ULMFiT_tutorial.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "ULMFiT tutorial.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gPb1g0g6WRSp",
        "colab_type": "text"
      },
      "source": [
        "# Using FastAI ULMFiT to make a state-of-the-art text classifier\n",
        "\n",
        "Refer to the full tutorial here:\n",
        "https://medium.com/technonerds/using-fastais-ulmfit-to-make-a-state-of-the-art-multi-label-text-classifier-bf54e2943e83"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "X7tKdQJdWeT_",
        "colab_type": "text"
      },
      "source": [
        "## 0. Setup"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NODTw5O0cP2j",
        "colab_type": "code",
        "outputId": "3fff9cbe-9dbb-47da-dc33-32ede9b5ca53",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "from fastai.text import *\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "pd.set_option('display.max_colwidth', -1)\n",
        "\n",
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_0FqsRgqWo2c",
        "colab_type": "text"
      },
      "source": [
        "## 1. Import the data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QtJrE3IAcsPM",
        "colab_type": "code",
        "outputId": "f4cfcdca-1203-493e-ed61-d70df64baada",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 215
        }
      },
      "source": [
        "text_path = 'gdrive/My Drive/Developer/Datasets/stackoverflow-dataset/title_StackOverflow.txt'\n",
        "label_path = 'gdrive/My Drive/Developer/Datasets/stackoverflow-dataset/label_StackOverflow.txt'\n",
        "\n",
        "df_text = pd.read_csv(text_path, sep='\\t', names=['text'], header=None)\n",
        "df_label = pd.read_csv(label_path, sep='\\t', names=['label'], header=None)\n",
        "\n",
        "df = pd.concat([df_label, df_text], axis=1, sort=False)\n",
        "print('Length of dataset: '+str(len(df.index)))\n",
        "df.head()"
      ],
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Length of dataset: 20000\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>label</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>18</td>\n",
              "      <td>How do I fill a DataSet or a DataTable from a LINQ query resultset ?</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>18</td>\n",
              "      <td>How do you page a collection with LINQ?</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>3</td>\n",
              "      <td>Best Subversion clients for Windows Vista (64bit)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>Best Practice: Collaborative Environment, Bin Directory, SVN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>7</td>\n",
              "      <td>Visual Studio Setup Project - Per User Registry Settings</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   label                                                                  text\n",
              "0  18     How do I fill a DataSet or a DataTable from a LINQ query resultset ?\n",
              "1  18     How do you page a collection with LINQ?                             \n",
              "2  3      Best Subversion clients for Windows Vista (64bit)                   \n",
              "3  3      Best Practice: Collaborative Environment, Bin Directory, SVN        \n",
              "4  7      Visual Studio Setup Project - Per User Registry Settings            "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZRjYAd0zATIu",
        "colab_type": "code",
        "outputId": "125837b4-96aa-4540-81fc-377e4b3e53e9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 198
        }
      },
      "source": [
        "mapping = {\n",
        "  1: 'wordpress',\n",
        "  2: 'oracle',\n",
        "  3: 'svn',\n",
        "  4: 'apache',\n",
        "  5: 'excel',\n",
        "  6: 'matlab',\n",
        "  7: 'visual-studio',\n",
        "  8: 'cocoa',\n",
        "  9: 'osx',\n",
        "  10: 'bash',\n",
        "  11: 'spring',\n",
        "  12: 'hibernate',\n",
        "  13: 'scala',\n",
        "  14: 'sharepoint',\n",
        "  15: 'ajax',\n",
        "  16: 'qt',\n",
        "  17: 'drupal',\n",
        "  18: 'linq',\n",
        "  19: 'haskell',\n",
        "  20: 'magento'\n",
        "}\n",
        "\n",
        "df['label'] = df['label'].map(mapping)\n",
        "\n",
        "df.head()"
      ],
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>label</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>linq</td>\n",
              "      <td>How do I fill a DataSet or a DataTable from a LINQ query resultset ?</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>linq</td>\n",
              "      <td>How do you page a collection with LINQ?</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>svn</td>\n",
              "      <td>Best Subversion clients for Windows Vista (64bit)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>svn</td>\n",
              "      <td>Best Practice: Collaborative Environment, Bin Directory, SVN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>visual-studio</td>\n",
              "      <td>Visual Studio Setup Project - Per User Registry Settings</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "           label                                                                  text\n",
              "0  linq           How do I fill a DataSet or a DataTable from a LINQ query resultset ?\n",
              "1  linq           How do you page a collection with LINQ?                             \n",
              "2  svn            Best Subversion clients for Windows Vista (64bit)                   \n",
              "3  svn            Best Practice: Collaborative Environment, Bin Directory, SVN        \n",
              "4  visual-studio  Visual Studio Setup Project - Per User Registry Settings            "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 37
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7SSQyf-RWwnG",
        "colab_type": "text"
      },
      "source": [
        "## 2. Create train & validation datasets and FastAI data bunch"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "86impBLW82I9",
        "colab_type": "code",
        "outputId": "901fe1c3-4898-46a0-964f-12419e9af846",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "df_trn, df_val = train_test_split(df, stratify = df['label'], test_size = 0.3)\n",
        "\n",
        "df_trn.shape, df_val.shape"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((14000, 2), (6000, 2))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yx5Z7N6v85y8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Language model data\n",
        "data_lm = TextLMDataBunch.from_df(train_df = df_trn, valid_df = df_val, path = \"\")\n",
        "\n",
        "# Classifier model data\n",
        "data_clas = TextClasDataBunch.from_df(train_df = df_trn, valid_df = df_val, path = \"\", vocab=data_lm.train_ds.vocab, bs=32)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YcbffU6e88k5",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 233
        },
        "outputId": "03cc8311-3ab4-4334-f170-96e040cf766f"
      },
      "source": [
        "data_clas.show_batch()"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th>text</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <td>xxbos xxmaj why do i receive a \" xxmaj out of xxmaj windows xxmaj resources \" warning when i open numerous figure windows in xxup matlab [ 7.0 ( xxup xxunk ) and beyond ] on a xxmaj microsoft xxmaj windows xxup pc ?</td>\n",
              "      <td>matlab</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>xxbos xxmaj remove xxmaj single xxmaj quotes xxmaj from xxmaj all xxmaj cells in a datatable - xxmaj creating xxmaj new xxmaj table - xxmaj using xxup linq in xxmaj vb . xxmaj net</td>\n",
              "      <td>linq</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>xxbos xxmaj what 's the best way to find if the processor type is xxup ppc or xxmaj intel in xxmaj cocoa ? ( xxmaj do i have to use xxmaj carbon ? )</td>\n",
              "      <td>osx</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>xxbos xxmaj in xxmaj oracle xxmaj xxunk ( xxup xxunk ) , is there a way to get the xxunk point of a polygon that is contained by the surface of the polygon ?</td>\n",
              "      <td>oracle</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>xxbos xxmaj configuring xxmaj apache to route \" xxup get / user / foo \" to / user / xxunk and \" xxup put / user / foo \" to / user / xxunk</td>\n",
              "      <td>apache</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Jo1xRTOP9J_2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "data_lm.save('data_lm_export.pkl')\n",
        "data_clas.save('data_clas_export.pkl')\n",
        "data_lm = load_data('./', 'data_lm_export.pkl')\n",
        "data_clas = load_data('./', 'data_clas_export.pkl', bs=64)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5UsS_KdIW6DS",
        "colab_type": "text"
      },
      "source": [
        "## 3. Create and Train the Language Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l-IrOyNv9UIL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "learn = language_model_learner(data_lm, AWD_LSTM, drop_mult=0.3)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vmCBf6aQ9X0J",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 335
        },
        "outputId": "d681b54d-e4f6-4463-8d65-6fcf7e5a7e1c"
      },
      "source": [
        "learn.lr_find()\n",
        "learn.recorder.plot(suggestion=True)\n",
        "min_grad_lr = learn.recorder.min_grad_lr"
      ],
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "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",
            "Min numerical gradient: 3.31E-02\n",
            "Min loss divided by 10: 7.59E-02\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nXuZxgM0AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dQoyexxK9bV5",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 108
        },
        "outputId": "97ab094f-9dde-43bc-cc0d-2b42fc817a66"
      },
      "source": [
        "learn.fit_one_cycle(2, min_grad_lr)"
      ],
      "execution_count": 44,
      "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>4.840412</td>\n",
              "      <td>4.063015</td>\n",
              "      <td>0.314076</td>\n",
              "      <td>00:14</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>4.135231</td>\n",
              "      <td>3.952360</td>\n",
              "      <td>0.322781</td>\n",
              "      <td>00:14</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sp0jzmcR9cF6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 108
        },
        "outputId": "7114486a-63df-4618-fcda-57cfcc6e4407"
      },
      "source": [
        "learn.unfreeze()\n",
        "learn.fit_one_cycle(2, 1e-3)"
      ],
      "execution_count": 45,
      "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>3.542882</td>\n",
              "      <td>3.825899</td>\n",
              "      <td>0.336423</td>\n",
              "      <td>00:18</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>3.368596</td>\n",
              "      <td>3.830364</td>\n",
              "      <td>0.338944</td>\n",
              "      <td>00:18</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h6mBSac69hDA",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "0891c408-b8fb-4f00-97ad-f7fbd1e36c38"
      },
      "source": [
        "learn.predict(\"How do\", n_words=10)"
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'How do i get status object record in my MS'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 46
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WgIcbM439kjI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "learn.save_encoder('ft_enc')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-qfpJf2fXAQY",
        "colab_type": "text"
      },
      "source": [
        "## 4. Using the Language Model to Train the Classifier"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-dJUmbri9n3v",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "learn = text_classifier_learner(data_clas, AWD_LSTM, drop_mult=0.5)\n",
        "learn.load_encoder('ft_enc')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jOBRWdea9s2L",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 335
        },
        "outputId": "94361726-d554-45ac-8cfd-d61871837360"
      },
      "source": [
        "learn.lr_find()\n",
        "learn.recorder.plot(suggestion=True)\n",
        "min_grad_lr = learn.recorder.min_grad_lr"
      ],
      "execution_count": 49,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              ""
            ],
            "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",
            "Min numerical gradient: 5.75E-02\n",
            "Min loss divided by 10: 1.74E-02\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIiKS8P8ByTHjC7/BReEAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3H7e89ab9u9K",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 108
        },
        "outputId": "bb87c897-b0f4-4930-98d2-2462d790d0a9"
      },
      "source": [
        "learn.fit_one_cycle(2, min_grad_lr)"
      ],
      "execution_count": 50,
      "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>1.806202</td>\n",
              "      <td>1.342678</td>\n",
              "      <td>0.605000</td>\n",
              "      <td>00:10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>1.391742</td>\n",
              "      <td>1.032389</td>\n",
              "      <td>0.720333</td>\n",
              "      <td>00:10</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U3AjT6C89yCU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "outputId": "c055a98b-b429-4cc5-c2f0-3205dbdcf318"
      },
      "source": [
        "learn.recorder.plot_losses()"
      ],
      "execution_count": 51,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cXzFnXvR91b8",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 168
        },
        "outputId": "8082e22d-b588-4f48-9af1-92e360a877dd"
      },
      "source": [
        "learn.freeze_to(-2)\n",
        "learn.fit_one_cycle(4, slice(5e-3, 2e-3), moms=(0.8,0.7))"
      ],
      "execution_count": 52,
      "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>1.309619</td>\n",
              "      <td>0.798372</td>\n",
              "      <td>0.779833</td>\n",
              "      <td>00:11</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>1.057222</td>\n",
              "      <td>0.642959</td>\n",
              "      <td>0.819667</td>\n",
              "      <td>00:12</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.841182</td>\n",
              "      <td>0.595715</td>\n",
              "      <td>0.832667</td>\n",
              "      <td>00:12</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>3</td>\n",
              "      <td>0.754000</td>\n",
              "      <td>0.586629</td>\n",
              "      <td>0.832667</td>\n",
              "      <td>00:12</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "17tFKFBW95md",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 284
        },
        "outputId": "d7f67ada-be2a-4057-8b62-fa23c48a04c4"
      },
      "source": [
        "learn.recorder.plot_losses()"
      ],
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b5QHV_gO98Ze",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 168
        },
        "outputId": "136fad32-0a8a-4ed1-eb9d-984078a4a822"
      },
      "source": [
        "learn.unfreeze()\n",
        "learn.fit_one_cycle(4, slice(2e-3/100, 2e-3), moms=(0.8,0.7))"
      ],
      "execution_count": 54,
      "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>0.653609</td>\n",
              "      <td>0.558160</td>\n",
              "      <td>0.838167</td>\n",
              "      <td>00:23</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>1</td>\n",
              "      <td>0.582657</td>\n",
              "      <td>0.532493</td>\n",
              "      <td>0.846000</td>\n",
              "      <td>00:22</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>2</td>\n",
              "      <td>0.516326</td>\n",
              "      <td>0.517132</td>\n",
              "      <td>0.851333</td>\n",
              "      <td>00:25</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <td>3</td>\n",
              "      <td>0.472621</td>\n",
              "      <td>0.516339</td>\n",
              "      <td>0.852833</td>\n",
              "      <td>00:26</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XUf_GA3bXIEn",
        "colab_type": "text"
      },
      "source": [
        "## 5. Analyzing our results"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "INCp-Ozi-BfK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 311
        },
        "outputId": "51b36540-93e7-4bd3-df51-04178b06ee3f"
      },
      "source": [
        "preds,y,losses = learn.get_preds(with_loss=True)\n",
        "interp = ClassificationInterpretation(learn, preds, y, losses)\n",
        "interp.plot_confusion_matrix()"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAARsAAAEmCAYAAABSyTRSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Etgy39mH-ED-",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "85ef628a-f24d-4c9a-aed5-387fc3d693de"
      },
      "source": [
        "interp.most_confused()"
      ],
      "execution_count": 56,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('osx', 'cocoa', 41),\n",
              " ('qt', 'cocoa', 30),\n",
              " ('cocoa', 'osx', 25),\n",
              " ('scala', 'haskell', 16),\n",
              " ('cocoa', 'qt', 11),\n",
              " ('osx', 'qt', 11),\n",
              " ('spring', 'hibernate', 11),\n",
              " ('ajax', 'apache', 10),\n",
              " ('ajax', 'cocoa', 10),\n",
              " ('bash', 'osx', 10),\n",
              " ('svn', 'visual-studio', 10),\n",
              " ('hibernate', 'spring', 9),\n",
              " ('matlab', 'bash', 9),\n",
              " ('osx', 'svn', 9),\n",
              " ('sharepoint', 'cocoa', 9),\n",
              " ('bash', 'cocoa', 8),\n",
              " ('cocoa', 'bash', 8),\n",
              " ('hibernate', 'oracle', 8),\n",
              " ('scala', 'qt', 8),\n",
              " ('sharepoint', 'visual-studio', 8),\n",
              " ('spring', 'cocoa', 8),\n",
              " ('apache', 'ajax', 7),\n",
              " ('bash', 'qt', 7),\n",
              " ('cocoa', 'oracle', 7),\n",
              " ('excel', 'cocoa', 7),\n",
              " ('osx', 'visual-studio', 7),\n",
              " ('spring', 'qt', 7),\n",
              " ('ajax', 'spring', 6),\n",
              " ('apache', 'osx', 6),\n",
              " ('bash', 'apache', 6),\n",
              " ('drupal', 'magento', 6),\n",
              " ('drupal', 'sharepoint', 6),\n",
              " ('haskell', 'cocoa', 6),\n",
              " ('hibernate', 'cocoa', 6),\n",
              " ('hibernate', 'linq', 6),\n",
              " ('matlab', 'cocoa', 6),\n",
              " ('osx', 'ajax', 6),\n",
              " ('osx', 'bash', 6),\n",
              " ('qt', 'bash', 6),\n",
              " ('qt', 'haskell', 6),\n",
              " ('sharepoint', 'qt', 6),\n",
              " ('svn', 'cocoa', 6),\n",
              " ('visual-studio', 'cocoa', 6),\n",
              " ('ajax', 'oracle', 5),\n",
              " ('apache', 'cocoa', 5),\n",
              " ('apache', 'spring', 5),\n",
              " ('apache', 'svn', 5),\n",
              " ('drupal', 'cocoa', 5),\n",
              " ('haskell', 'scala', 5),\n",
              " ('linq', 'cocoa', 5),\n",
              " ('linq', 'visual-studio', 5),\n",
              " ('osx', 'apache', 5),\n",
              " ('qt', 'ajax', 5),\n",
              " ('qt', 'osx', 5),\n",
              " ('spring', 'oracle', 5),\n",
              " ('spring', 'scala', 5),\n",
              " ('wordpress', 'ajax', 5),\n",
              " ('ajax', 'bash', 4),\n",
              " ('apache', 'bash', 4),\n",
              " ('bash', 'ajax', 4),\n",
              " ('bash', 'haskell', 4),\n",
              " ('cocoa', 'matlab', 4),\n",
              " ('drupal', 'apache', 4),\n",
              " ('drupal', 'wordpress', 4),\n",
              " ('excel', 'matlab', 4),\n",
              " ('haskell', 'oracle', 4),\n",
              " ('linq', 'excel', 4),\n",
              " ('linq', 'hibernate', 4),\n",
              " ('magento', 'cocoa', 4),\n",
              " ('qt', 'spring', 4),\n",
              " ('qt', 'wordpress', 4),\n",
              " ('sharepoint', 'excel', 4),\n",
              " ('spring', 'ajax', 4),\n",
              " ('svn', 'bash', 4),\n",
              " ('visual-studio', 'ajax', 4),\n",
              " ('visual-studio', 'excel', 4),\n",
              " ('visual-studio', 'oracle', 4),\n",
              " ('visual-studio', 'qt', 4),\n",
              " ('visual-studio', 'sharepoint', 4),\n",
              " ('wordpress', 'apache', 4),\n",
              " ('ajax', 'qt', 3),\n",
              " ('ajax', 'sharepoint', 3),\n",
              " ('bash', 'svn', 3),\n",
              " ('cocoa', 'drupal', 3),\n",
              " ('cocoa', 'haskell', 3),\n",
              " ('cocoa', 'linq', 3),\n",
              " ('cocoa', 'scala', 3),\n",
              " ('cocoa', 'sharepoint', 3),\n",
              " ('cocoa', 'spring', 3),\n",
              " ('cocoa', 'visual-studio', 3),\n",
              " ('drupal', 'ajax', 3),\n",
              " ('drupal', 'bash', 3),\n",
              " ('excel', 'sharepoint', 3),\n",
              " ('excel', 'visual-studio', 3),\n",
              " ('haskell', 'osx', 3),\n",
              " ('haskell', 'qt', 3),\n",
              " ('haskell', 'spring', 3),\n",
              " ('hibernate', 'scala', 3),\n",
              " ('linq', 'oracle', 3),\n",
              " ('linq', 'scala', 3),\n",
              " ('linq', 'spring', 3),\n",
              " ('magento', 'wordpress', 3),\n",
              " ('oracle', 'ajax', 3),\n",
              " ('oracle', 'bash', 3),\n",
              " ('oracle', 'cocoa', 3),\n",
              " ('oracle', 'excel', 3),\n",
              " ('oracle', 'hibernate', 3),\n",
              " ('oracle', 'linq', 3),\n",
              " ('osx', 'matlab', 3),\n",
              " ('qt', 'visual-studio', 3),\n",
              " ('scala', 'cocoa', 3),\n",
              " ('sharepoint', 'spring', 3),\n",
              " ('svn', 'apache', 3),\n",
              " ('svn', 'qt', 3),\n",
              " ('svn', 'spring', 3),\n",
              " ('visual-studio', 'bash', 3),\n",
              " ('wordpress', 'cocoa', 3),\n",
              " ('wordpress', 'drupal', 3),\n",
              " ('wordpress', 'qt', 3),\n",
              " ('wordpress', 'sharepoint', 3),\n",
              " ('ajax', 'drupal', 2),\n",
              " ('ajax', 'wordpress', 2),\n",
              " ('apache', 'wordpress', 2),\n",
              " ('bash', 'visual-studio', 2),\n",
              " ('cocoa', 'ajax', 2),\n",
              " ('cocoa', 'magento', 2),\n",
              " ('cocoa', 'wordpress', 2),\n",
              " ('drupal', 'oracle', 2),\n",
              " ('excel', 'bash', 2),\n",
              " ('excel', 'osx', 2),\n",
              " ('excel', 'spring', 2),\n",
              " ('haskell', 'bash', 2),\n",
              " ('haskell', 'linq', 2),\n",
              " ('haskell', 'visual-studio', 2),\n",
              " ('hibernate', 'haskell', 2),\n",
              " ('linq', 'bash', 2),\n",
              " ('linq', 'haskell', 2),\n",
              " ('linq', 'matlab', 2),\n",
              " ('linq', 'sharepoint', 2),\n",
              " ('linq', 'svn', 2),\n",
              " ('magento', 'apache', 2),\n",
              " ('magento', 'osx', 2),\n",
              " ('magento', 'sharepoint', 2),\n",
              " ('matlab', 'excel', 2),\n",
              " ('matlab', 'haskell', 2),\n",
              " ('matlab', 'magento', 2),\n",
              " ('matlab', 'osx', 2),\n",
              " ('matlab', 'qt', 2),\n",
              " ('matlab', 'spring', 2),\n",
              " ('oracle', 'matlab', 2),\n",
              " ('oracle', 'sharepoint', 2),\n",
              " ('oracle', 'svn', 2),\n",
              " ('oracle', 'visual-studio', 2),\n",
              " ('osx', 'drupal', 2),\n",
              " ('osx', 'haskell', 2),\n",
              " ('osx', 'oracle', 2),\n",
              " ('qt', 'scala', 2),\n",
              " ('scala', 'ajax', 2),\n",
              " ('scala', 'linq', 2),\n",
              " ('scala', 'matlab', 2),\n",
              " ('scala', 'oracle', 2),\n",
              " ('scala', 'spring', 2),\n",
              " ('scala', 'visual-studio', 2),\n",
              " ('sharepoint', 'drupal', 2),\n",
              " ('sharepoint', 'linq', 2),\n",
              " ('spring', 'visual-studio', 2),\n",
              " ('svn', 'haskell', 2),\n",
              " ('svn', 'magento', 2),\n",
              " ('svn', 'osx', 2),\n",
              " ('svn', 'sharepoint', 2),\n",
              " ('visual-studio', 'haskell', 2),\n",
              " ('visual-studio', 'magento', 2),\n",
              " ('visual-studio', 'svn', 2),\n",
              " ('wordpress', 'magento', 2),\n",
              " ('wordpress', 'spring', 2),\n",
              " ('ajax', 'haskell', 1),\n",
              " ('ajax', 'linq', 1),\n",
              " ('ajax', 'magento', 1),\n",
              " ('ajax', 'osx', 1),\n",
              " ('ajax', 'visual-studio', 1),\n",
              " ('apache', 'drupal', 1),\n",
              " ('apache', 'haskell', 1),\n",
              " ('apache', 'magento', 1),\n",
              " ('apache', 'matlab', 1),\n",
              " ('apache', 'oracle', 1),\n",
              " ('apache', 'qt', 1),\n",
              " ('bash', 'hibernate', 1),\n",
              " ('bash', 'linq', 1),\n",
              " ('bash', 'matlab', 1),\n",
              " ('bash', 'oracle', 1),\n",
              " ('bash', 'sharepoint', 1),\n",
              " ('bash', 'spring', 1),\n",
              " ('bash', 'wordpress', 1),\n",
              " ('cocoa', 'apache', 1),\n",
              " ('cocoa', 'svn', 1),\n",
              " ('drupal', 'linq', 1),\n",
              " ('drupal', 'qt', 1),\n",
              " ('drupal', 'spring', 1),\n",
              " ('excel', 'ajax', 1),\n",
              " ('excel', 'apache', 1),\n",
              " ('excel', 'linq', 1),\n",
              " ('excel', 'magento', 1),\n",
              " ('excel', 'qt', 1),\n",
              " ('haskell', 'apache', 1),\n",
              " ('haskell', 'magento', 1),\n",
              " ('haskell', 'matlab', 1),\n",
              " ('haskell', 'svn', 1),\n",
              " ('hibernate', 'ajax', 1),\n",
              " ('hibernate', 'apache', 1),\n",
              " ('hibernate', 'bash', 1),\n",
              " ('hibernate', 'drupal', 1),\n",
              " ('hibernate', 'excel', 1),\n",
              " ('linq', 'ajax', 1),\n",
              " ('linq', 'osx', 1),\n",
              " ('linq', 'qt', 1),\n",
              " ('magento', 'ajax', 1),\n",
              " ('magento', 'bash', 1),\n",
              " ('magento', 'matlab', 1),\n",
              " ('magento', 'oracle', 1),\n",
              " ('magento', 'qt', 1),\n",
              " ('magento', 'spring', 1),\n",
              " ('magento', 'visual-studio', 1),\n",
              " ('matlab', 'sharepoint', 1),\n",
              " ('oracle', 'apache', 1),\n",
              " ('oracle', 'drupal', 1),\n",
              " ('oracle', 'magento', 1),\n",
              " ('oracle', 'osx', 1),\n",
              " ('oracle', 'scala', 1),\n",
              " ('oracle', 'spring', 1),\n",
              " ('osx', 'excel', 1),\n",
              " ('osx', 'magento', 1),\n",
              " ('osx', 'sharepoint', 1),\n",
              " ('osx', 'spring', 1),\n",
              " ('qt', 'apache', 1),\n",
              " ('qt', 'drupal', 1),\n",
              " ('qt', 'excel', 1),\n",
              " ('qt', 'matlab', 1),\n",
              " ('qt', 'oracle', 1),\n",
              " ('qt', 'sharepoint', 1),\n",
              " ('scala', 'hibernate', 1),\n",
              " ('scala', 'svn', 1),\n",
              " ('sharepoint', 'apache', 1),\n",
              " ('sharepoint', 'haskell', 1),\n",
              " ('sharepoint', 'svn', 1),\n",
              " ('sharepoint', 'wordpress', 1),\n",
              " ('spring', 'apache', 1),\n",
              " ('spring', 'bash', 1),\n",
              " ('spring', 'drupal', 1),\n",
              " ('spring', 'osx', 1),\n",
              " ('spring', 'wordpress', 1),\n",
              " ('svn', 'ajax', 1),\n",
              " ('svn', 'excel', 1),\n",
              " ('svn', 'scala', 1),\n",
              " ('visual-studio', 'apache', 1),\n",
              " ('visual-studio', 'linq', 1),\n",
              " ('visual-studio', 'matlab', 1),\n",
              " ('wordpress', 'bash', 1),\n",
              " ('wordpress', 'excel', 1),\n",
              " ('wordpress', 'visual-studio', 1)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 56
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0otTxIuBXOvi",
        "colab_type": "text"
      },
      "source": [
        "## 6. Predictions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yGtVxuVh-JFT",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 104
        },
        "outputId": "a1cb29c8-ef9f-4087-e096-c5e7eaecdc83"
      },
      "source": [
        "learn.predict(\"homebrew not working\")"
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(Category cocoa,\n",
              " tensor(3),\n",
              " tensor([0.1329, 0.0255, 0.0075, 0.2108, 0.0212, 0.0048, 0.0121, 0.0144, 0.0074,\n",
              "         0.1576, 0.0061, 0.0078, 0.0666, 0.1396, 0.0068, 0.0484, 0.0539, 0.0068,\n",
              "         0.0418, 0.0280]))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 57
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aeRURqcqXUJY",
        "colab_type": "text"
      },
      "source": [
        "## 7. Export the model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EFIPJelMih7z",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "learn.export()"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}