Emociones_caras_fastai.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "Emociones_caras_fastai",
      "provenance": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "WBxwpVBaKGMW"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "from matplotlib import pyplot as plt\n",
        "import scipy.io as sio\n",
        "import seaborn as sns\n",
        "from fastai.vision import *"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eN_1tHL9KLjS"
      },
      "source": [
        "import zipfile\n",
        "from pathlib import Path\n",
        "data_path = Path(\"/content/drive/MyDrive/CARAS/archive.zip\")\n",
        "\n",
        "with zipfile.ZipFile(str(data_path), 'r') as zip_ref:\n",
        "    zip_ref.extractall(\"./data\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iYR9yuKyLyuj"
      },
      "source": [
        "path = ('./data')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "R86Jh33lL9IO",
        "outputId": "ad61c57a-5a2c-4ab1-c81a-2743cfeaee81"
      },
      "source": [
        "ls data/train"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\u001b[0m\u001b[01;34mangry\u001b[0m/  \u001b[01;34mdisgust\u001b[0m/  \u001b[01;34mfear\u001b[0m/  \u001b[01;34mhappy\u001b[0m/  \u001b[01;34mneutral\u001b[0m/  \u001b[01;34msad\u001b[0m/  \u001b[01;34msurprise\u001b[0m/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cTPyzFTyKOuY"
      },
      "source": [
        "# Load data\n",
        "batch_size = 256\n",
        "size = 48\n",
        "data = ImageDataBunch.from_folder(path=path, train=\"train\", test=\"test\", \n",
        "                                  valid_pct=0.2, \n",
        "                                  ds_tfms=get_transforms(), size=size, \n",
        "                                  bs=batch_size\n",
        "                                  ).normalize(imagenet_stats)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K15hMvxjj5Ed",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "32b5bddd-d5be-43ad-8dad-3b9f96ff2c71"
      },
      "source": [
        "data"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "ImageDataBunch;\n",
              "\n",
              "Train: LabelList (22968 items)\n",
              "x: ImageList\n",
              "Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48)\n",
              "y: CategoryList\n",
              "happy,happy,happy,happy,happy\n",
              "Path: data;\n",
              "\n",
              "Valid: LabelList (5741 items)\n",
              "x: ImageList\n",
              "Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48)\n",
              "y: CategoryList\n",
              "neutral,happy,surprise,happy,neutral\n",
              "Path: data;\n",
              "\n",
              "Test: LabelList (7178 items)\n",
              "x: ImageList\n",
              "Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48),Image (3, 48, 48)\n",
              "y: EmptyLabelList\n",
              ",,,,\n",
              "Path: data"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pTOQvm0ULhUc",
        "outputId": "b0c06743-8486-42f8-80c2-0b97987857cc"
      },
      "source": [
        "data.classes"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['angry', 'disgust', 'fear', 'happy', 'neutral', 'sad', 'surprise']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x1WRm1r6R0p3"
      },
      "source": [
        "### prueba 1\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KxORAEXaMUQd"
      },
      "source": [
        "learn = cnn_learner(data, models.densenet201, metrics=accuracy)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "of2RPWCuMtRN",
        "outputId": "6b63ec2a-ea34-4fac-c6bd-58adc08398ad"
      },
      "source": [
        "learn.lr_find()\n",
        "learn.recorder.plot()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='1' class='' max='2' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      50.00% [1/2 00:25<00:25]\n",
              "    </div>\n",
              "    \n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>8.541283</td>\n",
              "      <td>#na#</td>\n",
              "      <td>00:25</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table><p>\n",
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='1' class='' max='89' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      1.12% [1/89 00:00<01:01 10.5588]\n",
              "    </div>\n",
              "    "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EdONc6e3pC3G"
      },
      "source": [
        "lr = 1e-02"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 634
        },
        "id": "8z3mkVA2pHG9",
        "outputId": "489f20b5-1bc5-455c-f982-cf5e5f408b60"
      },
      "source": [
        "learn.fit_one_cycle(19, max_lr=slice(lr))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>2.468974</td>\n",
              "      <td>1.936624</td>\n",
              "      <td>0.335482</td>\n",
              "      <td>00:48</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>1.988539</td>\n",
              "      <td>1.674327</td>\n",
              "      <td>0.356906</td>\n",
              "      <td>00:48</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>1.763578</td>\n",
              "      <td>1.615072</td>\n",
              "      <td>0.390176</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>3</td>\n",
              "      <td>1.707965</td>\n",
              "      <td>1.519914</td>\n",
              "      <td>0.425710</td>\n",
              "      <td>00:48</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>4</td>\n",
              "      <td>1.664792</td>\n",
              "      <td>1.435571</td>\n",
              "      <td>0.459153</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>5</td>\n",
              "      <td>1.607781</td>\n",
              "      <td>1.411314</td>\n",
              "      <td>0.479533</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>6</td>\n",
              "      <td>1.535875</td>\n",
              "      <td>1.321792</td>\n",
              "      <td>0.499390</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>7</td>\n",
              "      <td>1.443728</td>\n",
              "      <td>1.343979</td>\n",
              "      <td>0.511932</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>8</td>\n",
              "      <td>1.394863</td>\n",
              "      <td>1.242817</td>\n",
              "      <td>0.531789</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>9</td>\n",
              "      <td>1.329616</td>\n",
              "      <td>1.220766</td>\n",
              "      <td>0.542414</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>10</td>\n",
              "      <td>1.283782</td>\n",
              "      <td>1.211516</td>\n",
              "      <td>0.539976</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>11</td>\n",
              "      <td>1.247748</td>\n",
              "      <td>1.184205</td>\n",
              "      <td>0.553736</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>12</td>\n",
              "      <td>1.213557</td>\n",
              "      <td>1.175735</td>\n",
              "      <td>0.549382</td>\n",
              "      <td>00:46</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>13</td>\n",
              "      <td>1.176526</td>\n",
              "      <td>1.161512</td>\n",
              "      <td>0.557394</td>\n",
              "      <td>00:46</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>14</td>\n",
              "      <td>1.155651</td>\n",
              "      <td>1.154452</td>\n",
              "      <td>0.555826</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>15</td>\n",
              "      <td>1.136914</td>\n",
              "      <td>1.141792</td>\n",
              "      <td>0.566103</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>16</td>\n",
              "      <td>1.131203</td>\n",
              "      <td>1.140371</td>\n",
              "      <td>0.565058</td>\n",
              "      <td>00:47</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>17</td>\n",
              "      <td>1.114794</td>\n",
              "      <td>1.136304</td>\n",
              "      <td>0.567323</td>\n",
              "      <td>00:46</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>18</td>\n",
              "      <td>1.107901</td>\n",
              "      <td>1.136619</td>\n",
              "      <td>0.566452</td>\n",
              "      <td>00:46</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GKN3UaJygDqb"
      },
      "source": [
        "learn.save(\"score_55+\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        },
        "id": "UfeNTLde3-Wk",
        "outputId": "a948ec9b-70ee-4afc-ab8f-86d88a6c7194"
      },
      "source": [
        "learn.lr_find()\n",
        "learn.recorder.plot()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='0' class='' max='2' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      0.00% [0/2 00:00<00:00]\n",
              "    </div>\n",
              "    \n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "  </tbody>\n",
              "</table><p>\n",
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='88' class='' max='89' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      98.88% [88/89 00:42<00:00 3.6426]\n",
              "    </div>\n",
              "    "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d-V4CLeK2Ce1"
      },
      "source": [
        "lr_2 = 1e-03"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O87r2WYjh9r9"
      },
      "source": [
        "### Prueba 2"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zOYpqlPliFxw"
      },
      "source": [
        "learn2 = cnn_learner(data, models.resnet50, metrics=accuracy)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 405
        },
        "id": "EpIILRZ6hgAJ",
        "outputId": "08381c76-2952-4a7f-cdd9-f34ecc3a03d5"
      },
      "source": [
        "learn2.lr_find()\n",
        "learn2.recorder.plot()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='1' class='' max='2' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      50.00% [1/2 00:25<00:25]\n",
              "    </div>\n",
              "    \n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>9.405434</td>\n",
              "      <td>#na#</td>\n",
              "      <td>00:25</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table><p>\n",
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='0' class='' max='89' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      0.00% [0/89 00:00<00:00]\n",
              "    </div>\n",
              "    "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EbfJk9hviCRG"
      },
      "source": [
        "lr2 = 0.05"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "7tPDj5lHhknW",
        "outputId": "e9dbcc71-8a99-483a-b2de-26dfa570ed18"
      },
      "source": [
        "learn2.fit_one_cycle(1, max_lr=slice(lr2))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>2.044993</td>\n",
              "      <td>1.855206</td>\n",
              "      <td>0.354990</td>\n",
              "      <td>00:27</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>1.916044</td>\n",
              "      <td>3.230865</td>\n",
              "      <td>0.383557</td>\n",
              "      <td>00:27</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>1.636576</td>\n",
              "      <td>1.385952</td>\n",
              "      <td>0.484062</td>\n",
              "      <td>00:27</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>3</td>\n",
              "      <td>1.455296</td>\n",
              "      <td>1.368102</td>\n",
              "      <td>0.500784</td>\n",
              "      <td>00:27</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>4</td>\n",
              "      <td>1.340249</td>\n",
              "      <td>1.255904</td>\n",
              "      <td>0.534924</td>\n",
              "      <td>00:27</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>5</td>\n",
              "      <td>1.255077</td>\n",
              "      <td>1.214092</td>\n",
              "      <td>0.540498</td>\n",
              "      <td>00:27</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>6</td>\n",
              "      <td>1.197735</td>\n",
              "      <td>1.187108</td>\n",
              "      <td>0.553388</td>\n",
              "      <td>00:27</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>7</td>\n",
              "      <td>1.163750</td>\n",
              "      <td>1.154857</td>\n",
              "      <td>0.557917</td>\n",
              "      <td>00:27</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V_EEetT0pU74"
      },
      "source": [
        "learn2.save(\"/content/drive/MyDrive/CARAS/model_score_55+\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T9fqQErwq_eP"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ld-oWFWfh4_j"
      },
      "source": [
        "learn2.unfreeze()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QXPivE_8h9mF",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        },
        "outputId": "dd8b0b39-9880-4a17-896b-10ffa8c24ac1"
      },
      "source": [
        "learn2.lr_find()\n",
        "learn2.recorder.plot()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='0' class='' max='2' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      0.00% [0/2 00:00<00:00]\n",
              "    </div>\n",
              "    \n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "  </tbody>\n",
              "</table><p>\n",
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='85' class='' max='89' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      95.51% [85/89 00:25<00:01 2.9853]\n",
              "    </div>\n",
              "    "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bZ5BolUmiA8-"
      },
      "source": [
        "lr2_2 = 1e-05"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zE_WRRp8iFad",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 218
        },
        "outputId": "92a6fc28-a4d2-4b25-f297-97eaccd5f475"
      },
      "source": [
        "learn2.fit_one_cycle(5, max_lr=slice(lr2_2/100,lr2_2))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='4' class='' max='5' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      80.00% [4/5 01:55<00:28]\n",
              "    </div>\n",
              "    \n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: left;\">\n",
              "      <th>epoch</th>\n",
              "      <th>train_loss</th>\n",
              "      <th>valid_loss</th>\n",
              "      <th>accuracy</th>\n",
              "      <th>time</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>0</td>\n",
              "      <td>1.140181</td>\n",
              "      <td>1.147283</td>\n",
              "      <td>0.562968</td>\n",
              "      <td>00:28</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>1.136412</td>\n",
              "      <td>1.138949</td>\n",
              "      <td>0.564884</td>\n",
              "      <td>00:28</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>1.137655</td>\n",
              "      <td>1.136727</td>\n",
              "      <td>0.566974</td>\n",
              "      <td>00:28</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>3</td>\n",
              "      <td>1.130904</td>\n",
              "      <td>1.141842</td>\n",
              "      <td>0.565755</td>\n",
              "      <td>00:28</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table><p>\n",
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='10' class='' max='23' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      43.48% [10/23 00:01<00:01]\n",
              "    </div>\n",
              "    "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}