose API
Domains	Model_Composition, Name_Management
Last Updated	2026-02-10 00:00 GMT

Overview

Concrete tool for adding namespace prefixes to ONNX model names provided by the ONNX compose module.

Description

The add_prefix function adds a prefix string to all named elements in an ONNX model, including nodes, edges, inputs, outputs, initializers, value infos, and local functions. Each category of names can be selectively enabled or disabled via boolean parameters. The function can operate in-place or return a copy. Internally, it delegates to add_prefix_graph for graph-level renaming and handles function renaming separately.

Usage

Import this function when preparing models for composition via merge_models. Apply distinct prefixes to each model before merging to prevent name collisions. Empty names are not prefixed.

Code Reference

Source Location

• Repository: onnx
• File: onnx/compose.py
• Lines: 564-635

Signature

def add_prefix(
    model: ModelProto,
    prefix: str,
    rename_nodes: bool | None = True,
    rename_edges: bool | None = True,
    rename_inputs: bool | None = True,
    rename_outputs: bool | None = True,
    rename_initializers: bool | None = True,
    rename_value_infos: bool | None = True,
    rename_functions: bool | None = True,
    inplace: bool | None = False,
) -> ModelProto:
    """Adds a prefix to names of elements in a model.

    Args:
        model: Model to prefix.
        prefix: Prefix string to prepend to each name.
        rename_nodes: Whether to prefix node names.
        rename_edges: Whether to prefix edge (input/output) names on nodes.
        rename_inputs: Whether to prefix graph input names.
        rename_outputs: Whether to prefix graph output names.
        rename_initializers: Whether to prefix initializer names.
        rename_value_infos: Whether to prefix value info names.
        rename_functions: Whether to prefix local function names.
        inplace: If True, mutate the model directly; otherwise copy first.

    Returns:
        ModelProto with prefixed names.
    """

Import

from onnx import compose

I/O Contract

Inputs

Name	Type	Required	Description
model	ModelProto	Yes	Model whose names to prefix
prefix	str	Yes	Prefix string to prepend
rename_nodes	bool	No	Prefix node names (default: True)
rename_edges	bool	No	Prefix edge names (default: True)
rename_inputs	bool	No	Prefix graph input names (default: True)
rename_outputs	bool	No	Prefix graph output names (default: True)
rename_initializers	bool	No	Prefix initializer names (default: True)
rename_value_infos	bool	No	Prefix value info names (default: True)
rename_functions	bool	No	Prefix local function names (default: True)
inplace	bool	No	Mutate directly or copy (default: False)

Outputs

Name	Type	Description
return	ModelProto	Model with all selected names prefixed

Usage Examples

Basic Prefixing

import onnx
from onnx import compose

model = onnx.load_model("model_a.onnx")

# Add "m1/" prefix to all names
prefixed = compose.add_prefix(model, prefix="m1/")

# Verify input names are prefixed
for inp in prefixed.graph.input:
    print(inp.name)  # e.g., "m1/X", "m1/W"

Selective Prefixing

from onnx import compose

# Only prefix internal names, keep inputs/outputs unchanged
prefixed = compose.add_prefix(
    model,
    prefix="net1_",
    rename_inputs=False,
    rename_outputs=False,
)

Related Pages

Implements Principle

• Principle:Onnx_Onnx_Namespace_Prefixing

Requires Environment

• Environment:Onnx_Onnx_Python_Runtime_Environment