notebooks/4.2_transformer_lastname.ipynb

{"cells":[{"cell_type":"code","execution_count":6,"metadata":{"executionInfo":{"elapsed":1205,"status":"ok","timestamp":1687907280818,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"6wTqaxXwRGBE"},"outputs":[],"source":["import string\n","\n","import torch\n","import torch.nn as nn\n","from torch.utils.data import Dataset, DataLoader\n","from torch.optim.adamw import AdamW\n","\n","from tqdm import tqdm\n","tqdm.pandas()\n","\n","import pandas as pd\n","import numpy as np\n","\n","from fastprogress import master_bar, progress_bar\n","\n","from sklearn.feature_extraction.text import CountVectorizer\n"]},{"cell_type":"code","source":["from google.colab import drive\n","drive.mount('/content/drive')"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"R9TvbZ-BEmX9","executionInfo":{"status":"ok","timestamp":1687906610403,"user_tz":420,"elapsed":25588,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}},"outputId":"47ba728f-537b-4c28-96f2-154b92ca70c1"},"execution_count":2,"outputs":[{"output_type":"stream","name":"stdout","text":["Mounted at /content/drive\n"]}]},{"cell_type":"markdown","metadata":{"id":"cSlSeNxpELos"},"source":["\n","\n","## load data"]},{"cell_type":"code","execution_count":3,"metadata":{"executionInfo":{"elapsed":7018,"status":"ok","timestamp":1687907223708,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"VVVfrmyYbL0-"},"outputs":[],"source":["train_df = pd.read_csv(\"/content/drive/MyDrive/Colab/ethnicolor/data/fl_2022_LastName_train.csv.gz\")\n","val_df = pd.read_csv(\"/content/drive/MyDrive/Colab/ethnicolor/data/fl_2022_LastName_val.csv.gz\")\n","test_df = pd.read_csv(\"/content/drive/MyDrive/Colab/ethnicolor/data/fl_2022_LastName_test.csv.gz\")"]},{"cell_type":"code","execution_count":4,"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":206},"executionInfo":{"elapsed":10,"status":"ok","timestamp":1687907223708,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"u0TJpyUQEAHz","outputId":"456c91ef-fab2-4ad6-e47b-96d8982204ac"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["           name_last  asian  hispanic  nh_black  nh_white     other  total_n  \\\n","0  Neaman-Piotrowicz    0.0       0.0       0.0  1.000000  0.000000      1.0   \n","1              Chaet    0.0       0.0       0.0  1.000000  0.000000     13.0   \n","2        Veloz Perez    0.0       1.0       0.0  0.000000  0.000000      1.0   \n","3            Spalter    0.0       0.0       0.0  0.928571  0.071429     42.0   \n","4     Kearns-Edwards    0.0       0.0       0.0  1.000000  0.000000      1.0   \n","\n","       race  race_code  \n","0  nh_white          3  \n","1  nh_white          3  \n","2  hispanic          1  \n","3  nh_white          3  \n","4  nh_white          3  "],"text/html":["\n","  <div id=\"df-d8280fab-8370-44f2-bf48-cc8ac153ed49\">\n","    <div class=\"colab-df-container\">\n","      <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>name_last</th>\n","      <th>asian</th>\n","      <th>hispanic</th>\n","      <th>nh_black</th>\n","      <th>nh_white</th>\n","      <th>other</th>\n","      <th>total_n</th>\n","      <th>race</th>\n","      <th>race_code</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>Neaman-Piotrowicz</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>1.000000</td>\n","      <td>0.000000</td>\n","      <td>1.0</td>\n","      <td>nh_white</td>\n","      <td>3</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>Chaet</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>1.000000</td>\n","      <td>0.000000</td>\n","      <td>13.0</td>\n","      <td>nh_white</td>\n","      <td>3</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>Veloz Perez</td>\n","      <td>0.0</td>\n","      <td>1.0</td>\n","      <td>0.0</td>\n","      <td>0.000000</td>\n","      <td>0.000000</td>\n","      <td>1.0</td>\n","      <td>hispanic</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>Spalter</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.928571</td>\n","      <td>0.071429</td>\n","      <td>42.0</td>\n","      <td>nh_white</td>\n","      <td>3</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>Kearns-Edwards</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>1.000000</td>\n","      <td>0.000000</td>\n","      <td>1.0</td>\n","      <td>nh_white</td>\n","      <td>3</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>\n","      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-d8280fab-8370-44f2-bf48-cc8ac153ed49')\"\n","              title=\"Convert this dataframe to an interactive table.\"\n","              style=\"display:none;\">\n","        \n","  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n","       width=\"24px\">\n","    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n","    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n","  </svg>\n","      </button>\n","      \n","  <style>\n","    .colab-df-container {\n","      display:flex;\n","      flex-wrap:wrap;\n","      gap: 12px;\n","    }\n","\n","    .colab-df-convert {\n","      background-color: #E8F0FE;\n","      border: none;\n","      border-radius: 50%;\n","      cursor: pointer;\n","      display: none;\n","      fill: #1967D2;\n","      height: 32px;\n","      padding: 0 0 0 0;\n","      width: 32px;\n","    }\n","\n","    .colab-df-convert:hover {\n","      background-color: #E2EBFA;\n","      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n","      fill: #174EA6;\n","    }\n","\n","    [theme=dark] .colab-df-convert {\n","      background-color: #3B4455;\n","      fill: #D2E3FC;\n","    }\n","\n","    [theme=dark] .colab-df-convert:hover {\n","      background-color: #434B5C;\n","      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n","      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n","      fill: #FFFFFF;\n","    }\n","  </style>\n","\n","      <script>\n","        const buttonEl =\n","          document.querySelector('#df-d8280fab-8370-44f2-bf48-cc8ac153ed49 button.colab-df-convert');\n","        buttonEl.style.display =\n","          google.colab.kernel.accessAllowed ? 'block' : 'none';\n","\n","        async function convertToInteractive(key) {\n","          const element = document.querySelector('#df-d8280fab-8370-44f2-bf48-cc8ac153ed49');\n","          const dataTable =\n","            await google.colab.kernel.invokeFunction('convertToInteractive',\n","                                                     [key], {});\n","          if (!dataTable) return;\n","\n","          const docLinkHtml = 'Like what you see? Visit the ' +\n","            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n","            + ' to learn more about interactive tables.';\n","          element.innerHTML = '';\n","          dataTable['output_type'] = 'display_data';\n","          await google.colab.output.renderOutput(dataTable, element);\n","          const docLink = document.createElement('div');\n","          docLink.innerHTML = docLinkHtml;\n","          element.appendChild(docLink);\n","        }\n","      </script>\n","    </div>\n","  </div>\n","  "]},"metadata":{},"execution_count":4}],"source":["train_df.head()"]},{"cell_type":"code","source":["print(train_df.shape)\n","print(val_df.shape)\n","print(test_df.shape)"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"QLsK0nn7HI2I","executionInfo":{"status":"ok","timestamp":1687907244859,"user_tz":420,"elapsed":5,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}},"outputId":"b250a616-2a8a-45f2-8b7a-39c547ec56a6"},"execution_count":5,"outputs":[{"output_type":"stream","name":"stdout","text":["(845316, 9)\n","(105664, 9)\n","(105665, 9)\n"]}]},{"cell_type":"code","source":["train_df.head()"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":206},"id":"xdILqu7bHexA","executionInfo":{"status":"ok","timestamp":1687907340207,"user_tz":420,"elapsed":7,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}},"outputId":"ca0bab2a-f30a-4d21-ec10-90d211661d62"},"execution_count":8,"outputs":[{"output_type":"execute_result","data":{"text/plain":["           name_last  asian  hispanic  nh_black  nh_white     other  total_n  \\\n","0  Neaman-Piotrowicz    0.0       0.0       0.0  1.000000  0.000000      1.0   \n","1              Chaet    0.0       0.0       0.0  1.000000  0.000000     13.0   \n","2        Veloz Perez    0.0       1.0       0.0  0.000000  0.000000      1.0   \n","3            Spalter    0.0       0.0       0.0  0.928571  0.071429     42.0   \n","4     Kearns-Edwards    0.0       0.0       0.0  1.000000  0.000000      1.0   \n","\n","       race  race_code  \n","0  nh_white          3  \n","1  nh_white          3  \n","2  hispanic          1  \n","3  nh_white          3  \n","4  nh_white          3  "],"text/html":["\n","  <div id=\"df-0302f208-8140-4103-84b0-fed47c778258\">\n","    <div class=\"colab-df-container\">\n","      <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>name_last</th>\n","      <th>asian</th>\n","      <th>hispanic</th>\n","      <th>nh_black</th>\n","      <th>nh_white</th>\n","      <th>other</th>\n","      <th>total_n</th>\n","      <th>race</th>\n","      <th>race_code</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>Neaman-Piotrowicz</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>1.000000</td>\n","      <td>0.000000</td>\n","      <td>1.0</td>\n","      <td>nh_white</td>\n","      <td>3</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>Chaet</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>1.000000</td>\n","      <td>0.000000</td>\n","      <td>13.0</td>\n","      <td>nh_white</td>\n","      <td>3</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>Veloz Perez</td>\n","      <td>0.0</td>\n","      <td>1.0</td>\n","      <td>0.0</td>\n","      <td>0.000000</td>\n","      <td>0.000000</td>\n","      <td>1.0</td>\n","      <td>hispanic</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>Spalter</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.928571</td>\n","      <td>0.071429</td>\n","      <td>42.0</td>\n","      <td>nh_white</td>\n","      <td>3</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>Kearns-Edwards</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>1.000000</td>\n","      <td>0.000000</td>\n","      <td>1.0</td>\n","      <td>nh_white</td>\n","      <td>3</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>\n","      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-0302f208-8140-4103-84b0-fed47c778258')\"\n","              title=\"Convert this dataframe to an interactive table.\"\n","              style=\"display:none;\">\n","        \n","  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n","       width=\"24px\">\n","    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n","    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n","  </svg>\n","      </button>\n","      \n","  <style>\n","    .colab-df-container {\n","      display:flex;\n","      flex-wrap:wrap;\n","      gap: 12px;\n","    }\n","\n","    .colab-df-convert {\n","      background-color: #E8F0FE;\n","      border: none;\n","      border-radius: 50%;\n","      cursor: pointer;\n","      display: none;\n","      fill: #1967D2;\n","      height: 32px;\n","      padding: 0 0 0 0;\n","      width: 32px;\n","    }\n","\n","    .colab-df-convert:hover {\n","      background-color: #E2EBFA;\n","      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n","      fill: #174EA6;\n","    }\n","\n","    [theme=dark] .colab-df-convert {\n","      background-color: #3B4455;\n","      fill: #D2E3FC;\n","    }\n","\n","    [theme=dark] .colab-df-convert:hover {\n","      background-color: #434B5C;\n","      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n","      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n","      fill: #FFFFFF;\n","    }\n","  </style>\n","\n","      <script>\n","        const buttonEl =\n","          document.querySelector('#df-0302f208-8140-4103-84b0-fed47c778258 button.colab-df-convert');\n","        buttonEl.style.display =\n","          google.colab.kernel.accessAllowed ? 'block' : 'none';\n","\n","        async function convertToInteractive(key) {\n","          const element = document.querySelector('#df-0302f208-8140-4103-84b0-fed47c778258');\n","          const dataTable =\n","            await google.colab.kernel.invokeFunction('convertToInteractive',\n","                                                     [key], {});\n","          if (!dataTable) return;\n","\n","          const docLinkHtml = 'Like what you see? Visit the ' +\n","            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n","            + ' to learn more about interactive tables.';\n","          element.innerHTML = '';\n","          dataTable['output_type'] = 'display_data';\n","          await google.colab.output.renderOutput(dataTable, element);\n","          const docLink = document.createElement('div');\n","          docLink.innerHTML = docLinkHtml;\n","          element.appendChild(docLink);\n","        }\n","      </script>\n","    </div>\n","  </div>\n","  "]},"metadata":{},"execution_count":8}]},{"cell_type":"markdown","metadata":{"id":"GzHXp_2UE_H6"},"source":["## Creating Vocab"]},{"cell_type":"code","execution_count":7,"metadata":{"executionInfo":{"elapsed":4,"status":"ok","timestamp":1687907318286,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"eQma_l_7KbGw"},"outputs":[],"source":["vectorizer = CountVectorizer(analyzer='char', lowercase=False)"]},{"cell_type":"code","execution_count":9,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":3629,"status":"ok","timestamp":1687907355680,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"acgCGLmiKePO","outputId":"6ba5f88c-6190-4a59-d600-da91f7a60ae6"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["<845316x55 sparse matrix of type '<class 'numpy.int64'>'\n","\twith 7105885 stored elements in Compressed Sparse Row format>"]},"metadata":{},"execution_count":9}],"source":["vectorizer.fit_transform(train_df['name_last'])"]},{"cell_type":"code","execution_count":10,"metadata":{"executionInfo":{"elapsed":2,"status":"ok","timestamp":1687907367442,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"NRIPVUoDKg6I"},"outputs":[],"source":["vocab = list(vectorizer.get_feature_names_out())"]},{"cell_type":"code","execution_count":11,"metadata":{"executionInfo":{"elapsed":2,"status":"ok","timestamp":1687907373262,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"54yuEsreKkDP"},"outputs":[],"source":["n_letters = len(vocab)"]},{"cell_type":"markdown","source":["## Creating Dataset"],"metadata":{"id":"gp8leAoGHrxF"}},{"cell_type":"code","execution_count":12,"metadata":{"id":"ZJKzxCO0EN2x","executionInfo":{"status":"ok","timestamp":1687907414644,"user_tz":420,"elapsed":488,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}}},"outputs":[],"source":["race_id_df = train_df[['race', 'race_code']].drop_duplicates().sort_values('race_code')\n","race_to_id = dict(race_id_df.values)\n","id_to_race = dict(race_id_df[['race_code', 'race']].values)\n","all_categories = list(id_to_race.values())\n","n_categories = len(all_categories)\n","seq_len = train_df['name_last'].str.len().max()"]},{"cell_type":"code","execution_count":13,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"kMnbkP7wEN2y","executionInfo":{"status":"ok","timestamp":1687907427495,"user_tz":420,"elapsed":5,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}},"outputId":"97f1effb-5af7-48d7-b336-7907c085c64d"},"outputs":[{"output_type":"stream","name":"stdout","text":["classes - ['asian', 'hispanic', 'nh_black', 'nh_white', 'other']\n","no of classes - 5\n","max seq len - 30\n"]}],"source":["print(f\"classes - {all_categories}\")\n","print(f\"no of classes - {n_categories}\")\n","print(f\"max seq len - {seq_len}\")"]},{"cell_type":"code","execution_count":14,"metadata":{"id":"ZFyLwWFmEN2z","executionInfo":{"status":"ok","timestamp":1687907438751,"user_tz":420,"elapsed":5,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}}},"outputs":[],"source":["# helper methods used for transform in dataset\n","\n","all_letters = ''.join(vocab)\n","oob = n_letters + 1\n","\n","def letterToIndex(letter):\n","    return all_letters.find(letter)\n","\n","def lineToTensor(line):\n","    tensor = torch.ones(seq_len) * oob\n","    for li, letter in enumerate(line):\n","        tensor[li] = letterToIndex(letter)\n","    return tensor"]},{"cell_type":"code","execution_count":15,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":3,"status":"ok","timestamp":1687907478163,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"fpvaQT5nK_2y","outputId":"e7eb4551-fa95-4383-d0ee-30c5b071dc0c"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["tensor([12., 29., 47., 43., 42., 56., 56., 56., 56., 56., 56., 56., 56., 56.,\n","        56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56.,\n","        56., 56.])"]},"metadata":{},"execution_count":15}],"source":["lineToTensor('Jason')"]},{"cell_type":"code","execution_count":16,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":4,"status":"ok","timestamp":1687907480693,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"NPAYYcHAK59U","outputId":"25fdf2c6-caa1-468b-be26-d44700308596"},"outputs":[{"output_type":"stream","name":"stdout","text":["torch.Size([30])\n"]}],"source":["print(lineToTensor('Jason').size())"]},{"cell_type":"code","execution_count":17,"metadata":{"executionInfo":{"elapsed":2,"status":"ok","timestamp":1687907501562,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"kKwuNTwjEzkV"},"outputs":[],"source":["# A custom Dataset class must implement three functions: __init__, __len__, and __getitem__\n","\n","class EthniColorDataset(Dataset):\n","  def __init__(self, data_df, transform=None):\n","    self.df = data_df\n","    self.transform = transform\n","\n","  def __len__(self):\n","    return len(self.df)\n","\n","  def __getitem__(self, idx):\n","    if torch.is_tensor(idx):\n","      idx = idx.tolist()\n","    name = self.df.iloc[idx, train_df.columns.get_loc('name_last')]\n","    if self.transform:\n","      name = self.transform(name)\n","    label = self.df.iloc[idx, train_df.columns.get_loc('race_code')]\n","    #target = getTarget(label)\n","    target = torch.tensor(label, dtype=torch.int64)\n","    return name, target"]},{"cell_type":"code","execution_count":18,"metadata":{"executionInfo":{"elapsed":13,"status":"ok","timestamp":1687907505297,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"Ue_MhX-7QN9I"},"outputs":[],"source":["train_dataset = EthniColorDataset(train_df, lineToTensor)\n","val_dataset = EthniColorDataset(val_df, lineToTensor)\n","test_dataset = EthniColorDataset(test_df, lineToTensor)"]},{"cell_type":"code","execution_count":19,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":14,"status":"ok","timestamp":1687907505300,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"U9fp-gC9Mnft","outputId":"7819ed10-ce47-4835-b47b-507cf9ac3e6a"},"outputs":[{"output_type":"stream","name":"stdout","text":["0 tensor([16., 33., 29., 41., 29., 42.,  2., 18., 37., 43., 48., 46., 43., 51.,\n","        37., 31., 54., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56.,\n","        56., 56.]) tensor(3)\n","1 tensor([ 5., 36., 29., 33., 48., 56., 56., 56., 56., 56., 56., 56., 56., 56.,\n","        56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56.,\n","        56., 56.]) tensor(3)\n","2 tensor([24., 33., 40., 43., 54.,  0., 18., 33., 46., 33., 54., 56., 56., 56.,\n","        56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56., 56.,\n","        56., 56.]) tensor(1)\n"]}],"source":["for i in range(3):\n","  name, label = train_dataset[i]\n","  print(i, name, label)"]},{"cell_type":"code","execution_count":20,"metadata":{"executionInfo":{"elapsed":4,"status":"ok","timestamp":1687907530799,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"tGyzWNzgNzjc"},"outputs":[],"source":["import os\n","# The Dataset retrieves our dataset’s features and labels one sample at a time.\n","# While training a model, we typically want to pass samples in “minibatches”,\n","# reshuffle the data at every epoch to reduce model overfitting, and\n","# use Python’s multiprocessing to speed up data retrieval.\n","\n","batch_size=128\n","\n","train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True, drop_last=True, num_workers=os.cpu_count())\n","val_dataloader = DataLoader(val_dataset, batch_size=batch_size, shuffle=True, drop_last=True, num_workers=os.cpu_count())\n","test_dataloader = DataLoader(test_dataset, batch_size=batch_size, shuffle=True, drop_last=True, num_workers=os.cpu_count())"]},{"cell_type":"code","execution_count":21,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":609,"status":"ok","timestamp":1687907533125,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"C7c7-RtYlE3G","outputId":"a6d626a1-635f-4a47-8d8e-9fe05ff611ea"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["[tensor([[10., 37., 46.,  ..., 56., 56., 56.],\n","         [13., 43., 47.,  ..., 56., 56., 56.],\n","         [ 9., 49., 54.,  ..., 56., 56., 56.],\n","         ...,\n","         [28., 29., 53.,  ..., 56., 56., 56.],\n","         [ 6., 49.,  0.,  ..., 56., 56., 56.],\n","         [ 9., 49., 42.,  ..., 56., 56., 56.]]),\n"," tensor([1, 3, 2, 1, 3, 3, 3, 3, 4, 3, 3, 3, 1, 3, 3, 1, 1, 1, 2, 3, 1, 1, 1, 1,\n","         3, 2, 3, 3, 0, 1, 1, 3, 3, 4, 1, 3, 3, 1, 3, 2, 1, 1, 2, 2, 0, 3, 3, 3,\n","         3, 3, 3, 3, 4, 1, 3, 3, 0, 2, 3, 1, 2, 1, 4, 1, 2, 3, 1, 0, 3, 3, 3, 3,\n","         1, 2, 3, 0, 1, 3, 3, 3, 3, 1, 3, 1, 3, 3, 1, 3, 3, 3, 2, 1, 1, 3, 2, 3,\n","         1, 1, 3, 0, 1, 1, 1, 2, 1, 2, 0, 3, 1, 1, 1, 3, 3, 3, 3, 1, 4, 3, 3, 1,\n","         3, 1, 3, 3, 1, 1, 3, 3])]"]},"metadata":{},"execution_count":21}],"source":["next(iter(train_dataloader))"]},{"cell_type":"code","execution_count":22,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":1032,"status":"ok","timestamp":1687907538897,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"UAebM2_6kIr1","outputId":"79e5d41e-b98e-494a-90d1-7c62539faa11"},"outputs":[{"output_type":"stream","name":"stdout","text":["0 torch.Size([128, 30]) torch.Size([128])\n","1 torch.Size([128, 30]) torch.Size([128])\n","2 torch.Size([128, 30]) torch.Size([128])\n"]}],"source":["for i_batch, sample_batched in enumerate(train_dataloader):\n","    print(i_batch, sample_batched[0].size(), sample_batched[1].size())\n","    if i_batch == 2:\n","        break"]},{"cell_type":"code","execution_count":23,"metadata":{"executionInfo":{"elapsed":3,"status":"ok","timestamp":1687907547069,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"IUewq69bGlIR"},"outputs":[],"source":["import math\n","\n","class PositionalEncoding(nn.Module):\n","    def __init__(self, d_model, dropout=0.1, max_len=5000):\n","        super(PositionalEncoding, self).__init__()\n","        self.dropout = nn.Dropout(p=dropout)\n","\n","        pe = torch.zeros(max_len, d_model)\n","        position = torch.arange(0, max_len, dtype=torch.float).unsqueeze(1)\n","        div_term = torch.exp(torch.arange(0, d_model, 2).float() * -(math.log(10000.0) / d_model))\n","        pe[:, 0::2] = torch.sin(position * div_term)\n","        pe[:, 1::2] = torch.cos(position * div_term)\n","        pe = pe.unsqueeze(0).transpose(0, 1)\n","        self.register_buffer('pe', pe)\n","\n","    def forward(self, x):\n","        x = x + self.pe[:x.size(0), :]\n","        return self.dropout(x)\n"]},{"cell_type":"code","execution_count":29,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":4,"status":"ok","timestamp":1687908244255,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"5zvwGXG2M-Gu","outputId":"d06bca6f-2ed6-4717-e0cf-fd2443a5292d"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["Transformer(\n","  (embedding): Embedding(57, 256)\n","  (pos_encoder): PositionalEncoding(\n","    (dropout): Dropout(p=0.1, inplace=False)\n","  )\n","  (transformer_encoder): TransformerEncoder(\n","    (layers): ModuleList(\n","      (0-1): 2 x TransformerEncoderLayer(\n","        (self_attn): MultiheadAttention(\n","          (out_proj): NonDynamicallyQuantizableLinear(in_features=256, out_features=256, bias=True)\n","        )\n","        (linear1): Linear(in_features=256, out_features=2048, bias=True)\n","        (dropout): Dropout(p=0.1, inplace=False)\n","        (linear2): Linear(in_features=2048, out_features=256, bias=True)\n","        (norm1): LayerNorm((256,), eps=1e-05, elementwise_affine=True)\n","        (norm2): LayerNorm((256,), eps=1e-05, elementwise_affine=True)\n","        (dropout1): Dropout(p=0.1, inplace=False)\n","        (dropout2): Dropout(p=0.1, inplace=False)\n","      )\n","    )\n","  )\n","  (fc): Linear(in_features=256, out_features=5, bias=True)\n","  (softmax): LogSoftmax(dim=1)\n",")"]},"metadata":{},"execution_count":29}],"source":["# Set the random seed for reproducible results\n","torch.manual_seed(42)\n","device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n","\n","# Define the Transformer model\n","class Transformer(nn.Module):\n","    def __init__(self, input_size, hidden_size, output_size, nhead, num_layers=1):\n","        super(Transformer, self).__init__()\n","        self.hidden_size = hidden_size\n","\n","        self.embedding = nn.Embedding(input_size, hidden_size)\n","        self.pos_encoder = PositionalEncoding(hidden_size)\n","        encoder_layers = nn.TransformerEncoderLayer(hidden_size, nhead, batch_first=True)\n","        self.transformer_encoder = nn.TransformerEncoder(encoder_layers, num_layers)\n","        self.fc = nn.Linear(hidden_size, output_size)\n","        self.softmax = nn.LogSoftmax(dim=1)\n","\n","    def forward(self, input):\n","        embedded = self.embedding(input.type(torch.IntTensor).to(input.device))\n","        pos_encoded = self.pos_encoder(embedded)\n","        out = self.transformer_encoder(pos_encoded) # (batch, seq, feature)\n","        out = out[:, -1, :] # drop seq\n","        out = self.fc(out)\n","        out = self.softmax(out) # (batch, out_cats)\n","        return out\n","\n","n_hidden = 256 # 128\n","seq_len = seq_len\n","vocab_size = n_letters + 1 + 1 # vocab + oob + 1\n","\n","model = Transformer(vocab_size, n_hidden, n_categories, nhead=4, num_layers=2)\n","model.to(device)"]},{"cell_type":"code","execution_count":26,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":4285,"status":"ok","timestamp":1687908191463,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"G2v1iG1PTDsO","outputId":"cdd4dcf9-c420-4bb8-d5c5-9aa92d5d5b8b"},"outputs":[{"output_type":"stream","name":"stdout","text":["input shape :  torch.Size([30])\n","input shape w batch :  torch.Size([1, 30])\n","torch.Size([1, 5])\n","tensor([[-1.7602, -1.9035, -2.4558, -0.9444, -1.5886]], device='cuda:0',\n","       grad_fn=<LogSoftmaxBackward0>)\n","tensor(3, device='cuda:0')\n","model predicted w/o train - nh_white\n"]}],"source":["input = lineToTensor('Rogers Renee')\n","\n","print(\"input shape : \",input.shape)\n","print(\"input shape w batch : \", input.unsqueeze(0).shape)\n","#print(hidden.shape)\n","\n","# sending with batch 1\n","output = model(input.unsqueeze(0).to(device))\n","# print(output) - has 47x8 log values\n","print(output.shape)\n","print(output)\n","print(torch.argmax(output))\n","print(f\"model predicted w/o train - {all_categories[torch.argmax(output).item()]}\")"]},{"cell_type":"markdown","metadata":{"id":"_qsEnfPKgYgt"},"source":["## Training\n"]},{"cell_type":"code","execution_count":27,"metadata":{"id":"zRcu2U_PUKsP","executionInfo":{"status":"ok","timestamp":1687908205899,"user_tz":420,"elapsed":432,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}}},"outputs":[],"source":["class EarlyStopper:\n","    def __init__(self, patience=1, min_delta=0):\n","        self.patience = patience\n","        self.min_delta = min_delta\n","        self.counter = 0\n","        self.min_validation_loss = np.inf\n","\n","    def early_stop(self, validation_loss):\n","        if validation_loss < self.min_validation_loss:\n","            self.min_validation_loss = validation_loss\n","            self.counter = 0\n","        elif validation_loss > (self.min_validation_loss + self.min_delta):\n","            self.counter += 1\n","            if self.counter >= self.patience:\n","                return True\n","        return False"]},{"cell_type":"code","execution_count":31,"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"RUkgRkq-TuWB","outputId":"79082c0e-5599-4429-d22b-30ab725e042e","executionInfo":{"status":"ok","timestamp":1687915156403,"user_tz":420,"elapsed":6837050,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}}},"outputs":[{"output_type":"display_data","data":{"text/plain":["<IPython.core.display.HTML object>"],"text/html":["\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:not([value]), progress:not([value])::-webkit-progress-bar {\n","        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n","    }\n","    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n","        background: #F44336;\n","    }\n","</style>\n"]},"metadata":{}},{"output_type":"display_data","data":{"text/plain":["<IPython.core.display.HTML object>"],"text/html":["\n","Epoch 0: Training loss 0.972470 validation loss 0.911322 with lr 0.000500<p>\n","Epoch 1: Training loss 0.915568 validation loss 0.888493 with lr 0.000500<p>\n","Epoch 2: Training loss 0.895168 validation loss 0.866666 with lr 0.000500<p>\n","Epoch 3: Training loss 0.872330 validation loss 0.841949 with lr 0.000500<p>\n","Epoch 4: Training loss 0.854262 validation loss 0.829526 with lr 0.000500<p>\n","Epoch 5: Training loss 0.843302 validation loss 0.821801 with lr 0.000500<p>\n","Epoch 6: Training loss 0.834950 validation loss 0.812855 with lr 0.000500<p>\n","Epoch 7: Training loss 0.828024 validation loss 0.808916 with lr 0.000500<p>\n","Epoch 8: Training loss 0.823799 validation loss 0.805657 with lr 0.000500<p>\n","Epoch 9: Training loss 0.819669 validation loss 0.802810 with lr 0.000500<p>\n","Epoch 10: Training loss 0.817051 validation loss 0.801424 with lr 0.000500<p>\n","Epoch 11: Training loss 0.814956 validation loss 0.798371 with lr 0.000500<p>\n","Epoch 12: Training loss 0.813273 validation loss 0.799631 with lr 0.000500<p>\n","Epoch 13: Training loss 0.811256 validation loss 0.796599 with lr 0.000500<p>\n","Epoch 14: Training loss 0.809805 validation loss 0.795919 with lr 0.000500<p>\n","Epoch 15: Training loss 0.808765 validation loss 0.795227 with lr 0.000500<p>\n","Epoch 16: Training loss 0.808330 validation loss 0.794181 with lr 0.000500<p>\n","Epoch 17: Training loss 0.806994 validation loss 0.795025 with lr 0.000500<p>\n","Epoch 18: Training loss 0.806395 validation loss 0.793764 with lr 0.000500<p>\n","Epoch 19: Training loss 0.805544 validation loss 0.793158 with lr 0.000500<p>\n","Epoch 20: Training loss 0.805383 validation loss 0.792545 with lr 0.000500<p>\n","Epoch 21: Training loss 0.804947 validation loss 0.792656 with lr 0.000500<p>\n","Epoch 22: Training loss 0.804385 validation loss 0.792179 with lr 0.000500<p>\n","Epoch 23: Training loss 0.804427 validation loss 0.792198 with lr 0.000500<p>\n","Epoch 24: Training loss 0.804183 validation loss 0.792826 with lr 0.000500<p>\n","Epoch 25: Training loss 0.804154 validation loss 0.791984 with lr 0.000500<p>\n","Epoch 26: Training loss 0.804150 validation loss 0.791692 with lr 0.000500<p>\n","Epoch 27: Training loss 0.804115 validation loss 0.791665 with lr 0.000500<p>\n","Epoch 28: Training loss 0.803623 validation loss 0.791838 with lr 0.000500<p>\n","Epoch 29: Training loss 0.803948 validation loss 0.792142 with lr 0.000500<p>\n","Epoch 30: Training loss 0.804133 validation loss 0.791528 with lr 0.000500<p>\n","Epoch 31: Training loss 0.803645 validation loss 0.791660 with lr 0.000500<p>\n","Epoch 32: Training loss 0.803960 validation loss 0.791802 with lr 0.000500<p>\n","Epoch 33: Training loss 0.803761 validation loss 0.791895 with lr 0.000500<p>\n","Epoch 34: Training loss 0.803798 validation loss 0.791241 with lr 0.000500<p>\n","Epoch 35: Training loss 0.803509 validation loss 0.791275 with lr 0.000500<p>\n","Epoch 36: Training loss 0.803665 validation loss 0.790943 with lr 0.000500<p>\n","Epoch 37: Training loss 0.803412 validation loss 0.791265 with lr 0.000500<p>\n","Epoch 38: Training loss 0.803196 validation loss 0.791112 with lr 0.000500<p>\n","Epoch 39: Training loss 0.802799 validation loss 0.791207 with lr 0.000500<p>\n","Epoch 40: Training loss 0.802705 validation loss 0.790584 with lr 0.000500<p>\n","Epoch 41: Training loss 0.802259 validation loss 0.790191 with lr 0.000500<p>\n","Epoch 42: Training loss 0.801756 validation loss 0.790041 with lr 0.000500<p>\n","Epoch 43: Training loss 0.801449 validation loss 0.789915 with lr 0.000500<p>\n","Epoch 44: Training loss 0.801043 validation loss 0.788661 with lr 0.000500<p>\n","Epoch 45: Training loss 0.800379 validation loss 0.788286 with lr 0.000500<p>\n","Epoch 46: Training loss 0.799999 validation loss 0.788367 with lr 0.000500<p>\n","Epoch 47: Training loss 0.799388 validation loss 0.788814 with lr 0.000500<p>\n","Epoch 48: Training loss 0.799000 validation loss 0.787275 with lr 0.000500<p>\n","Epoch 49: Training loss 0.798471 validation loss 0.786696 with lr 0.000500<p>\n","Epoch 50: Training loss 0.797666 validation loss 0.786980 with lr 0.000500<p>\n","Epoch 51: Training loss 0.797208 validation loss 0.785429 with lr 0.000500<p>\n","Epoch 52: Training loss 0.796282 validation loss 0.785050 with lr 0.000500<p>\n","Epoch 53: Training loss 0.795368 validation loss 0.785245 with lr 0.000500<p>\n","Epoch 54: Training loss 0.794924 validation loss 0.785039 with lr 0.000500<p>\n","Epoch 55: Training loss 0.794137 validation loss 0.784433 with lr 0.000500<p>\n","Epoch 56: Training loss 0.793751 validation loss 0.782589 with lr 0.000500<p>\n","Epoch 57: Training loss 0.793303 validation loss 0.783357 with lr 0.000500<p>\n","Epoch 58: Training loss 0.792360 validation loss 0.781823 with lr 0.000500<p>\n","Epoch 59: Training loss 0.792088 validation loss 0.780293 with lr 0.000500<p>\n","Epoch 60: Training loss 0.791235 validation loss 0.782502 with lr 0.000500<p>\n","Epoch 61: Training loss 0.790647 validation loss 0.780508 with lr 0.000500<p>\n","Epoch 62: Training loss 0.790433 validation loss 0.779639 with lr 0.000500<p>\n","Epoch 63: Training loss 0.789731 validation loss 0.779744 with lr 0.000500<p>\n","Epoch 64: Training loss 0.789132 validation loss 0.778712 with lr 0.000500<p>\n","Epoch 65: Training loss 0.788726 validation loss 0.779982 with lr 0.000500<p>\n","Epoch 66: Training loss 0.788330 validation loss 0.777589 with lr 0.000500<p>\n","Epoch 67: Training loss 0.788081 validation loss 0.777171 with lr 0.000500<p>\n","Epoch 68: Training loss 0.787282 validation loss 0.777606 with lr 0.000500<p>\n","Epoch 69: Training loss 0.787113 validation loss 0.776812 with lr 0.000500<p>\n","Epoch 70: Training loss 0.786447 validation loss 0.777026 with lr 0.000500<p>\n","Epoch 71: Training loss 0.786361 validation loss 0.776379 with lr 0.000500<p>\n","Epoch 72: Training loss 0.785965 validation loss 0.775424 with lr 0.000500<p>\n","Epoch 73: Training loss 0.785768 validation loss 0.775655 with lr 0.000500<p>\n","Epoch 74: Training loss 0.785338 validation loss 0.775396 with lr 0.000500<p>\n","Epoch 75: Training loss 0.785348 validation loss 0.774754 with lr 0.000500<p>\n","Epoch 76: Training loss 0.784870 validation loss 0.774651 with lr 0.000500<p>\n","Epoch 77: Training loss 0.784814 validation loss 0.775531 with lr 0.000500<p>\n","Epoch 78: Training loss 0.784791 validation loss 0.775029 with lr 0.000500<p>\n","Epoch 79: Training loss 0.784385 validation loss 0.775875 with lr 0.000500<p>\n","Epoch 80: Training loss 0.784187 validation loss 0.775855 with lr 0.000500<p>\n","Epoch 81: Training loss 0.784446 validation loss 0.775094 with lr 0.000500<p>\n","Epoch 82: Training loss 0.784099 validation loss 0.774851 with lr 0.000500<p>\n","Epoch 83: Training loss 0.784167 validation loss 0.774891 with lr 0.000500<p>\n","Epoch 84: Training loss 0.784137 validation loss 0.774757 with lr 0.000500<p>\n","Epoch 85: Training loss 0.784538 validation loss 0.774530 with lr 0.000500<p>\n","Epoch 86: Training loss 0.784087 validation loss 0.774349 with lr 0.000500<p>\n","Epoch 87: Training loss 0.783850 validation loss 0.774546 with lr 0.000500<p>\n","Epoch 88: Training loss 0.783920 validation loss 0.774921 with lr 0.000500<p>\n","Epoch 89: Training loss 0.783996 validation loss 0.774365 with lr 0.000500<p>\n","Epoch 90: Training loss 0.783885 validation loss 0.774594 with lr 0.000500<p>\n","Epoch 91: Training loss 0.784014 validation loss 0.774499 with lr 0.000500<p>\n","Epoch 92: Training loss 0.783798 validation loss 0.774811 with lr 0.000500<p>\n","Epoch 93: Training loss 0.783983 validation loss 0.774592 with lr 0.000500<p>\n","Epoch 94: Training loss 0.784164 validation loss 0.774804 with lr 0.000500<p>\n","Epoch 95: Training loss 0.784054 validation loss 0.774872 with lr 0.000500<p>\n","Epoch 96: Training loss 0.783752 validation loss 0.774308 with lr 0.000500<p>\n","Epoch 97: Training loss 0.783994 validation loss 0.774413 with lr 0.000500<p>\n","Epoch 98: Training loss 0.783707 validation loss 0.774745 with lr 0.000500<p>\n","Epoch 99: Training loss 0.783485 validation loss 0.774636 with lr 0.000500"]},"metadata":{}},{"output_type":"display_data","data":{"text/plain":["<Figure size 600x400 with 1 Axes>"],"image/png":"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\n"},"metadata":{}},{"output_type":"stream","name":"stderr","text":["/usr/local/lib/python3.10/dist-packages/numpy/core/shape_base.py:65: FutureWarning: The input object of type 'Tensor' is an array-like implementing one of the corresponding protocols (`__array__`, `__array_interface__` or `__array_struct__`); but not a sequence (or 0-D). In the future, this object will be coerced as if it was first converted using `np.array(obj)`. To retain the old behaviour, you have to either modify the type 'Tensor', or assign to an empty array created with `np.empty(correct_shape, dtype=object)`.\n","  ary = asanyarray(ary)\n"]},{"output_type":"stream","name":"stdout","text":["Validation loss decreased (inf --> 0.911322).  Saving model ...\n"]},{"output_type":"stream","name":"stderr","text":["/usr/local/lib/python3.10/dist-packages/numpy/core/shape_base.py:65: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.\n","  ary = asanyarray(ary)\n"]},{"output_type":"stream","name":"stdout","text":["Validation loss decreased (0.911322 --> 0.888493).  Saving model ...\n","Validation loss decreased (0.888493 --> 0.866666).  Saving model ...\n","Validation loss decreased (0.866666 --> 0.841949).  Saving model ...\n","Validation loss decreased (0.841949 --> 0.829526).  Saving model ...\n","Validation loss decreased (0.829526 --> 0.821801).  Saving model ...\n","Validation loss decreased (0.821801 --> 0.812855).  Saving model ...\n","Validation loss decreased (0.812855 --> 0.808916).  Saving model ...\n","Validation loss decreased (0.808916 --> 0.805657).  Saving model ...\n","Validation loss decreased (0.805657 --> 0.802810).  Saving model ...\n","Validation loss decreased (0.802810 --> 0.801424).  Saving model ...\n","Validation loss decreased (0.801424 --> 0.798371).  Saving model ...\n","Validation loss decreased (0.798371 --> 0.796599).  Saving model ...\n","Validation loss decreased (0.796599 --> 0.795919).  Saving model ...\n","Validation loss decreased (0.795919 --> 0.795227).  Saving model ...\n","Validation loss decreased (0.795227 --> 0.794181).  Saving model ...\n","Validation loss decreased (0.794181 --> 0.793764).  Saving model ...\n","Validation loss decreased (0.793764 --> 0.793158).  Saving model ...\n","Validation loss decreased (0.793158 --> 0.792545).  Saving model ...\n","Validation loss decreased (0.792545 --> 0.792179).  Saving model ...\n","Validation loss decreased (0.792179 --> 0.791984).  Saving model ...\n","Validation loss decreased (0.791984 --> 0.791692).  Saving model ...\n","Validation loss decreased (0.791692 --> 0.791665).  Saving model ...\n","Validation loss decreased (0.791665 --> 0.791528).  Saving model ...\n","Validation loss decreased (0.791528 --> 0.791241).  Saving model ...\n","Validation loss decreased (0.791241 --> 0.790943).  Saving model ...\n","Validation loss decreased (0.790943 --> 0.790584).  Saving model ...\n","Validation loss decreased (0.790584 --> 0.790191).  Saving model ...\n","Validation loss decreased (0.790191 --> 0.790041).  Saving model ...\n","Validation loss decreased (0.790041 --> 0.789915).  Saving model ...\n","Validation loss decreased (0.789915 --> 0.788661).  Saving model ...\n","Validation loss decreased (0.788661 --> 0.788286).  Saving model ...\n","Validation loss decreased (0.788286 --> 0.787275).  Saving model ...\n","Validation loss decreased (0.787275 --> 0.786696).  Saving model ...\n","Validation loss decreased (0.786696 --> 0.785429).  Saving model ...\n","Validation loss decreased (0.785429 --> 0.785050).  Saving model ...\n","Validation loss decreased (0.785050 --> 0.785039).  Saving model ...\n","Validation loss decreased (0.785039 --> 0.784433).  Saving model ...\n","Validation loss decreased (0.784433 --> 0.782589).  Saving model ...\n","Validation loss decreased (0.782589 --> 0.781823).  Saving model ...\n","Validation loss decreased (0.781823 --> 0.780293).  Saving model ...\n","Validation loss decreased (0.780293 --> 0.779639).  Saving model ...\n","Validation loss decreased (0.779639 --> 0.778712).  Saving model ...\n","Validation loss decreased (0.778712 --> 0.777589).  Saving model ...\n","Validation loss decreased (0.777589 --> 0.777171).  Saving model ...\n","Validation loss decreased (0.777171 --> 0.776812).  Saving model ...\n","Validation loss decreased (0.776812 --> 0.776379).  Saving model ...\n","Validation loss decreased (0.776379 --> 0.775424).  Saving model ...\n","Validation loss decreased (0.775424 --> 0.775396).  Saving model ...\n","Validation loss decreased (0.775396 --> 0.774754).  Saving model ...\n","Validation loss decreased (0.774754 --> 0.774651).  Saving model ...\n","Validation loss decreased (0.774651 --> 0.774530).  Saving model ...\n","Validation loss decreased (0.774530 --> 0.774349).  Saving model ...\n","Validation loss decreased (0.774349 --> 0.774308).  Saving model ...\n"]}],"source":["epochs = 100\n","lr = 0.0005 # 0.005 is very slow\n","\n","\n","# CrossEntropyLoss expects raw prediction values while NLLLoss expects log probabilities.\n","# criterion = nn.CrossEntropyLoss() # nn.NLLLoss()\n","# since we are using nn.LogSoftmax as final layer at model\n","criterion = nn.NLLLoss()\n","\n","#optimizer = AdamW(model.parameters(), lr)\n","optimizer = torch.optim.SGD(model.parameters(), lr=lr)\n","scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(optimizer, T_max=30, eta_min=1e-5)\n","\n","early_stopper = EarlyStopper(patience=10)\n","\n","mb = master_bar(range(epochs))\n","mb.names = ['Training loss', 'Validation loss']\n","\n","x = []\n","training_losses = []\n","validation_losses = []\n","\n","valid_mean_min = np.Inf\n","\n","till_batch = 100\n","\n","for epoch in mb:\n","  x.append(epoch)\n","  # Train\n","  i = 0\n","  model.train()\n","  total_loss = torch.Tensor([0.0]).to(device)\n","  for batch in progress_bar(train_dataloader, parent=mb):\n","    model.zero_grad()\n","    input = batch[0].to(device)\n","    label = batch[1].to(device)\n","    output = model(input)\n","    loss = criterion(output, label)\n","    # backward propagation\n","    loss.backward()\n","    optimizer.step()\n","    with torch.no_grad():\n","      total_loss += loss.item()\n","    i += 1\n","    #if i == till_batch:\n","    #  break\n","\n","  # decay lr\n","  scheduler.step()\n","\n","  mean = total_loss / len(train_dataloader)\n","  #mean = total_loss / till_batch\n","  training_losses.append(mean.cpu())\n","\n","  # Evaluate\n","  i = 0\n","  model.eval()\n","  validation_loss = torch.Tensor([0.0]).to(device)\n","  with torch.no_grad():\n","    for batch in progress_bar(val_dataloader, parent=mb):\n","      input = batch[0].to(device)\n","      label = batch[1].to(device)\n","      output = model(input)\n","      loss = criterion(output, label)\n","      validation_loss += loss.item()\n","      i += 1\n","      #if i == till_batch:\n","      #  break\n","\n","  val_mean = validation_loss / len(val_dataloader)\n","  #val_mean = validation_loss / till_batch\n","\n","  validation_losses.append(val_mean.cpu())\n","  # Update training chart\n","  mb.update_graph([[x, training_losses], [x, validation_losses]], [0,epochs])\n","  mb.write(f\"\\nEpoch {epoch}: Training loss {mean.item():.6f} validation loss {val_mean.item():.6f} with lr {lr:.6f}\")\n","  # save model if validation loss has decreased\n","  if val_mean.item() <= valid_mean_min:\n","      print('Validation loss decreased ({:.6f} --> {:.6f}).  Saving model ...'.format(\n","      valid_mean_min,\n","      val_mean.item()))\n","      torch.save(model.state_dict(), '/content/drive/MyDrive/Colab/ethnicolor/transformer_lastname.pt')\n","      valid_mean_min = val_mean.item()\n","\n","  if early_stopper.early_stop(val_mean.item()):\n","        break"]},{"cell_type":"markdown","metadata":{"id":"BAjnrNfQqsjQ"},"source":["## Save Model"]},{"cell_type":"code","execution_count":32,"metadata":{"id":"9ahPb_GaquBz","executionInfo":{"status":"ok","timestamp":1687915156404,"user_tz":420,"elapsed":18,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"}}},"outputs":[],"source":["torch.save(model.state_dict(), '/content/drive/MyDrive/Colab/ethnicolor/transformer_lastname_after_train.pt')"]},{"cell_type":"code","execution_count":null,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":1903,"status":"ok","timestamp":1685656311704,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"00kkuZuq8OEh","outputId":"c37d0d13-cb44-40db-809e-e6780ffa9df4"},"outputs":[{"data":{"text/plain":["<All keys matched successfully>"]},"execution_count":40,"metadata":{},"output_type":"execute_result"}],"source":["model.load_state_dict(torch.load('/content/drive/MyDrive/Colab/ethnicolor/transformer_fullname_after_train.pt'))"]},{"cell_type":"markdown","metadata":{"id":"YkVR5WaPBRin"},"source":["## Testing"]},{"cell_type":"code","execution_count":38,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":4,"status":"ok","timestamp":1687915178692,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"WClV6_m9yFsG","outputId":"f0a49dad-1450-4900-e83c-21dfe0b09285"},"outputs":[{"output_type":"stream","name":"stdout","text":["CUDA is available!  Training on GPU ...\n"]}],"source":["# check if CUDA is available\n","train_on_gpu = torch.cuda.is_available()\n","\n","if not train_on_gpu:\n","    print('CUDA is not available.  Training on CPU ...')\n","else:\n","    print('CUDA is available!  Training on GPU ...')"]},{"cell_type":"code","execution_count":34,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":16392,"status":"ok","timestamp":1687915173186,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"eDUpH4eNgb5O","outputId":"c4108bc6-dbf7-4b8d-94c6-ef6084ff9b7b"},"outputs":[{"output_type":"stream","name":"stderr","text":["Testing: 100%|██████████| 825/825 [00:14<00:00, 55.27it/s]\n","/usr/local/lib/python3.10/dist-packages/sklearn/metrics/_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n","  _warn_prf(average, modifier, msg_start, len(result))\n","/usr/local/lib/python3.10/dist-packages/sklearn/metrics/_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n","  _warn_prf(average, modifier, msg_start, len(result))\n"]},{"output_type":"stream","name":"stdout","text":["              precision    recall  f1-score   support\n","\n","       asian       0.38      0.03      0.06      3646\n","    hispanic       0.80      0.80      0.80     32431\n","    nh_black       0.48      0.09      0.16     10398\n","    nh_white       0.72      0.91      0.80     56236\n","       other       0.00      0.00      0.00      2889\n","\n","    accuracy                           0.74    105600\n","   macro avg       0.48      0.37      0.36    105600\n","weighted avg       0.69      0.74      0.69    105600\n","\n"]},{"output_type":"stream","name":"stderr","text":["/usr/local/lib/python3.10/dist-packages/sklearn/metrics/_classification.py:1344: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n","  _warn_prf(average, modifier, msg_start, len(result))\n"]}],"source":["from sklearn.metrics import confusion_matrix, classification_report, ConfusionMatrixDisplay\n","\n","# track test loss\n","test_loss = 0.0\n","\n","\n","class_correct = list(0. for i in range(n_categories))\n","class_total = list(0. for i in range(n_categories))\n","\n","\n","actual = []\n","predictions = []\n","\n","model.eval()\n","# iterate over test data\n","pbar = tqdm(test_dataloader, total=len(test_dataloader), position=0, desc=\"Testing\", leave=True)\n","for batch in pbar:\n","    # move tensors to GPU if CUDA is available\n","    input = batch[0].to(device)\n","    label = batch[1].to(device)\n","    output = model(input)\n","    loss = criterion(output, label)\n","    test_loss += loss.item()\n","    pred = torch.argmax(output, dim=1)\n","    correct_tensor = pred.eq(label.data.view_as(pred))\n","    correct = np.squeeze(correct_tensor.numpy()) if not train_on_gpu else np.squeeze(correct_tensor.cpu().numpy())\n","    # calculate test accuracy for each object class\n","    for i in range(label.shape[0]):\n","        l = label.data[i]\n","        class_correct[l.long()] += correct[i].item()\n","        class_total[l.long()] += 1\n","        # for confusion matrix\n","        actual.append(all_categories[label.data[i].item()])\n","        predictions.append(all_categories[pred.data[i].item()])\n","\n","\n","# plot confusion matrix\n","cm = confusion_matrix(actual, predictions, labels=all_categories)\n","print(classification_report(actual, predictions))"]},{"cell_type":"code","execution_count":35,"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":702},"executionInfo":{"elapsed":21,"status":"ok","timestamp":1687915173187,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"TrS2WIqxyJPL","outputId":"60b78b1a-0701-4962-f969-a3b62cb04375"},"outputs":[{"output_type":"execute_result","data":{"text/plain":["<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7fe25f2ee9e0>"]},"metadata":{},"execution_count":35},{"output_type":"display_data","data":{"text/plain":["<Figure size 800x800 with 2 Axes>"],"image/png":"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\n"},"metadata":{}}],"source":["import matplotlib.pyplot as plt\n","%matplotlib inline\n","\n","cmp = ConfusionMatrixDisplay(cm, display_labels=all_categories)\n","fig, ax = plt.subplots(figsize=(8,8))\n","cmp.plot(ax=ax, xticks_rotation='vertical')"]},{"cell_type":"code","execution_count":36,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":843,"status":"ok","timestamp":1687915174024,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"Kjz7DX4o9qlO","outputId":"daf51729-0055-41ad-d9f4-159c03441498"},"outputs":[{"output_type":"stream","name":"stdout","text":["Test Loss: 0.777511\n","\n","Test Accuracy of asian:  3% (127/3646)\n","Test Accuracy of hispanic: 79% (25849/32431)\n","Test Accuracy of nh_black:  9% (984/10398)\n","Test Accuracy of nh_white: 90% (51052/56236)\n","Test Accuracy of other:  0% ( 0/2889)\n","\n","Test Accuracy (Overall): 73% (78012/105600)\n"]}],"source":["# average test loss\n","test_loss = test_loss/len(test_dataloader)\n","print('Test Loss: {:.6f}\\n'.format(test_loss))\n","\n","for i in range(len(all_categories)):\n","    if class_total[i] > 0:\n","        print('Test Accuracy of %5s: %2d%% (%2d/%2d)' % (\n","            all_categories[i], 100 * class_correct[i] / class_total[i],\n","            np.sum(class_correct[i]), np.sum(class_total[i])))\n","    else:\n","        print('Test Accuracy of %5s: N/A (no training examples)' % (all_categories[i]))\n","\n","print('\\nTest Accuracy (Overall): %2d%% (%2d/%2d)' % (\n","    100. * np.sum(class_correct) / np.sum(class_total),\n","    np.sum(class_correct), np.sum(class_total)))"]},{"cell_type":"code","execution_count":37,"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"elapsed":13,"status":"ok","timestamp":1687915174024,"user":{"displayName":"Rajashekar Chintalapati","userId":"03596288833202137831"},"user_tz":420},"id":"Hy8oB6Kg-YM8","outputId":"fe76223d-abef-4406-e619-b0df178ecb37"},"outputs":[{"output_type":"stream","name":"stdout","text":["torch.Size([30])\n","torch.Size([1, 5])\n","tensor(3, device='cuda:0')\n","nh_white\n"]}],"source":["name = \"Ward Stephanie\"\n","name_tokens = lineToTensor(name)\n","inp = name_tokens\n","print(inp.shape)\n","out = model(inp.unsqueeze(0).to(device))\n","print(out.shape)\n","out = torch.argmax(out)\n","print(out)\n","print(all_categories[out.item()])"]},{"cell_type":"code","execution_count":null,"metadata":{"id":"dNfTPeMc_jbo"},"outputs":[],"source":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","name":"python3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.8.10"},"colab":{"provenance":[],"machine_shape":"hm","gpuType":"A100"},"accelerator":"GPU"},"nbformat":4,"nbformat_minor":0}