MNIST_fastai-2.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "MNIST-fastai.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyMe/lvR4fs6F/J6k07W9Nyg",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/MohammedNayaz/Machine-learning-Internship-2/blob/master/MNIST_fastai-2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QDtL1fQHH9c9"
      },
      "source": [
        "# MNIST CNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PoDZgcR5HuA6"
      },
      "source": [
        "%reload_ext autoreload\n",
        "%autoreload 2\n",
        "%matplotlib inline"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KVBHsMe5IHrG"
      },
      "source": [
        "from fastai.vision import *"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KPZqNMRRIMkM"
      },
      "source": [
        "##Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qz3KBewuIHtd"
      },
      "source": [
        "path = untar_data(URLs.MNIST)"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lAR2JdBXIHvv",
        "outputId": "824a5d3c-e25c-4372-f234-35648166234e",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "path.ls()"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[PosixPath('/root/.fastai/data/mnist_png/testing'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training')]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TjCc3UJdIHyP"
      },
      "source": [
        "il = ImageList.from_folder(path, convert_mode='L')"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "P6nWmhNLIH0n",
        "outputId": "e20cdcb8-d748-4ce2-da53-58a1918181e0",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "il.items[0]"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "PosixPath('/root/.fastai/data/mnist_png/testing/1/9274.png')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FFNIpfN5IH28"
      },
      "source": [
        "defaults.cmap='binary'"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3bhrcg5kIH5Z",
        "outputId": "e1ad6527-44ed-4994-d5ac-f0bb094f39d1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "il"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "ImageList (70000 items)\n",
              "Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28)\n",
              "Path: /root/.fastai/data/mnist_png"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rHp-nWgSIH7t",
        "outputId": "949cc197-7419-476d-8621-67c5d9f84eb6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 194
        }
      },
      "source": [
        "il[0].show()"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 216x216 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W2C_5kNFIH-P",
        "outputId": "fb9ffa72-6691-4776-e9cd-c75e67c32637",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 194
        }
      },
      "source": [
        "il[1].show()"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAALEAAACxCAYAAACLKVzFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAESUlEQVR4nO3cP0vVbxzHYf3xAyczfAJuriI+gTZBBxdJEEHaWvyDLm7SEDgYNhaBQxJFuIWOgThIQU/AqRraWpsUW9rO56jneNDzPl7X+PFwcw8vbvA+3/Ptv7i46INk/931BuCmREw8ERNPxMQTMfFETLz/r/i7+ze6SX81dBITT8TEEzHxREw8ERNPxMQTMfFETDwRE0/ExBMx8URMPBETT8TEEzHxREw8ERNPxMQTMfFETDwRE++qXzvT5dbW1sr5zs5OOX/37l05n5+f79iebpuTmHgiJp6IiSdi4omYeG4nwvX3l68nazr/9OlTOXc7AXdIxMQTMfFETDwRE8/tRJCvX782zD5+/NjSGktLS53aTtdwEhNPxMQTMfFETDwRE8/tRBf6/v17OV9cXGyY/fr1q6W1Hzx40M6WupqTmHgiJp6IiSdi4omYeG4n7pnnz5+X8/fv39/yTjrHSUw8ERNPxMQTMfFETLz+i4uLy/5+6R+5mdPT03I+OTlZzn/8+HHttcfHx8v558+fy/nQ0NC1175D5cs0nMTEEzHxREw8ERNPxMTz7MQd2tvbK+c/f/4s583edFkJv4VoiZOYeCImnoiJJ2Li+cfuFmxtbZXz7e3tltYZHR1tmD158qT8bC/+A9eMk5h4IiaeiIknYuKJmHgeim/D2dlZOX/16lU5X15eLufNvkYeGBgo5/v7+w2zqamp8rM9ykPx9CYRE0/ExBMx8URMPM9OtKHZLcTKykpH1p+YmCjn9+wm4tqcxMQTMfFETDwRE0/ExHM70Ybj4+Ny3uw5lGbzx48fl/MPHz60t7F7yklMPBETT8TEEzHxREw8txOX+PbtWzk/PDws56288K+vr69vbGys5T3RyElMPBETT8TEEzHxREw8txP//P79u2H27Nmz8rN//vxpae2NjY1yvr6+3tI61JzExBMx8URMPBETT8TE81bMf968edMwe/r0aUfWPj8/78g6eCsmPUrExBMx8URMvHv3tfPa2lo5393dvfHaMzMzN16D1jmJiSdi4omYeCImnoiJ17NfOx8dHZXzR48elfNWfm6/sLBQzt++fXvtNWiLr53pTSImnoiJJ2LiiZh48c9OnJyclPMXL16U82a3ECMjIw2zubm58rObm5vX3B23wUlMPBETT8TEEzHxREy8+NuJly9flvODg4OW1qluIra2ttraE7fLSUw8ERNPxMQTMfFETLyY24nT09Ny/uXLl3L+8OHDcv769etyPj093d7GuHNOYuKJmHgiJp6IiSdi4sXcTgwODpbz4eHhcr66ulrOZ2dnO7YnuoOTmHgiJp6IiSdi4omYeD37Vkx6krdi0ptETDwRE0/ExBMx8URMPBETT8TEEzHxREw8ERNPxMQTMfFETDwRE0/ExBMx8URMPBETT8TEEzHxREy8q14oWP5EGrqJk5h4IiaeiIknYuKJmHgiJt5f/hiW+C4xmTsAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 216x216 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VaUnZJY1IfKY",
        "outputId": "77b11500-46d3-4924-c419-53ba0c997c65",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 194
        }
      },
      "source": [
        "il[00].show()"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 216x216 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C6RW1cVWIfND"
      },
      "source": [
        "sd = il.split_by_folder(train='training', valid='testing')"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jedI4p_hIfPR",
        "outputId": "a184df94-0d81-4b1c-9576-e66a829cbac9",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "sd"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "ItemLists;\n",
              "\n",
              "Train: ImageList (60000 items)\n",
              "Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28)\n",
              "Path: /root/.fastai/data/mnist_png;\n",
              "\n",
              "Valid: ImageList (10000 items)\n",
              "Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28)\n",
              "Path: /root/.fastai/data/mnist_png;\n",
              "\n",
              "Test: None"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WLukVKQEIfRy",
        "outputId": "60545333-086c-4766-a113-283b6399e881",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "(path/'training').ls()"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[PosixPath('/root/.fastai/data/mnist_png/training/1'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/7'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/5'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/3'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/0'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/6'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/2'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/4'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/8'),\n",
              " PosixPath('/root/.fastai/data/mnist_png/training/9')]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QuOhL0GYIfUb"
      },
      "source": [
        "ll = sd.label_from_folder()"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MNT8JhzOIfXG",
        "outputId": "e0fcfa8e-a328-4289-c460-7550abdfb6ab",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "ll"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LabelLists;\n",
              "\n",
              "Train: LabelList (60000 items)\n",
              "x: ImageList\n",
              "Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28)\n",
              "y: CategoryList\n",
              "1,1,1,1,1\n",
              "Path: /root/.fastai/data/mnist_png;\n",
              "\n",
              "Valid: LabelList (10000 items)\n",
              "x: ImageList\n",
              "Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28),Image (1, 28, 28)\n",
              "y: CategoryList\n",
              "1,1,1,1,1\n",
              "Path: /root/.fastai/data/mnist_png;\n",
              "\n",
              "Test: None"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "njCoAZ-TIfZ8"
      },
      "source": [
        "x,y = ll.train[0]"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5BVNsXl4Ifca",
        "outputId": "da308518-c521-4900-c133-fd4d9184253d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        }
      },
      "source": [
        "x.show()\n",
        "print(y,x.shape)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1 torch.Size([1, 28, 28])\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 216x216 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "59WOw27BIvy7"
      },
      "source": [
        "tfms = ([*rand_pad(padding=3, size=28, mode='zeros')], [])"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "laKGEaYOIv08"
      },
      "source": [
        "ll = ll.transform(tfms)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G9h6xVAsIv3W"
      },
      "source": [
        "bs = 128"
      ],
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "S5B3rwI8Iv5t"
      },
      "source": [
        "# not using imagenet_stats because not using pretrained model\n",
        "data = ll.databunch(bs=bs).normalize()"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1w86EmhcIv79"
      },
      "source": [
        "x,y = data.train_ds[0]"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ikWKfBgnIv-S",
        "outputId": "d7d2b875-deca-4658-b12f-71550f7d6646",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 211
        }
      },
      "source": [
        "x.show()\n",
        "print(y)"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 216x216 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TF6rvIlbIwAd",
        "outputId": "73db529d-7ee4-4068-e6c3-afdd74f0ff48",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 466
        }
      },
      "source": [
        "def _plot(i,j,ax): data.train_ds[0][0].show(ax, cmap='gray')\n",
        "plot_multi(_plot, 3, 3, figsize=(8,8))"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7MuRyQ9TIwCk",
        "outputId": "ac8e3d68-f0bd-4b0d-9ee8-b42a2c791dbd",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "xb,yb = data.one_batch()\n",
        "xb.shape,yb.shape"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(torch.Size([128, 1, 28, 28]), torch.Size([128]))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rQ7Qev1dIffb",
        "outputId": "ff6fc06e-5298-45a1-c988-5e5e4c6c9943",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        }
      },
      "source": [
        "data.show_batch(rows=3, figsize=(5,5))"
      ],
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 360x360 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7pGqJYHqJOXH"
      },
      "source": [
        "##Basic CNN with batchnorm"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7QPj6a_DJLgK"
      },
      "source": [
        "def conv(ni,nf): return nn.Conv2d(ni, nf, kernel_size=3, stride=2, padding=1)"
      ],
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PMEc9xzpJLik"
      },
      "source": [
        "model = nn.Sequential(\n",
        "    conv(1, 8), # 14\n",
        "    nn.BatchNorm2d(8),\n",
        "    nn.ReLU(),\n",
        "    conv(8, 16), # 7\n",
        "    nn.BatchNorm2d(16),\n",
        "    nn.ReLU(),\n",
        "    conv(16, 32), # 4\n",
        "    nn.BatchNorm2d(32),\n",
        "    nn.ReLU(),\n",
        "    conv(32, 16), # 2\n",
        "    nn.BatchNorm2d(16),\n",
        "    nn.ReLU(),\n",
        "    conv(16, 10), # 1\n",
        "    nn.BatchNorm2d(10),\n",
        "    Flatten()     # remove (1,1) grid\n",
        ")"
      ],
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MbbUBzXqJLlb"
      },
      "source": [
        "learn = Learner(data, model, loss_func = nn.CrossEntropyLoss(), metrics=accuracy)"
      ],
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c-PjoYvaJLp5",
        "outputId": "4b921ba8-c187-4979-d595-7dd8c4117e98",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "print(learn.summary())\n"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sequential\n",
            "======================================================================\n",
            "Layer (type)         Output Shape         Param #    Trainable \n",
            "======================================================================\n",
            "Conv2d               [8, 14, 14]          80         True      \n",
            "______________________________________________________________________\n",
            "BatchNorm2d          [8, 14, 14]          16         True      \n",
            "______________________________________________________________________\n",
            "ReLU                 [8, 14, 14]          0          False     \n",
            "______________________________________________________________________\n",
            "Conv2d               [16, 7, 7]           1,168      True      \n",
            "______________________________________________________________________\n",
            "BatchNorm2d          [16, 7, 7]           32         True      \n",
            "______________________________________________________________________\n",
            "ReLU                 [16, 7, 7]           0          False     \n",
            "______________________________________________________________________\n",
            "Conv2d               [32, 4, 4]           4,640      True      \n",
            "______________________________________________________________________\n",
            "BatchNorm2d          [32, 4, 4]           64         True      \n",
            "______________________________________________________________________\n",
            "ReLU                 [32, 4, 4]           0          False     \n",
            "______________________________________________________________________\n",
            "Conv2d               [16, 2, 2]           4,624      True      \n",
            "______________________________________________________________________\n",
            "BatchNorm2d          [16, 2, 2]           32         True      \n",
            "______________________________________________________________________\n",
            "ReLU                 [16, 2, 2]           0          False     \n",
            "______________________________________________________________________\n",
            "Conv2d               [10, 1, 1]           1,450      True      \n",
            "______________________________________________________________________\n",
            "BatchNorm2d          [10, 1, 1]           20         True      \n",
            "______________________________________________________________________\n",
            "Flatten              [10]                 0          False     \n",
            "______________________________________________________________________\n",
            "\n",
            "Total params: 12,126\n",
            "Total trainable params: 12,126\n",
            "Total non-trainable params: 0\n",
            "Optimized with 'torch.optim.adam.Adam', betas=(0.9, 0.99)\n",
            "Using true weight decay as discussed in https://www.fast.ai/2018/07/02/adam-weight-decay/ \n",
            "Loss function : CrossEntropyLoss\n",
            "======================================================================\n",
            "Callbacks functions applied \n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MrNOJGPmJLoC",
        "outputId": "70f6e95b-e287-4283-ace1-a2298afb3152",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        }
      },
      "source": [
        "xb = xb.cuda()"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "error",
          "ename": "RuntimeError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-32-76fc84feb73b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mxb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mxb\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcuda\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[0;32m/usr/local/lib/python3.6/dist-packages/torch/cuda/__init__.py\u001b[0m in \u001b[0;36m_lazy_init\u001b[0;34m()\u001b[0m\n\u001b[1;32m    188\u001b[0m             raise AssertionError(\n\u001b[1;32m    189\u001b[0m                 \"libcudart functions unavailable. It looks like you have a broken build?\")\n\u001b[0;32m--> 190\u001b[0;31m         \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_C\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_cuda_init\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    191\u001b[0m         \u001b[0;31m# Some of the queued calls may reentrantly call _lazy_init();\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    192\u001b[0m         \u001b[0;31m# we need to just return without initializing in that case.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mRuntimeError\u001b[0m: cuda runtime error (100) : no CUDA-capable device is detected at /pytorch/aten/src/THC/THCGeneral.cpp:47"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "24vjPPBuJlXe"
      },
      "source": [
        "model(xb).shape"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IOIJzJXIJlZ4"
      },
      "source": [
        "learn.lr_find(end_lr=100)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qwinzVgvJlcS"
      },
      "source": [
        "learn.recorder.plot()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B3UgpPceJle0"
      },
      "source": [
        "learn.fit_one_cycle(3, max_lr=0.1)"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}