{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "\n# Generating a FunctionProto\n\nThe example below shows how we can define Selu as a function in onnxscript.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "First, import the ONNX opset used to define the function.\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "from onnxscript import opset15 as op\nfrom onnxscript import script"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Next, define Selu as an ONNXScript function.\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "@script()\ndef Selu(X, alpha: float, gamma: float):\n    alphaX = op.CastLike(alpha, X)\n    gammaX = op.CastLike(gamma, X)\n    neg = gammaX * (alphaX * op.Exp(X) - alphaX)\n    pos = gammaX * X\n    zero = op.CastLike(0, X)\n    return op.Where(zero >= X, neg, pos)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "We can convert the ONNXScript function to an ONNX function (FunctionProto) as below:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "onnx_fun = Selu.to_function_proto()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Let's see what the translated function looks like:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "import onnx  # noqa: E402\n\nprint(onnx.printer.to_text(onnx_fun))"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.10.15"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}