ch02_1_NumPy에서부터_JAX_시작하기.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "authorship_tag": "ABX9TyMaxGX0aIoobZlwDrlvDyut",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/yblee110/jax-flax-book/blob/main/ch02_1_NumPy%EC%97%90%EC%84%9C%EB%B6%80%ED%84%B0_JAX_%EC%8B%9C%EC%9E%91%ED%95%98%EA%B8%B0.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install jax==0.4.26\n",
        "import jax\n",
        "print(jax.__version__)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gVQk9h-O9KMh",
        "outputId": "eb991f9a-9575-470a-a89c-4065d3f9f2d2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: jax==0.4.26 in /usr/local/lib/python3.10/dist-packages (0.4.26)\n",
            "Requirement already satisfied: ml-dtypes>=0.2.0 in /usr/local/lib/python3.10/dist-packages (from jax==0.4.26) (0.2.0)\n",
            "Requirement already satisfied: numpy>=1.22 in /usr/local/lib/python3.10/dist-packages (from jax==0.4.26) (1.25.2)\n",
            "Requirement already satisfied: opt-einsum in /usr/local/lib/python3.10/dist-packages (from jax==0.4.26) (3.3.0)\n",
            "Requirement already satisfied: scipy>=1.9 in /usr/local/lib/python3.10/dist-packages (from jax==0.4.26) (1.11.4)\n",
            "0.4.26\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "5K5EbEQmgpOv",
        "outputId": "3e07ac6d-6ab7-446d-a6d0-5142f2a659d8"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "array([ 6.,  8., 10., 12.])"
            ]
          },
          "metadata": {}
        }
      ],
      "source": [
        "import jax\n",
        "import jax.numpy as jnp\n",
        "import numpy as np\n",
        "from IPython.display import display\n",
        "\n",
        "x = np.array([1.0, 2.0, 3.0, 4.0])\n",
        "y = np.array([5.0, 6.0, 7.0, 8.0])\n",
        "z = x + y\n",
        "\n",
        "display(z)  # 출력: array([6., 8., 10., 12.])"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "x = jnp.array([1.0, 2.0, 3.0, 4.0])\n",
        "y = jnp.array([5.0, 6.0, 7.0, 8.0])\n",
        "z = x + y\n",
        "display(z)  # 출력: Array([6., 8., 10., 12.], dtype=float32)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "AO_ZxtB5hB7S",
        "outputId": "46638aed-07ce-4d70-ec36-55b049ebde38"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array([ 6.,  8., 10., 12.], dtype=float32)"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def f(x):\n",
        "    return x ** 2\n",
        "\n",
        "df = jax.grad(f)\n",
        "\n",
        "display(df(2.0))  # 출력: Array(4., dtype=float32)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "TeCs2i-phEC9",
        "outputId": "3344a288-09ec-498f-887b-bfb5bdbf403e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array(4., dtype=float32, weak_type=True)"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def f(x):\n",
        "    return x ** 2\n",
        "\n",
        "d2f = jax.grad(jax.grad(f))\n",
        "\n",
        "display(d2f(2.0))  # 출력: Array(2., dtype=float32)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "rEzmARY6hGHl",
        "outputId": "13275fb7-07cf-43b4-aa99-059ef7c52aa6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array(2., dtype=float32, weak_type=True)"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def mse_loss(y_true, y_pred):\n",
        "    return jnp.mean((y_true - y_pred) ** 2)\n",
        "\n",
        "# 임의의 예측값과 실제값\n",
        "y_true = jnp.array([1.0, 2.0, 3.0])\n",
        "y_pred = jnp.array([1.5, 1.5, 3.5])\n",
        "\n",
        "# 손실함수의 그래디언트를 계산하기 위한 함수 정의\n",
        "grad_mse_loss = jax.grad(mse_loss, argnums=1)\n",
        "\n",
        "# 그래디언트 계산\n",
        "gradient = grad_mse_loss(y_true, y_pred)\n",
        "\n",
        "display(gradient)  # 출력: Array([ 0.33333334, -0.33333334,  0.33333334], dtype=float32)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "zi0rsthphHdB",
        "outputId": "1771d13e-2297-41c6-8bcf-83feec569098"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array([ 0.33333334, -0.33333334,  0.33333334], dtype=float32)"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def mse_loss(y_true, y_pred):\n",
        "    return jnp.mean((y_true - y_pred) ** 2)\n",
        "\n",
        "# 임의의 예측값과 실제값\n",
        "y_true = jnp.array([1.0, 2.0, 3.0])\n",
        "y_pred = jnp.array([1.5, 1.5, 3.5])\n",
        "\n",
        "# 손실함수와 그래디언트를 동시에 계산하기 위한 함수 정의\n",
        "value_and_grad_mse_loss = jax.value_and_grad(mse_loss, argnums=1)\n",
        "\n",
        "# 손실값과 그래디언트 동시에 계산\n",
        "loss_value, gradient = value_and_grad_mse_loss(y_true, y_pred)\n",
        "\n",
        "display('Loss Value:', loss_value)  # 출력: Loss Value: 0.25\n",
        "display('Gradient:', gradient)  # 출력: Gradient: Array([ 0.33333334, -0.33333334,  0.33333334], dtype=float32)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        },
        "id": "HNSfPck9hJMt",
        "outputId": "999583fe-bddc-408f-dd06-f6008f4fd95b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "'Loss Value:'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array(0.25, dtype=float32)"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "'Gradient:'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array([ 0.33333334, -0.33333334,  0.33333334], dtype=float32)"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def mse_with_aux(pred, target):\n",
        "    error = pred - target\n",
        "    mse = np.mean(error ** 2)\n",
        "    return mse, error  # MSE 결과와 보조 결과인 에러를 반환\n",
        "\n",
        "# 가정된 예측값과 타겟값\n",
        "pred = np.array([1.0, 2.0, 3.0])\n",
        "target = np.array([1.5, 2.5, 3.5])\n",
        "\n",
        "grad_fun, aux = jax.grad(mse_with_aux)(pred, target) # TypeError 발생"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 478
        },
        "id": "uWRbXovfhYSQ",
        "outputId": "acdbace2-6366-410b-fb17-2b23d1f5b6be"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "TypeError",
          "evalue": "Gradient only defined for scalar-output functions. Output was (Array(0.25, dtype=float32), Array([-0.5, -0.5, -0.5], dtype=float32)).",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
            "    \u001b[0;31m[... skipping hidden 1 frame]\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/jax/_src/core.py\u001b[0m in \u001b[0;36mget_aval\u001b[0;34m(x)\u001b[0m\n\u001b[1;32m   1479\u001b[0m   \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1480\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0mconcrete_aval\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1481\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/jax/_src/core.py\u001b[0m in \u001b[0;36mconcrete_aval\u001b[0;34m(x)\u001b[0m\n\u001b[1;32m   1471\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mconcrete_aval\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__jax_array__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1472\u001b[0;31m   raise TypeError(f\"Value {x!r} with type {type(x)} is not a valid JAX \"\n\u001b[0m\u001b[1;32m   1473\u001b[0m                    \"type\")\n",
            "\u001b[0;31mTypeError\u001b[0m: Value (Array(0.25, dtype=float32), Array([-0.5, -0.5, -0.5], dtype=float32)) with type <class 'tuple'> is not a valid JAX type",
            "\nThe above exception was the direct cause of the following exception:\n",
            "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-12-e16f80781df3>\u001b[0m in \u001b[0;36m<cell line: 10>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0mtarget\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1.5\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m2.5\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m3.5\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 10\u001b[0;31m \u001b[0mgrad_fun\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maux\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mjax\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mgrad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmse_with_aux\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpred\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;31m# TypeError 발생\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "    \u001b[0;31m[... skipping hidden 4 frame]\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/jax/_src/api.py\u001b[0m in \u001b[0;36m_check_scalar\u001b[0;34m(x)\u001b[0m\n\u001b[1;32m    735\u001b[0m     \u001b[0maval\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcore\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget_aval\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    736\u001b[0m   \u001b[0;32mexcept\u001b[0m \u001b[0mTypeError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 737\u001b[0;31m     \u001b[0;32mraise\u001b[0m \u001b[0mTypeError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmsg\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34mf\"was {x}\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    738\u001b[0m   \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    739\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0maval\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mShapedArray\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mTypeError\u001b[0m: Gradient only defined for scalar-output functions. Output was (Array(0.25, dtype=float32), Array([-0.5, -0.5, -0.5], dtype=float32))."
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "grad_fun, aux = jax.grad(mse_with_aux, has_aux=True)(pred, target) # 정상적으로 동작"
      ],
      "metadata": {
        "id": "18qoHbMphlUd"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def f(x):\n",
        "    return x ** 2\n",
        "\n",
        "df = jax.grad(f)\n",
        "d2f = jax.grad(df) # jax.grad(jax.grad(f)) 와 같다\n",
        "\n",
        "display(df(2.0))  # 출력: Array(4., dtype=float32)\n",
        "display(d2f(2.0))  # 출력: Array(2., dtype=float32)\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "id": "26OGJ2uPhqTP",
        "outputId": "78e4a8c0-6003-4c3d-d294-ac1ab008e44c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array(4., dtype=float32, weak_type=True)"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array(2., dtype=float32, weak_type=True)"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "x = np.array([1, 2, 3])\n",
        "\n",
        "def in_place_modify(x):\n",
        "  x[0] = 123\n",
        "  return None\n",
        "\n",
        "in_place_modify(x)\n",
        "display(x)  # 출력: array([123, 2, 3])\n",
        "\n",
        "in_place_modify(jnp.array(x)) # TypeError 발생"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 356
        },
        "id": "XomzN2Xdhqn9",
        "outputId": "1d1d2d63-582c-49a6-a579-2f8808421269"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "array([123,   2,   3])"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "error",
          "ename": "TypeError",
          "evalue": "'<class 'jaxlib.xla_extension.ArrayImpl'>' object does not support item assignment. JAX arrays are immutable. Instead of ``x[idx] = y``, use ``x = x.at[idx].set(y)`` or another .at[] method: https://jax.readthedocs.io/en/latest/_autosummary/jax.numpy.ndarray.at.html",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-15-489e6a88d940>\u001b[0m in \u001b[0;36m<cell line: 10>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0mdisplay\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# 출력: array([123, 2, 3])\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 10\u001b[0;31m \u001b[0min_place_modify\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mjnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;31m# TypeError 발생\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m<ipython-input-15-489e6a88d940>\u001b[0m in \u001b[0;36min_place_modify\u001b[0;34m(x)\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0min_place_modify\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 4\u001b[0;31m   \u001b[0mx\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m123\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      5\u001b[0m   \u001b[0;32mreturn\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/jax/_src/numpy/array_methods.py\u001b[0m in \u001b[0;36m_unimplemented_setitem\u001b[0;34m(self, i, x)\u001b[0m\n\u001b[1;32m    276\u001b[0m          \u001b[0;34m\"or another .at[] method: \"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    277\u001b[0m          \"https://jax.readthedocs.io/en/latest/_autosummary/jax.numpy.ndarray.at.html\")\n\u001b[0;32m--> 278\u001b[0;31m   \u001b[0;32mraise\u001b[0m \u001b[0mTypeError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmsg\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mformat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    279\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    280\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0m_operator_round\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnumber\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mArrayLike\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mndigits\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mint\u001b[0m \u001b[0;34m|\u001b[0m \u001b[0;32mNone\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m->\u001b[0m \u001b[0mArray\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mTypeError\u001b[0m: '<class 'jaxlib.xla_extension.ArrayImpl'>' object does not support item assignment. JAX arrays are immutable. Instead of ``x[idx] = y``, use ``x = x.at[idx].set(y)`` or another .at[] method: https://jax.readthedocs.io/en/latest/_autosummary/jax.numpy.ndarray.at.html"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import jax.numpy as jnp\n",
        "\n",
        "x = jnp.array([1, 2, 3])\n",
        "\n",
        "def jax_modify(x):\n",
        "  return x.at[0].set(123)\n",
        "\n",
        "y = jax_modify(x)\n",
        "display(x)  # 출력: Array([1, 2, 3], dtype=int32)\n",
        "display(y)  # 출력: Array([123,   2,   3], dtype=int32)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "id": "t4eMtfBuhqqP",
        "outputId": "a0151f82-288c-4f9a-9662-24c9c4d7e9f1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array([1, 2, 3], dtype=int32)"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Array([123,   2,   3], dtype=int32)"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "xs = np.random.normal(size=(100,))\n",
        "noise = np.random.normal(scale=0.1, size=(100,))\n",
        "ys = xs * 3 - 1 + noise\n",
        "\n",
        "plt.scatter(xs, ys);\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 430
        },
        "id": "REVYOERVhqsj",
        "outputId": "92352e8a-e612-4149-88f9-720d9d61386d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def model(theta, x):\n",
        "  \"\"\"Computes wx + b on a batch of input x.\"\"\"\n",
        "  w, b = theta\n",
        "  return w * x + b"
      ],
      "metadata": {
        "id": "L-qexSiNhqvA"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def loss_fn(theta, x, y):\n",
        "  prediction = model(theta, x)\n",
        "  return jnp.mean((prediction-y)**2)"
      ],
      "metadata": {
        "id": "YzNbhuP-hqxR"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def update(theta, x, y, lr=0.1):\n",
        "  return theta - lr * jax.grad(loss_fn)(theta, x, y)"
      ],
      "metadata": {
        "id": "xxoPl3Pkhqzx"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "theta = jnp.array([1., 1.])\n",
        "\n",
        "for _ in range(1000):\n",
        "  theta = update(theta, xs, ys)\n",
        "\n",
        "plt.scatter(xs, ys)\n",
        "plt.plot(xs, model(theta, xs))\n",
        "\n",
        "w, b = theta\n",
        "print(f\"w: {w:<.2f}, b: {b:<.2f}\")  # 출력: w: 3.00, b: -1.00\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 447
        },
        "id": "Lh_p3Njrhq2B",
        "outputId": "2c388282-eb51-46b4-f61c-3f008c3019c3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "w: 3.01, b: -1.02\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "6hd-6alIhq4Y"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}