{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wJcYs_ERTnnI"
   },
   "source": [
    "##### Copyright 2021 The TensorFlow Authors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "cellView": "form",
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:43.743890Z",
     "iopub.status.busy": "2022-12-14T22:24:43.743253Z",
     "iopub.status.idle": "2022-12-14T22:24:43.747067Z",
     "shell.execute_reply": "2022-12-14T22:24:43.746542Z"
    },
    "id": "HMUDt0CiUJk9"
   },
   "outputs": [],
   "source": [
    "#@title Licensed under the Apache License, Version 2.0 (the \"License\");\n",
    "# you may not use this file except in compliance with the License.\n",
    "# You may obtain a copy of the License at\n",
    "#\n",
    "# https://www.apache.org/licenses/LICENSE-2.0\n",
    "#\n",
    "# Unless required by applicable law or agreed to in writing, software\n",
    "# distributed under the License is distributed on an \"AS IS\" BASIS,\n",
    "# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
    "# See the License for the specific language governing permissions and\n",
    "# limitations under the License."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "77z2OchJTk0l"
   },
   "source": [
    "# Estimator から Keras API に移行する\n",
    "\n",
    "<table class=\"tfo-notebook-buttons\" align=\"left\">\n",
    "  <td>     <a target=\"_blank\" href=\"https://www.tensorflow.org/guide/migrate/migrating_estimator\">     <img src=\"https://www.tensorflow.org/images/tf_logo_32px.png\">     TensorFlow.org で表示</a>\n",
    "</td>\n",
    "  <td>     <a target=\"_blank\" href=\"https://colab.research.google.com/github/tensorflow/docs-l10n/blob/master/site/ja/guide/migrate/migrating_estimator.ipynb\">     <img src=\"https://www.tensorflow.org/images/colab_logo_32px.png\">     Google Colab で実行</a>\n",
    "</td>\n",
    "  <td><a target=\"_blank\" href=\"https://github.com/tensorflow/docs-l10n/blob/master/site/ja/guide/migrate/migrating_estimator.ipynb\"><img src=\"https://www.tensorflow.org/images/GitHub-Mark-32px.png\"> GitHub でソースを表示</a></td>\n",
    "  <td><a href=\"https://storage.googleapis.com/tensorflow_docs/docs-l10n/site/ja/guide/migrate/migrating_estimator.ipynb\"><img src=\"https://www.tensorflow.org/images/download_logo_32px.png\">ノートブックをダウンロード</a></td>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "meUTrR4I6m1C"
   },
   "source": [
    "このガイドでは、TensorFlow 1 の `tf.estimator.Estimator` API から TensorFlow 2 の `tf.keras` API に移行する方法を示します。最初に、`tf.estimator.Estimator` を使用してトレーニングと評価のための基本モデルをセットアップして実行します。次に、`tf.keras` API を使用して  TensorFlow 2 で同等の手順を実行します。また、`tf.GradientTape` をサブクラス化し、`tf.keras.Model` を使用してトレーニングの手順をカスタマイズする方法も学びます。\n",
    "\n",
    "- TensorFlow 1 では、高レベルの `tf.estimator.Estimator` API を使用して、モデルのトレーニングと評価、推論の実行、およびモデルの保存（提供用）を行うことができます。\n",
    "- TensorFlow 2 では、Keras API を使用して、[モデルの構築](https://www.tensorflow.org/guide/keras/custom_layers_and_models)、勾配の適用、 [トレーニング](https://www.tensorflow.org/guide/keras/customizing_what_happens_in_fit)、評価、予測などの前述のタスクを実行します。\n",
    "\n",
    "（モデル/チェックポイント保存ワークフローを TensorFlow 2 に移行するには、[SavedModel](saved_model.ipynb) および [Checkpoint](checkpoint_saved.ipynb) 移行ガイドを確認してください。）"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "YdZSoIXEbhg-"
   },
   "source": [
    "## セットアップ\n",
    "\n",
    "インポートと単純なデータセットから始めます。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:43.750656Z",
     "iopub.status.busy": "2022-12-14T22:24:43.750108Z",
     "iopub.status.idle": "2022-12-14T22:24:45.655171Z",
     "shell.execute_reply": "2022-12-14T22:24:45.654333Z"
    },
    "id": "iE0vSfMXumKI"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-12-14 22:24:44.683460: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer.so.7'; dlerror: libnvinfer.so.7: cannot open shared object file: No such file or directory\n",
      "2022-12-14 22:24:44.683565: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer_plugin.so.7'; dlerror: libnvinfer_plugin.so.7: cannot open shared object file: No such file or directory\n",
      "2022-12-14 22:24:44.683576: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import tensorflow.compat.v1 as tf1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:45.659460Z",
     "iopub.status.busy": "2022-12-14T22:24:45.659049Z",
     "iopub.status.idle": "2022-12-14T22:24:45.663106Z",
     "shell.execute_reply": "2022-12-14T22:24:45.662536Z"
    },
    "id": "m7rnGxsXtDkV"
   },
   "outputs": [],
   "source": [
    "features = [[1., 1.5], [2., 2.5], [3., 3.5]]\n",
    "labels = [[0.3], [0.5], [0.7]]\n",
    "eval_features = [[4., 4.5], [5., 5.5], [6., 6.5]]\n",
    "eval_labels = [[0.8], [0.9], [1.]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4uXff1BEssdE"
   },
   "source": [
    "## TensorFlow 1: tf.estimator.Estimator でトレーニングと評価を行う\n",
    "\n",
    "この例では、TensorFlow 1 で `tf.estimator.Estimator` を使用してトレーニングと評価を実行する方法を示します。\n",
    "\n",
    "いくつかの関数を定義することから始めます。トレーニングデータの入力関数、評価データの評価入力関数、および特徴量とラベルを使用してトレーニング演算がどのように定義されるかを `Estimator` に伝えるモデル関数です。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:45.666523Z",
     "iopub.status.busy": "2022-12-14T22:24:45.665951Z",
     "iopub.status.idle": "2022-12-14T22:24:45.670880Z",
     "shell.execute_reply": "2022-12-14T22:24:45.670332Z"
    },
    "id": "lqe9obf7suIj"
   },
   "outputs": [],
   "source": [
    "def _input_fn():\n",
    "  return tf1.data.Dataset.from_tensor_slices((features, labels)).batch(1)\n",
    "\n",
    "def _eval_input_fn():\n",
    "  return tf1.data.Dataset.from_tensor_slices(\n",
    "      (eval_features, eval_labels)).batch(1)\n",
    "\n",
    "def _model_fn(features, labels, mode):\n",
    "  logits = tf1.layers.Dense(1)(features)\n",
    "  loss = tf1.losses.mean_squared_error(labels=labels, predictions=logits)\n",
    "  optimizer = tf1.train.AdagradOptimizer(0.05)\n",
    "  train_op = optimizer.minimize(loss, global_step=tf1.train.get_global_step())\n",
    "  return tf1.estimator.EstimatorSpec(mode, loss=loss, train_op=train_op)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "44bf417bf9c0"
   },
   "source": [
    "`Estimator` をインスタンス化し、モデルをトレーニングします。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:45.674157Z",
     "iopub.status.busy": "2022-12-14T22:24:45.673638Z",
     "iopub.status.idle": "2022-12-14T22:24:50.086817Z",
     "shell.execute_reply": "2022-12-14T22:24:50.085933Z"
    },
    "id": "922720812527"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Using default config.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Using temporary folder as model directory: /tmpfs/tmp/tmpzy9ux6di\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Using config: {'_model_dir': '/tmpfs/tmp/tmpzy9ux6di', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': None, '_save_checkpoints_secs': 600, '_session_config': allow_soft_placement: true\n",
      "graph_options {\n",
      "  rewrite_options {\n",
      "    meta_optimizer_iterations: ONE\n",
      "  }\n",
      "}\n",
      ", '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_experimental_max_worker_delay_secs': None, '_session_creation_timeout_secs': 7200, '_checkpoint_save_graph_def': True, '_service': None, '_cluster_spec': ClusterSpec({}), '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/training/training_util.py:396: Variable.initialized_value (from tensorflow.python.ops.variables) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use Variable.read_value. Variables in 2.X are initialized automatically both in eager and graph (inside tf.defun) contexts.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling model_fn.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/training/adagrad.py:138: calling Constant.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Done calling model_fn.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Create CheckpointSaverHook.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Graph was finalized.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Running local_init_op.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Done running local_init_op.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling checkpoint listeners before saving checkpoint 0...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Saving checkpoints for 0 into /tmpfs/tmp/tmpzy9ux6di/model.ckpt.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling checkpoint listeners after saving checkpoint 0...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:loss = 0.3302933, step = 0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling checkpoint listeners before saving checkpoint 3...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Saving checkpoints for 3 into /tmpfs/tmp/tmpzy9ux6di/model.ckpt.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling checkpoint listeners after saving checkpoint 3...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Loss for final step: 0.9824529.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow_estimator.python.estimator.estimator.Estimator at 0x7fdb44fb6e50>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator = tf1.estimator.Estimator(model_fn=_model_fn)\n",
    "estimator.train(_input_fn)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "17c9933c2d89"
   },
   "source": [
    "評価セットを使用してプログラムを評価します。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:50.090488Z",
     "iopub.status.busy": "2022-12-14T22:24:50.089913Z",
     "iopub.status.idle": "2022-12-14T22:24:50.484692Z",
     "shell.execute_reply": "2022-12-14T22:24:50.483816Z"
    },
    "id": "HsOpjW5plH9Q"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Calling model_fn.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Done calling model_fn.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Starting evaluation at 2022-12-14T22:24:50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Graph was finalized.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from /tmpfs/tmp/tmpzy9ux6di/model.ckpt-3\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Running local_init_op.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Done running local_init_op.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Inference Time : 0.25545s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Finished evaluation at 2022-12-14-22:24:50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Saving dict for global step 3: global_step = 3, loss = 2.2362373\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Saving 'checkpoint_path' summary for global step 3: /tmpfs/tmp/tmpzy9ux6di/model.ckpt-3\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 2.2362373, 'global_step': 3}"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator.evaluate(_eval_input_fn)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KEmzBjfnsxwT"
   },
   "source": [
    "## TensorFlow 2: 組み込みの Keras メソッドを使用してトレーニングと評価を行う\n",
    "\n",
    "この例では、TensorFlow 2 で Keras `Model.fit` と `Model.evaluate` を使用してトレーニングと評価を実行する方法を示します（詳細については、[組み込みメソッドを使用したトレーニングと評価](https://www.tensorflow.org/guide/keras/train_and_evaluate)ガイドを参照してください）。\n",
    "\n",
    "- `tf.data.Dataset` API を使用してデータセットパイプラインを準備することから始めます。\n",
    "- 1 つの線形（<code>tf.keras.layers.Dense</code>）レイヤーを持つ単純な Keras <a>Sequential</a> モデルを定義します。\n",
    "- Adagrad オプティマイザをインスタンス化します（`tf.keras.optimizers.Adagrad`）。\n",
    "- `optimizer` 変数と平均二乗誤差（`\"mse\"`）損失を `Model.compile` に渡して、トレーニング用のモデルを構成します。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:50.488564Z",
     "iopub.status.busy": "2022-12-14T22:24:50.487956Z",
     "iopub.status.idle": "2022-12-14T22:24:50.518528Z",
     "shell.execute_reply": "2022-12-14T22:24:50.517901Z"
    },
    "id": "atVciNgPs0fw"
   },
   "outputs": [],
   "source": [
    "dataset = tf.data.Dataset.from_tensor_slices((features, labels)).batch(1)\n",
    "eval_dataset = tf.data.Dataset.from_tensor_slices(\n",
    "      (eval_features, eval_labels)).batch(1)\n",
    "\n",
    "model = tf.keras.models.Sequential([tf.keras.layers.Dense(1)])\n",
    "optimizer = tf.keras.optimizers.Adagrad(learning_rate=0.05)\n",
    "\n",
    "model.compile(optimizer=optimizer, loss=\"mse\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ed17a6291959"
   },
   "source": [
    "これで、`Model.fit` を呼び出してモデルをトレーニングする準備が整いました。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:50.522081Z",
     "iopub.status.busy": "2022-12-14T22:24:50.521514Z",
     "iopub.status.idle": "2022-12-14T22:24:51.043078Z",
     "shell.execute_reply": "2022-12-14T22:24:51.042327Z"
    },
    "id": "a0b732534501"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "1/3 [=========>....................] - ETA: 0s - loss: 3.7549"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
      "3/3 [==============================] - 0s 5ms/step - loss: 10.1551\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fdb48799a60>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(dataset)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "74767288a2ea"
   },
   "source": [
    "最後に、`Model.evaluate` を使用してモデルを評価します。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:51.046749Z",
     "iopub.status.busy": "2022-12-14T22:24:51.046061Z",
     "iopub.status.idle": "2022-12-14T22:24:51.144209Z",
     "shell.execute_reply": "2022-12-14T22:24:51.143536Z"
    },
    "id": "Kip65sYBlKiu"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "1/3 [=========>....................] - ETA: 0s - loss: 25.7323"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
      "3/3 [==============================] - 0s 3ms/step - loss: 41.3360\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 41.33598709106445}"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(eval_dataset, return_dict=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BuVYN0CHs5sD"
   },
   "source": [
    "## TensorFlow 2: カスタムトレーニングステップと組み込みの Keras メソッドを使用してトレーニングと評価を行う"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gHx_RUL8xcJ3"
   },
   "source": [
    "TensorFlow 2 では、`tf.keras.callbacks.Callback` や `tf.distribute.Strategy` などの組み込みのトレーニングサポートを引き続き利用しながら、`tf.GradientTape` を使用して独自のカスタムトレーニングステップ関数を作成して、フォワードパスとバックワードパスを実行することもできます。（詳細については、[Model.fit の処理をカスタマイズする](https://www.tensorflow.org/guide/keras/customizing_what_happens_in_fit)および[トレーニングループの新規作成](https://www.tensorflow.org/guide/keras/writing_a_training_loop_from_scratch)を参照してください。)\n",
    "\n",
    "この例では、`tf.keras.Sequential` をオーバーライドする `Model.train_step` をサブクラス化することにより、カスタム `tf.keras.Model` を作成することから始めます。（[tf.keras.Model のサブクラス化](https://www.tensorflow.org/guide/keras/custom_layers_and_models)について詳しくご覧ください)。そのクラス内で、データのバッチごとに 1 つのトレーニングステップでフォワードパスとバックワードパスを実行するカスタムの `train_step` 関数を定義します。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:51.147808Z",
     "iopub.status.busy": "2022-12-14T22:24:51.147249Z",
     "iopub.status.idle": "2022-12-14T22:24:51.152083Z",
     "shell.execute_reply": "2022-12-14T22:24:51.151459Z"
    },
    "id": "rSz_y0zOs8h2"
   },
   "outputs": [],
   "source": [
    "class CustomModel(tf.keras.Sequential):\n",
    "  \"\"\"A custom sequential model that overrides `Model.train_step`.\"\"\"\n",
    "\n",
    "  def train_step(self, data):\n",
    "    batch_data, labels = data\n",
    "\n",
    "    with tf.GradientTape() as tape:\n",
    "      predictions = self(batch_data, training=True)\n",
    "      # Compute the loss value (the loss function is configured\n",
    "      # in `Model.compile`).\n",
    "      loss = self.compiled_loss(labels, predictions)\n",
    "\n",
    "    # Compute the gradients of the parameters with respect to the loss.\n",
    "    gradients = tape.gradient(loss, self.trainable_variables)\n",
    "    # Perform gradient descent by updating the weights/parameters.\n",
    "    self.optimizer.apply_gradients(zip(gradients, self.trainable_variables))\n",
    "    # Update the metrics (includes the metric that tracks the loss).\n",
    "    self.compiled_metrics.update_state(labels, predictions)\n",
    "    # Return a dict mapping metric names to the current values.\n",
    "    return {m.name: m.result() for m in self.metrics}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ee7c4f94d69b"
   },
   "source": [
    "次に、前と同じように以下を実行します。\n",
    "\n",
    "- `tf.data.Dataset` でデータセットパイプラインを準備します。\n",
    "- 1 つの `tf.keras.layers.Dense` レイヤーで単純なモデルを定義します。\n",
    "- Adagrad のインスタンス化（`tf.keras.optimizers.Adagrad`）\n",
    "- 損失関数として平均二乗誤差（`\"mse\"`）を使用しながら、`Model.compile` でトレーニング用のモデルを構成します。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:51.155385Z",
     "iopub.status.busy": "2022-12-14T22:24:51.154893Z",
     "iopub.status.idle": "2022-12-14T22:24:51.170611Z",
     "shell.execute_reply": "2022-12-14T22:24:51.169983Z"
    },
    "id": "01fcc2b1292c"
   },
   "outputs": [],
   "source": [
    "dataset = tf.data.Dataset.from_tensor_slices((features, labels)).batch(1)\n",
    "eval_dataset = tf.data.Dataset.from_tensor_slices(\n",
    "      (eval_features, eval_labels)).batch(1)\n",
    "\n",
    "model = CustomModel([tf.keras.layers.Dense(1)])\n",
    "optimizer = tf.keras.optimizers.Adagrad(learning_rate=0.05)\n",
    "\n",
    "model.compile(optimizer=optimizer, loss=\"mse\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "844543802ff5"
   },
   "source": [
    "`Model.fit` を呼び出してモデルをトレーニングします。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:51.174102Z",
     "iopub.status.busy": "2022-12-14T22:24:51.173582Z",
     "iopub.status.idle": "2022-12-14T22:24:51.532228Z",
     "shell.execute_reply": "2022-12-14T22:24:51.531556Z"
    },
    "id": "211be3620765"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "1/3 [=========>....................] - ETA: 0s - loss: 0.2560"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
      "3/3 [==============================] - 0s 3ms/step - loss: 0.4927\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fdb44170880>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(dataset)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "c93b9d6fc9d7"
   },
   "source": [
    "最後に、`Model.evaluate` を使用してプログラムを評価します。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-12-14T22:24:51.535877Z",
     "iopub.status.busy": "2022-12-14T22:24:51.535158Z",
     "iopub.status.idle": "2022-12-14T22:24:51.616781Z",
     "shell.execute_reply": "2022-12-14T22:24:51.616056Z"
    },
    "id": "nYO2wI1SlNCG"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "1/3 [=========>....................] - ETA: 0s - loss: 0.7179"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
      "3/3 [==============================] - 0s 3ms/step - loss: 1.4714\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 1.4713608026504517}"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(eval_dataset, return_dict=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "e9b5c9a4747b"
   },
   "source": [
    "## Next steps\n",
    "\n",
    "役に立つと思われる追加の Keras リソースは次のとおりです。\n",
    "\n",
    "- ガイド: [組み込みメソッドを使用したトレーニングと評価](https://www.tensorflow.org/guide/keras/train_and_evaluate)\n",
    "- ガイド: [Model.fit の処理をカスタマイズする](https://www.tensorflow.org/guide/keras/customizing_what_happens_in_fit)\n",
    "- ガイド: [トレーニングループの新規作成](https://www.tensorflow.org/guide/keras/writing_a_training_loop_from_scratch)\n",
    "- ガイド: [サブクラス化によるレイヤーとモデルの新規作成](https://www.tensorflow.org/guide/keras/custom_layers_and_models)\n",
    "\n",
    "次のガイドは、`tf.estimator` API から分散ストラテジーのワークフローを移行するのに役立ちます。\n",
    "\n",
    "- [TPUEstimator から TPUStrategy への移行](tpu_estimator.ipynb)\n",
    "- [シングルワーカーのマルチ GPUトレーニングを移行する](mirrored_strategy.ipynb)\n",
    "- [マルチワーカー CPU/GPUトレーニングを移行する](multi_worker_cpu_gpu_training.ipynb)"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "migrating_estimator.ipynb",
   "toc_visible": true
  },
  "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": "3.9.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}