ModelProto Properties

A ModelProto, in ONNX, usually stores extra information beyond the computational graph, such as ir_version or producer_name. Such properties of a generated ModelProto can be set by passing in extra named parameters to the call to script (or the call to to_model_proto), as illustrated by the example below. Only the valid fields defined in the protobuf message ModelProto should be specified in this fashion.

First, we define the implementation of a square-loss function in onnxscript.

import onnx

from onnxscript import FLOAT, script
from onnxscript import opset15 as op


@script(ir_version=7, producer_name="OnnxScript", producer_version="0.1")
def square_loss(X: FLOAT["N"], Y: FLOAT["N"]) -> FLOAT[1]:  # noqa: F821
    diff = X - Y
    return op.ReduceSum(diff * diff, keepdims=1)

/opt/hostedtoolcache/Python/3.10.15/x64/lib/python3.10/site-packages/onnxscript/converter.py:820: FutureWarning: 'onnxscript.values.Op.param_schemas' is deprecated in version 0.1 and will be removed in the future. Please use '.op_signature' instead.
  param_schemas = callee.param_schemas()

Let’s see what the generated model looks like.

model = square_loss.to_model_proto()
print(onnx.printer.to_text(model))

<
   ir_version: 7,
   opset_import: ["" : 15],
   producer_name: "OnnxScript",
   producer_version: "0.1"
>
square_loss (float[N] X, float[N] Y) => (float[1] return_val) {
   diff = Sub (X, Y)
   tmp = Mul (diff, diff)
   return_val = ReduceSum <keepdims: int = 1> (tmp)
}