Implementation:Onnx Onnx Printer To Text

Knowledge Sources	ONNX ONNX Printer API
Domains	Debugging, Model_Analysis
Last Updated	2026-02-10 00:00 GMT

Overview

Concrete tool for converting ONNX protobuf structures to human-readable text format, wrapping a C++ printer engine.

Description

The to_text function converts ModelProto, FunctionProto, or GraphProto objects into their ONNX text format representation. It is a Python wrapper around the C++ printer implementation (onnx_cpp2py_export.printer), which serializes the proto to bytes and then generates the text output. The function dispatches to model_to_text, function_to_text, or graph_to_text based on the input type.

Usage

Import this function when you need to inspect the structure of an ONNX model in a human-readable format. It is particularly useful after shape inference (to verify inferred types) or after model transformations (to verify correctness).

Code Reference

Source Location

Repository: onnx
File: onnx/printer.py
Lines: 10-17

Signature

def to_text(proto: onnx.ModelProto | onnx.FunctionProto | onnx.GraphProto) -> str:
    """Convert an ONNX protobuf to its text representation.

    Args:
        proto: ModelProto, FunctionProto, or GraphProto to convert.

    Returns:
        Human-readable text representation of the protobuf.

    Raises:
        TypeError: If proto is not a supported type.
    """

Import

from onnx import printer

I/O Contract

Inputs

Name	Type	Required	Description
proto	ModelProto or FunctionProto or GraphProto	Yes	The protobuf structure to convert to text

Outputs

Name	Type	Description
return	str	Human-readable ONNX text format string

Usage Examples

Print Model Structure

import onnx
from onnx import printer

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

# Convert to text and print
text = printer.to_text(model)
print(text)

Print Graph Only

import onnx
from onnx import printer

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

# Print just the graph (without model-level metadata)
graph_text = printer.to_text(model.graph)
print(graph_text)

Related Pages

Implements Principle

Principle:Onnx_Onnx_Model_Inspection

Requires Environment

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment