notebooks/nmt_with_attention_tensorflow.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "s_qNSzzyaCbD"
   },
   "source": [
    "##### Copyright 2019 The TensorFlow Authors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "cellView": "form",
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:11.066314Z",
     "iopub.status.busy": "2023-12-07T12:22:11.066033Z",
     "iopub.status.idle": "2023-12-07T12:22:11.070372Z",
     "shell.execute_reply": "2023-12-07T12:22:11.069673Z"
    },
    "id": "jmjh290raIky"
   },
   "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": "J0Qjg6vuaHNt"
   },
   "source": [
    "# Neural machine translation with attention"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "AOpGoE2T-YXS"
   },
   "source": [
    "<table class=\"tfo-notebook-buttons\" align=\"left\">\n",
    "  <td>\n",
    "    <a target=\"_blank\" href=\"https://www.tensorflow.org/text/tutorials/nmt_with_attention\">\n",
    "    <img src=\"https://www.tensorflow.org/images/tf_logo_32px.png\" />\n",
    "    View on TensorFlow.org</a>\n",
    "  </td>\n",
    "  <td>\n",
    "    <a target=\"_blank\" href=\"https://colab.research.google.com/github/tensorflow/text/blob/master/docs/tutorials/nmt_with_attention.ipynb\">\n",
    "    <img src=\"https://www.tensorflow.org/images/colab_logo_32px.png\" />\n",
    "    Run in Google Colab</a>\n",
    "  </td>\n",
    "  <td>\n",
    "    <a target=\"_blank\" href=\"/~https://github.com/tensorflow/text/blob/master/docs/tutorials/nmt_with_attention.ipynb\">\n",
    "    <img src=\"https://www.tensorflow.org/images/GitHub-Mark-32px.png\" />\n",
    "    View source on GitHub</a>\n",
    "  </td>\n",
    "  <td>\n",
    "    <a href=\"https://storage.googleapis.com/tensorflow_docs/text/docs/tutorials/nmt_with_attention.ipynb\"><img src=\"https://www.tensorflow.org/images/download_logo_32px.png\" />Download notebook</a>\n",
    "  </td>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Xh8WNEwYA3BW"
   },
   "source": [
    "This tutorial demonstrates how to train a sequence-to-sequence (seq2seq) model for Spanish-to-English translation roughly based on [Effective Approaches to Attention-based Neural Machine Translation](https://arxiv.org/abs/1508.04025v5) (Luong et al., 2015). \n",
    "\n",
    "<table>\n",
    "<tr>\n",
    "  <td>\n",
    "   <img width=400 src=\"https://www.tensorflow.org/images/tutorials/transformer/RNN%2Battention-words-spa.png\"/>\n",
    "  </td>\n",
    "</tr>\n",
    "<tr>\n",
    "  <th colspan=1>This tutorial: An encoder/decoder connected by attention.</th>\n",
    "<tr>\n",
    "</table>\n",
    "\n",
    "While this architecture is somewhat outdated, it is still a very useful project to work through to get a deeper understanding of sequence-to-sequence models and attention mechanisms (before going on to [Transformers](transformer.ipynb))."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "yAmSR1FaqKrl"
   },
   "source": [
    "## Setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:15.711634Z",
     "iopub.status.busy": "2023-12-07T12:22:15.711335Z",
     "iopub.status.idle": "2023-12-07T12:22:18.532312Z",
     "shell.execute_reply": "2023-12-07T12:22:18.531513Z"
    },
    "id": "tnxXKDjq3jEL"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-23 17:55:08.760301: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
      "2024-01-23 17:55:08.760349: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
      "2024-01-23 17:55:08.779875: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
      "2024-01-23 17:55:08.860652: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2024-01-23 17:55:09.943636: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "import typing\n",
    "from typing import Any, Tuple\n",
    "\n",
    "import einops\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.ticker as ticker\n",
    "\n",
    "import tensorflow as tf\n",
    "import tensorflow_text as tf_text\n",
    "\n",
    "from nltk.translate.bleu_score import sentence_bleu, corpus_bleu"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "l_yq8kvIqoqQ"
   },
   "source": [
    "This tutorial uses a lot of low level API's where it's easy to get shapes wrong. This class is used to check shapes throughout the tutorial.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:18.536971Z",
     "iopub.status.busy": "2023-12-07T12:22:18.536253Z",
     "iopub.status.idle": "2023-12-07T12:22:18.541973Z",
     "shell.execute_reply": "2023-12-07T12:22:18.541286Z"
    },
    "id": "KqFqKi4fqN9X"
   },
   "outputs": [],
   "source": [
    "#@title\n",
    "class ShapeChecker():\n",
    "  def __init__(self):\n",
    "    # Keep a cache of every axis-name seen\n",
    "    self.shapes = {}\n",
    "\n",
    "  def __call__(self, tensor, names, broadcast=False):\n",
    "    if not tf.executing_eagerly():\n",
    "      return\n",
    "\n",
    "    parsed = einops.parse_shape(tensor, names)\n",
    "\n",
    "    for name, new_dim in parsed.items():\n",
    "      old_dim = self.shapes.get(name, None)\n",
    "      \n",
    "      if (broadcast and new_dim == 1):\n",
    "        continue\n",
    "\n",
    "      if old_dim is None:\n",
    "        # If the axis name is new, add its length to the cache.\n",
    "        self.shapes[name] = new_dim\n",
    "        continue\n",
    "\n",
    "      if new_dim != old_dim:\n",
    "        raise ValueError(f\"Shape mismatch for dimension: '{name}'\\n\"\n",
    "                         f\"    found: {new_dim}\\n\"\n",
    "                         f\"    expected: {old_dim}\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gjUROhJfH3ML"
   },
   "source": [
    "## The data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "puE_K74DIE9W"
   },
   "source": [
    "The tutorial uses a language dataset provided by [Anki](http://www.manythings.org/anki/). This dataset contains language translation pairs in the format:\n",
    "\n",
    "```\n",
    "May I borrow this book?\t¿Puedo tomar prestado este libro?\n",
    "```\n",
    "\n",
    "They have a variety of languages available, but this example uses the English-Spanish dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wfodePkj3jEa"
   },
   "source": [
    "### Download and prepare the dataset\n",
    "\n",
    "For convenience, a copy of this dataset is hosted on Google Cloud, but you can also download your own copy. After downloading the dataset, here are the steps you need to take to prepare the data:\n",
    "\n",
    "1. Add a *start* and *end* token to each sentence.\n",
    "2. Clean the sentences by removing special characters.\n",
    "3. Create a word index and reverse word index (dictionaries mapping from word → id and id → word).\n",
    "4. Pad each sentence to a maximum length."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:18.545276Z",
     "iopub.status.busy": "2023-12-07T12:22:18.545023Z",
     "iopub.status.idle": "2023-12-07T12:22:18.667063Z",
     "shell.execute_reply": "2023-12-07T12:22:18.666394Z"
    },
    "id": "kRVATYOgJs1b"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/marwan/.local/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    },
    {
     "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>en</th>\n",
       "      <th>fr</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Jamaica: “I am HIV”</td>\n",
       "      <td>Jamaïque : J’ai le VIH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>It's widely acknowledged, in the Caribbean and...</td>\n",
       "      <td>Il est largement reconnu, dans les Caraïbes et...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>For this woman, however, photographed in the s...</td>\n",
       "      <td>Pour cette femme, cependant, photographiée dan...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>As Bacon writes on her blog:</td>\n",
       "      <td>Comme Bacon écrit sur son blog:</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>“When I asked to take her picture, I suggested...</td>\n",
       "      <td>“Quand je lui ai demandé de la prendre en phot...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  en  \\\n",
       "0                                Jamaica: “I am HIV”   \n",
       "1  It's widely acknowledged, in the Caribbean and...   \n",
       "2  For this woman, however, photographed in the s...   \n",
       "3                       As Bacon writes on her blog:   \n",
       "4  “When I asked to take her picture, I suggested...   \n",
       "\n",
       "                                                  fr  \n",
       "0                             Jamaïque : J’ai le VIH  \n",
       "1  Il est largement reconnu, dans les Caraïbes et...  \n",
       "2  Pour cette femme, cependant, photographiée dan...  \n",
       "3                    Comme Bacon écrit sur son blog:  \n",
       "4  “Quand je lui ai demandé de la prendre en phot...  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from datasets import load_dataset\n",
    "\n",
    "dataset = load_dataset(\"Nicolas-BZRD/Parallel_Global_Voices_English_French\", split='train').to_pandas()\n",
    "dataset.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_data(dataset):\n",
    "  context = np.array([context for context in dataset['fr']])\n",
    "  target = np.array([target for target in dataset['en']])\n",
    "\n",
    "  return target, context"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:18.677543Z",
     "iopub.status.busy": "2023-12-07T12:22:18.676961Z",
     "iopub.status.idle": "2023-12-07T12:22:19.206716Z",
     "shell.execute_reply": "2023-12-07T12:22:19.205959Z"
    },
    "id": "cTbSbBz55QtF"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Seulement, aujourd'hui, elle ne porte pas de bracelets.\n"
     ]
    }
   ],
   "source": [
    "target_raw, context_raw = load_data(dataset)\n",
    "print(context_raw[-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:19.209949Z",
     "iopub.status.busy": "2023-12-07T12:22:19.209658Z",
     "iopub.status.idle": "2023-12-07T12:22:19.213499Z",
     "shell.execute_reply": "2023-12-07T12:22:19.212865Z"
    },
    "id": "lH_dPY8TRp3c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "But now she's not wearing bangles.\n"
     ]
    }
   ],
   "source": [
    "print(target_raw[-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "342060"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(context_raw)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rgCLkfv5uO3d"
   },
   "source": [
    "### Create a tf.data dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "PfVWx3WaI5Df"
   },
   "source": [
    "From these arrays of strings you can create a `tf.data.Dataset` of strings that shuffles and batches them efficiently:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:19.216791Z",
     "iopub.status.busy": "2023-12-07T12:22:19.216230Z",
     "iopub.status.idle": "2023-12-07T12:22:22.042643Z",
     "shell.execute_reply": "2023-12-07T12:22:22.041764Z"
    },
    "id": "3rZFgz69nMPa"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-23 17:55:20.754355: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:20.952212: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:20.952273: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:20.954039: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:20.954089: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:20.954128: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:21.320288: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:21.320689: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:21.320722: I tensorflow/core/common_runtime/gpu/gpu_device.cc:2022] Could not identify NUMA node of platform GPU id 0, defaulting to 0.  Your kernel may not have been built with NUMA support.\n",
      "2024-01-23 17:55:21.320871: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-01-23 17:55:21.320937: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1929] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 5606 MB memory:  -> device: 0, name: NVIDIA GeForce RTX 3060 Ti, pci bus id: 0000:01:00.0, compute capability: 8.6\n"
     ]
    }
   ],
   "source": [
    "BUFFER_SIZE = len(context_raw)\n",
    "BATCH_SIZE = 64\n",
    "\n",
    "# Set a random seed for reproducibility\n",
    "np.random.seed(42)\n",
    "\n",
    "# Shuffle the indices\n",
    "indices = np.arange(len(target_raw))\n",
    "np.random.shuffle(indices)\n",
    "\n",
    "# Determine the sizes of the train, validation, and test sets\n",
    "train_size = int(0.8 * len(target_raw))\n",
    "val_size = int(0.1 * len(target_raw))\n",
    "test_size = len(target_raw) - train_size - val_size\n",
    "\n",
    "# Use the shuffled indices to create subsets\n",
    "train_indices = indices[:train_size]\n",
    "val_indices = indices[train_size:train_size + val_size]\n",
    "test_indices = indices[train_size + val_size:]\n",
    "\n",
    "# Create datasets\n",
    "train_raw = (\n",
    "    tf.data.Dataset\n",
    "    .from_tensor_slices((context_raw[train_indices], target_raw[train_indices]))\n",
    "    .shuffle(BUFFER_SIZE)\n",
    "    .batch(BATCH_SIZE))\n",
    "\n",
    "val_raw = (\n",
    "    tf.data.Dataset\n",
    "    .from_tensor_slices((context_raw[val_indices], target_raw[val_indices]))\n",
    "    .shuffle(BUFFER_SIZE)\n",
    "    .batch(BATCH_SIZE))\n",
    "\n",
    "test_raw = (\n",
    "    tf.data.Dataset\n",
    "    .from_tensor_slices((context_raw[test_indices], target_raw[test_indices]))\n",
    "    .shuffle(BUFFER_SIZE)\n",
    "    .batch(BATCH_SIZE))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of context strings in the test dataset: 34206\n",
      "Total number of context strings in the validation dataset: 34206\n",
      "Total number of context strings in the training dataset: 273648\n",
      "Total number of context strings in the dataset: 342060\n"
     ]
    }
   ],
   "source": [
    "nb_context_test = test_raw.reduce(0, lambda x, batch: x + tf.shape(batch[0])[0]).numpy()\n",
    "nb_context_val = val_raw.reduce(0, lambda x, batch: x + tf.shape(batch[0])[0]).numpy()\n",
    "nb_context_train = train_raw.reduce(0, lambda x, batch: x + tf.shape(batch[0])[0]).numpy()\n",
    "\n",
    "print(\"Total number of context strings in the test dataset:\", nb_context_test)\n",
    "print(\"Total number of context strings in the validation dataset:\", nb_context_val)\n",
    "print(\"Total number of context strings in the training dataset:\", nb_context_train)\n",
    "\n",
    "print(\"Total number of context strings in the dataset:\", nb_context_test + nb_context_val + nb_context_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:22.046591Z",
     "iopub.status.busy": "2023-12-07T12:22:22.046323Z",
     "iopub.status.idle": "2023-12-07T12:22:22.178092Z",
     "shell.execute_reply": "2023-12-07T12:22:22.177347Z"
    },
    "id": "qc6-NK1GtWQt"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tf.Tensor(\n",
      "[b\"\\xc3\\x87a semble toujours impossible jusqu'\\xc3\\xa0 ce que \\xc3\\xa7a devienne une r\\xc3\\xa9alit\\xc3\\xa9, si vous croyez en vous-m\\xc3\\xaame et persistez, vous verrez les r\\xc3\\xa9sultats.\"\n",
      " b\"Les pays \\xc3\\xa9trangers qui ont soutenu les Khmers Rouges, ou \\xc3\\xa9t\\xc3\\xa9 les catalyseurs principaux de l'\\xc3\\xa9mergence de ce r\\xc3\\xa9gime cruel, ne seront pas train\\xc3\\xa9s en justice.\"\n",
      " b\"De toute fa\\xc3\\xa7on, nous esp\\xc3\\xa9rons qu'il existe un juste milieu qui n'implique pas de schtroumpfs, des nains ou l'am\\xc3\\xa9ricanisation compl\\xc3\\xa8te des traditions dans le monde.\"\n",
      " b'Pointant du doigt la nette fracture entre la classe dirigeante et le reste du pays, le blogeur Alick Nyasulu \\xc3\\xa9crit dans un billet publi\\xc3\\xa9 dans le journal The Nation que le gouvernement ne remplit pas le contrat social qui le lie au peuple malawite :'\n",
      " b'Bataille de Kiev.'], shape=(5,), dtype=string)\n",
      "\n",
      "tf.Tensor(\n",
      "[b'It always does seem it is impossible until it is done but if you believe in yourself and persist, you will see it come to fruition.'\n",
      " b'The foreign countries that supported the Khmer Rouge, or acted as the main catalyst for the emergence of this cruel regime, will not be brought to court.'\n",
      " b\"On the blog Tiger Beatdown, beneath Flavia's post \\xe2\\x80\\x9cIf you protest racism during Black Face season in The Netherlands, you will be beaten up and arrested\\xe2\\x80\\x9d a comment by Elfe echoes the above:\"\n",
      " b'Pointing to the clear divide between the ruling class and everyone else, blogger Alick Nyasulu wrote in a post that appeared in The Nation newspaper that the government is not fulfilling the social contract that it holds with the Malawian people:'\n",
      " b'The Kyiv Battle.'], shape=(5,), dtype=string)\n"
     ]
    }
   ],
   "source": [
    "for example_context_strings, example_target_strings in train_raw.take(1):\n",
    "  print(example_context_strings[:5])\n",
    "  print()\n",
    "  print(example_target_strings[:5])\n",
    "  break"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "zCoxLcuN3bwv"
   },
   "source": [
    "### Text preprocessing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7kwdPcHvzz_a"
   },
   "source": [
    "One of the goals of this tutorial is to build a model that can be exported as a `tf.saved_model`. To make that exported model useful it should take `tf.string` inputs, and return `tf.string` outputs: All the text processing happens inside the model. Mainly using a `layers.TextVectorization` layer."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "EOQ5n55X4uDB"
   },
   "source": [
    "#### Standardization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "upKhKAMK4zzI"
   },
   "source": [
    "The model is dealing with multilingual text with a limited vocabulary. So it will be important to standardize the input text.\n",
    "\n",
    "The first step is Unicode normalization to split accented characters and replace compatibility characters with their ASCII equivalents.\n",
    "\n",
    "The `tensorflow_text` package contains a unicode normalize operation:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:22.182025Z",
     "iopub.status.busy": "2023-12-07T12:22:22.181390Z",
     "iopub.status.idle": "2023-12-07T12:22:22.186397Z",
     "shell.execute_reply": "2023-12-07T12:22:22.185723Z"
    },
    "id": "mD0e-DWGQ2Vo"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "b'Aujourd\\xe2\\x80\\x99hui, il fait beau.'\n",
      "b'Aujourd\\xe2\\x80\\x99hui, il fait beau.'\n"
     ]
    }
   ],
   "source": [
    "example_text = tf.constant('Aujourd’hui, il fait beau.')\n",
    "\n",
    "print(example_text.numpy())\n",
    "print(tf_text.normalize_utf8(example_text, 'NFKD').numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6hTllEjK6RSo"
   },
   "source": [
    "Unicode normalization will be the first step in the text standardization function:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:22.189478Z",
     "iopub.status.busy": "2023-12-07T12:22:22.189225Z",
     "iopub.status.idle": "2023-12-07T12:22:22.193927Z",
     "shell.execute_reply": "2023-12-07T12:22:22.193252Z"
    },
    "id": "chTF5N885F0P"
   },
   "outputs": [],
   "source": [
    "def tf_lower_and_split_punct(text):\n",
    "  # Split accented characters.\n",
    "  text = tf_text.normalize_utf8(text, 'NFKD')\n",
    "  text = tf.strings.lower(text)\n",
    "  # Keep space, a to z, and select punctuation.\n",
    "  text = tf.strings.regex_replace(text, '[^ a-z.?!,¿]', '')\n",
    "  # Add spaces around punctuation.\n",
    "  text = tf.strings.regex_replace(text, '[.?!,¿]', r' \\0 ')\n",
    "  # Strip whitespace.\n",
    "  text = tf.strings.strip(text)\n",
    "\n",
    "  text = tf.strings.join(['[START]', text, '[END]'], separator=' ')\n",
    "  return text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:22.197138Z",
     "iopub.status.busy": "2023-12-07T12:22:22.196577Z",
     "iopub.status.idle": "2023-12-07T12:22:22.205704Z",
     "shell.execute_reply": "2023-12-07T12:22:22.205030Z"
    },
    "id": "UREvDg3sEKYa"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Aujourd’hui, il fait beau.\n",
      "[START] aujourdhui ,  il fait beau . [END]\n"
     ]
    }
   ],
   "source": [
    "print(example_text.numpy().decode())\n",
    "print(tf_lower_and_split_punct(example_text).numpy().decode())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4q-sKsSI7xRZ"
   },
   "source": [
    "#### Text Vectorization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6aKn8qd37abi"
   },
   "source": [
    "This standardization function will be wrapped up in a `tf.keras.layers.TextVectorization` layer which will handle the vocabulary extraction and conversion of input text to sequences of tokens."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:22.208909Z",
     "iopub.status.busy": "2023-12-07T12:22:22.208642Z",
     "iopub.status.idle": "2023-12-07T12:22:22.222918Z",
     "shell.execute_reply": "2023-12-07T12:22:22.222243Z"
    },
    "id": "eAY9k49G3jE_"
   },
   "outputs": [],
   "source": [
    "max_vocab_size = 5000\n",
    "\n",
    "context_text_processor = tf.keras.layers.TextVectorization(\n",
    "    standardize=tf_lower_and_split_punct,\n",
    "    max_tokens=max_vocab_size,\n",
    "    ragged=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7kbC6ODP8IK_"
   },
   "source": [
    "The `TextVectorization` layer and many other [Keras preprocessing layers](https://www.tensorflow.org/guide/keras/preprocessing_layers) have an `adapt` method. This method reads one epoch of the training data, and works a lot like `Model.fit`. This `adapt` method initializes the layer based on the data. Here it determines the vocabulary:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:22.226625Z",
     "iopub.status.busy": "2023-12-07T12:22:22.225976Z",
     "iopub.status.idle": "2023-12-07T12:22:24.601360Z",
     "shell.execute_reply": "2023-12-07T12:22:24.600568Z"
    },
    "id": "bmsI1Yql8FYe"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['', '[UNK]', 'de', ',', '[START]', '[END]', '.', 'la', 'a', 'le']"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "context_text_processor.adapt(train_raw.map(lambda context, target: context))\n",
    "context_text_processor.adapt(test_raw.map(lambda context, target: context))\n",
    "\n",
    "# Here are the first 10 words from the vocabulary:\n",
    "context_text_processor.get_vocabulary()[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "9kGjIFjX8_Wp"
   },
   "source": [
    "That's the Spanish `TextVectorization` layer, now build and `.adapt()` the English one:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:24.604989Z",
     "iopub.status.busy": "2023-12-07T12:22:24.604641Z",
     "iopub.status.idle": "2023-12-07T12:22:26.852853Z",
     "shell.execute_reply": "2023-12-07T12:22:26.852115Z"
    },
    "id": "jlC4xuZnKLBS"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['', '[UNK]', 'the', '[START]', '[END]', ',', '.', 'of', 'to', 'and']"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target_text_processor = tf.keras.layers.TextVectorization(\n",
    "    standardize=tf_lower_and_split_punct,\n",
    "    max_tokens=max_vocab_size,\n",
    "    ragged=True)\n",
    "\n",
    "target_text_processor.adapt(train_raw.map(lambda context, target: target))\n",
    "target_text_processor.adapt(test_raw.map(lambda context, target: target))\n",
    "\n",
    "target_text_processor.get_vocabulary()[:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BWQqlP_s9eIv"
   },
   "source": [
    "Now these layers can convert a batch of strings into a batch of token IDs:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:26.856522Z",
     "iopub.status.busy": "2023-12-07T12:22:26.856004Z",
     "iopub.status.idle": "2023-12-07T12:22:27.557144Z",
     "shell.execute_reply": "2023-12-07T12:22:27.556492Z"
    },
    "id": "9KZxj8IrNZ9S"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-23 17:55:52.567114: I external/local_tsl/tsl/platform/default/subprocess.cc:304] Start cannot spawn child process: No such file or directory\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tf.RaggedTensor [[4, 175, 325, 137, 1324, 309, 26, 14, 175, 4506, 21, 573, 3, 50, 49, 1, 13,\n",
       "  1, 11, 1, 3, 49, 1, 10, 510, 6, 5]                                        ,\n",
       " [4, 10, 45, 624, 18, 27, 1840, 10, 1, 1979, 3, 36, 33, 10, 1, 1013, 2, 1,\n",
       "  2, 26, 339, 1, 3, 29, 508, 23, 1, 13, 301, 6, 5]                        ,\n",
       " [4, 2, 190, 235, 3, 32, 2115, 58, 694, 16, 327, 1131, 18, 1, 23, 2, 1, 3,\n",
       "  12, 1, 36, 1, 2126, 12, 2022, 20, 9, 71, 6, 5]                          ]>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "example_tokens = context_text_processor(example_context_strings)\n",
    "example_tokens[:3, :]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "AA9rUn9G9n78"
   },
   "source": [
    "The `get_vocabulary` method can be used to convert token IDs back to text:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "target_vocab = np.array(target_text_processor.get_vocabulary())\n",
    "context_vocab = np.array(context_text_processor.get_vocabulary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:27.560512Z",
     "iopub.status.busy": "2023-12-07T12:22:27.560241Z",
     "iopub.status.idle": "2023-12-07T12:22:27.577208Z",
     "shell.execute_reply": "2023-12-07T12:22:27.576612Z"
    },
    "id": "98g9rcxGQY0I"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'[START] ca semble toujours impossible jusqua ce que ca devienne une realite , si vous [UNK] en [UNK] et [UNK] , vous [UNK] les resultats . [END]'"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tokens = context_vocab[example_tokens[0].numpy()]\n",
    "' '.join(tokens)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Ot0aCL9t-Ghi"
   },
   "source": [
    "The returned token IDs are zero-padded. This can easily be turned into a mask:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:27.580623Z",
     "iopub.status.busy": "2023-12-07T12:22:27.580079Z",
     "iopub.status.idle": "2023-12-07T12:22:27.907741Z",
     "shell.execute_reply": "2023-12-07T12:22:27.906943Z"
    },
    "id": "_jx4Or_eFRSz"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Mask')"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAigAAAGzCAYAAAAFROyYAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAABBKUlEQVR4nO3de3xU1bn/8e9MhlwgN64JEYggCIiCGBWiWCmmIiKiYotoW2yt9hKsgD1WWpXCUUFrRa0KrbVwbH+UFnuASouWokCtgBCgXlAERUBCgnJJuOU2s39/eJi6Zg3OTJxk9oTP+/Wa18u15pm9nwll9WFnPXt7HMdxBAAA4CLeRCcAAAAQigIFAAC4DgUKAABwHQoUAADgOhQoAADAdShQAACA61CgAAAA16FAAQAArkOBAgAAXIcCBZIkj8ejCRMmJDoNAIjJypUr5fF49Pzzzyc6FcQZBUoS83g8Ub1WrlyZ6FRjMnToUJ199tnG3Omnnx78Pl6vV7m5uTrnnHN02223ad26dQnKFMC8efOCfzdfffVV633HcdS1a1d5PB5dddVVCcgQycqX6ATQeL/73e+M8XPPPafly5db83379m3OtJrMueeeqzvvvFOSdPjwYb3zzjtauHChnnnmGU2aNEmPPvpogjMETl3p6emaP3++hgwZYsyvWrVKH330kdLS0hKUGZIVBUoS+/rXv26M165dq+XLl1vzLcVpp51mfbeHHnpIN954o2bNmqVevXrp+9//foKyA05tV155pRYuXKgnnnhCPt9//q9l/vz5Kioq0ieffJLA7JCM+BVPC3f06FHdeeed6tq1q9LS0tS7d2898sgjiuYh1vfff7+8Xq9++ctfBueWLVumSy65RG3atFFWVpZGjhypt99+2/jczTffrMzMTO3Zs0fXXHONMjMz1bFjR/3oRz+S3++P6/fLyMjQ7373O7Vr104PPPCA8b0WLFigoqIiZWVlKTs7W+ecc44ef/zxuJ4fwKfGjRun/fv3a/ny5cG5uro6Pf/887rxxhut+EceeUQXXXSR2rdvr4yMDBUVFYXdR7J8+XINGTJEubm5yszMVO/evfWTn/zkc3Opra3VVVddpZycHL322mtf/MshIShQWjDHcXT11Vdr1qxZuuKKK/Too4+qd+/e+q//+i9Nnjz5cz97zz336L777tOvfvUr3X777ZI+/ZXSyJEjlZmZqYceekj33nuvtmzZoiFDhujDDz80Pu/3+zV8+HC1b99ejzzyiC699FL94he/0K9//eu4f8/MzExde+212rNnj7Zs2SLp00Vt3Lhxatu2rR566CHNnDlTQ4cO1b/+9a+4nx/Ap/vEiouL9Yc//CE4t2zZMlVVVemGG26w4h9//HENHDhQ06dP14MPPiifz6evfvWr+utf/xqMefvtt3XVVVeptrZW06dP1y9+8QtdffXVn/v3+Pjx4xo1apRee+01/eMf/9BFF10U3y+K5uOgxSgtLXU++0e6ePFiR5Jz//33G3HXX3+94/F4nO3btwfnJDmlpaWO4zjOnXfe6Xi9XmfevHnB9w8fPuzk5uY6t956q3GsiooKJycnx5gfP368I8mZPn26ETtw4ECnqKgo4ve49NJLnX79+hlzhYWFzsiRI0/6mVmzZjmSnCVLljiO4zh33HGHk52d7TQ0NEQ8H4DGmzt3riPJWb9+vfPkk086WVlZzrFjxxzHcZyvfvWrzpe//GXHcey/wydiTqirq3POPvtsZ9iwYcG5E3+vP/7445Oe/5VXXnEkOQsXLnQOHz7sXHrppU6HDh2cTZs2xfFbIhG4gtKC/e1vf1NKSop++MMfGvN33nmnHMfRsmXLjHnHcTRhwgQ9/vjj+v3vf6/x48cH31u+fLkOHTqkcePG6ZNPPgm+UlJSNGjQIL3yyivW+b/3ve8Z40suuUQffPBBHL/hf2RmZkr6dPOsJOXm5uro0aPG5WYATetrX/uajh8/rqVLl+rw4cNaunRp2F/vSJ/+evaEgwcPqqqqSpdccok2btwYnM/NzZUkLVmyRIFA4HPPXVVVpcsvv1zvvvuuVq5cqXPPPfcLfx8kFptkW7CdO3eqoKBAWVlZxvyJrp6dO3ca888995yOHDmi2bNna9y4ccZ727ZtkyQNGzYs7Lmys7ONcXp6ujp27GjMtW3bVgcPHoz9i0ThyJEjkhT8rj/4wQ/0pz/9SSNGjNBpp52myy+/XF/72td0xRVXNMn5AUgdO3ZUSUmJ5s+fr2PHjsnv9+v6668PG7t06VLdf//92rx5s2pra4PzHo8n+N9jx47Vb37zG33nO9/R3Xffrcsuu0zXXXedrr/+enm95r+vJ06cqJqaGm3atEn9+vVrmi+IZsUVFARdfPHFysvL05NPPqkDBw4Y753418vvfvc7LV++3HotWbLEiE9JSWm2vCXprbfekiT17NlTktSpUydt3rxZf/nLX3T11VfrlVde0YgRI4yrQgDi78Ybb9SyZcs0Z84cjRgxIngV5LP++c9/6uqrr1Z6erqefvpp/e1vf9Py5ct14403GhvdMzIytHr1av3jH//QN77xDb3xxhsaO3asvvKVr1gb7kePHi3HcTRz5syIV1uQHChQWrDCwkKVl5cHf+1xwrvvvht8/7N69uypv//97yovL9cVV1xhfO6MM86Q9On/8ZeUlFivoUOHNu2X+RxHjhzRokWL1LVrV+OeL6mpqRo1apSefvppvf/++/rud7+r5557Ttu3b09YrkBLd+2118rr9Wrt2rUn/fXOn//8Z6Wnp+ull17St7/9bY0YMUIlJSVhY71ery677DI9+uij2rJlix544AG9/PLL1q+Vr7nmGv32t7/V/PnzVVpaGvfvheZHgdKCXXnllfL7/XryySeN+VmzZsnj8WjEiBHWZ/r376+//e1veueddzRq1CgdP35ckjR8+HBlZ2frwQcfVH19vfW5jz/+uGm+RATHjx/XN77xDR04cEA//elPg5eH9+/fb8R5vV71799fkozLyQDiKzMzU7Nnz9bPfvYzjRo1KmxMSkqKPB6PcRXkww8/1OLFi4240Cu5koJ7S8L9Pf7mN7+pJ554QnPmzNGPf/zjxn8JuAJ7UFqwUaNG6ctf/rJ++tOf6sMPP9SAAQP097//XUuWLNHEiRODV0VCDR48WEuWLNGVV16p66+/XosXL1Z2drZmz56tb3zjGzrvvPN0ww03qGPHjtq1a5f++te/6uKLL7YKoXjbs2ePfv/730v69KrJli1btHDhQlVUVOjOO+/Ud7/73WDsd77zHR04cEDDhg1Tly5dtHPnTv3yl7/Uueee22LurAu4VaRfpY4cOVKPPvqorrjiCt14443at2+fnnrqKfXs2VNvvPFGMG769OlavXq1Ro4cqcLCQu3bt09PP/20unTpYt2x9oQJEyaourpaP/3pT5WTkxPxnilwscQ2ESGeQtuMHefT9uBJkyY5BQUFTqtWrZxevXo5P//5z51AIGDE6TNtxicsWbLE8fl8ztixYx2/3+84zqctfcOHD3dycnKc9PR054wzznBuvvlmZ8OGDcHPjR8/3mnTpo2V39SpU638wjlZm7EkR5Lj8Xic7Oxsp1+/fs6tt97qrFu3zjrG888/71x++eVOp06dnNTUVKdbt27Od7/7XWfv3r0Rzw8gep9tM/48oW3Gzz77rNOrVy8nLS3N6dOnjzN37lxrjVixYoUzevRop6CgwElNTXUKCgqccePGOe+9914w5rNtxp911113OZKcJ598Mk7fFM3N4zhR3FIUAACgGbEHBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHUoUAAAgOtQoAAAANdx3Y3aAoGAysvLlZWVZTw0CkDzcRxHhw8fVkFBgfVQNrdi7QASK97rhusKlPLycnXt2jXRaQCQtHv3bnXp0iXRaUSFtQNwh3itG64rULKysiRJQ3SlfGoVnPemphpxgbq6iMc68K1B1ly7ueuMsa9HoRXTsGO3Mf5gZpEV0+PH6yOeP1Tg4v4RY7z/eiNiDNDUGlSvV/W34N/HZHAi150bT1d2ZnJc9Wlq1555TqJTwCkk3uuG6wqUE5dmfWoln+czBcpn/luSAp7IN8BNSU235nwhx/F508IkYcZ4MyIfJxoBn32cUKHfE0iI//vrlUy/KjmRa3amV9lZKQnOxh0as04BjRbndYN/ZgAAANehQAEAAK7jul/xnEw0e05Ctf/1axFjGt7fYc15W7c2xj0nvh7zuSUppd+Zxnjbjfbl1l7fa9yxASCSl8r/bYyHFwxIUCZA7LiCAgAAXIcCBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHWSpounqXjP62fNeT4sNyeOHbNjBpl3hXXW2XeAPTKr3hj3S91txcTemwQANjp00NJwBQUAALgOBQoAAHAdChQAAOA6FCgAAMB1KFAAAIDrJE0Xjzc11RiHezbP+78faIzP+Ma/rRg5AWN4vKCNFZK28aAxTsnNtWL8Ybp2QrUZtdcY19XWWDG+zvnGOHCoyor56/vmM4VGFAy0YsKpXNLXGOeNfieqzwFIPqHP3WludBEh3riCAgAAXCfmAmXPnj36+te/rvbt2ysjI0PnnHOONmzYEHzfcRzdd9996ty5szIyMlRSUqJt27bFNWkAyYe1A0AsYipQDh48qIsvvlitWrXSsmXLtGXLFv3iF79Q27ZtgzEPP/ywnnjiCc2ZM0fr1q1TmzZtNHz4cNXU2L/eAHBqYO0AEKuY9qA89NBD6tq1q+bOnRuc6969e/C/HcfRY489pnvuuUejR4+WJD333HPKy8vT4sWLdcMNN8QpbQDJhLUDQKxiuoLyl7/8Reeff76++tWvqlOnTho4cKCeeeaZ4Ps7duxQRUWFSkpKgnM5OTkaNGiQ1qxZE/aYtbW1qq6uNl4AWhbWDgCxiukKygcffKDZs2dr8uTJ+slPfqL169frhz/8oVJTUzV+/HhVVFRIkvLy8ozP5eXlBd8LNWPGDE2bNi3iucN17YQ64+ubovgWpvRlZdac4zHrNv+hQzEfV5ICYbp2QjXsNX8u6avyrZjQrp29d15kxXT+xWvWXKO6djyRa1Zf1wJrruLKbsa4wxw7H5y6Erl2oHk0ZxcRHUOnhpiuoAQCAZ133nl68MEHNXDgQN1222269dZbNWfOnEYnMGXKFFVVVQVfu3fbD9QDkNxYOwDEKqYCpXPnzjrrrLOMub59+2rXrl2SpPz8T//1X1lZacRUVlYG3wuVlpam7Oxs4wWgZWHtABCrmAqUiy++WFu3bjXm3nvvPRUWFkr6dNNbfn6+VqxYEXy/urpa69atU3FxcRzSBZCMWDsAxCqmPSiTJk3SRRddpAcffFBf+9rX9Prrr+vXv/61fv3rX0uSPB6PJk6cqPvvv1+9evVS9+7dde+996qgoEDXXHNNU+QPIAmwdgCIVUwFygUXXKBFixZpypQpmj59urp3767HHntMN910UzDmrrvu0tGjR3Xbbbfp0KFDGjJkiF588UWlp6fHPXkAyYG1A0CsPI7jOIlO4rOqq6uVk5OjoRotn6dVcN6TkmIGDjxLobx1DebErr1WTGhHTugzfqToOoaAlqzBqddKLVFVVVXS7O04sXYcfK+HsrNSIn8gAeg+QUsW73WDZ/EAAADXoUABAACuQ4ECAABchwIFAAC4TkxdPIkUukk2sOFNK8Y5r58ZE8Ut6sNuiA253buvQzsrpOGTA8Y4pU1rK+adx/oa4zO/sz5iPgBarqa8HTwbcNHScAUFAAC4DgUKAABwHQoUAADgOhQoAADAdShQAACA6yRNF48TCLkjv8eurXZcl2OMCzfax/nopxcZ4y4PvBbuZMbQf7DKCjl23YXGuPWf11oxdO0AiAc6dHAq4goKAABwHQoUAADgOhQoAADAdShQAACA61CgAAAA10meLh6/P2QiYMUU3hOmIydE2K6diCe3zxXateNNS7diQp/z4+taYMXcvfKvxvj+HvZu/SMvnmGMM694/+S5AmhxmvIZPqHoGIJbcAUFAAC4DgUKAABwHQoUAADgOhQoAADAdShQAACA6yRNF0+4TppQKWf3NsaVl7S3YjrMNrt4UrKyrJhdE84xxqfNiNz5E6itseY8F5rHcY7VWTH3nzEw9EhWjNW1E+Y5ROF+Pn/Zs8EYX9P9IismUN9gjGv/3s2KSfvKh/b5Qnh8rcx0GuqtGF9eJ2N8ZHB3KyZ9yTrzM926WDEfPNrWGHf76tt2QlH87yU0n4bKfRE/E44nJcU8dWjHGZBEmrNjCJ/vVO+o4goKAABwHQoUAADgOhQoAADAdShQAACA6yTPJtkQ+2+1N3y2f8bczNrhrcjH8bxgb5I97cvmbeyzX+1oxVRfst+cCLMp03n9TWMct62TUWwAlaSf7zc36Ybeej+caDbEhhNuU2yo0E2o6Usib0pt2PWRNdftenuuMRq7KTYUm2IB26m+wRNfHFdQAACA61CgAAAA16FAAQAArkOBAgAAXIcCBQAAuE7SdvG03Xo8Yow3Ld2as25JP6bW/mBIl0z1kI8jnmv3VLurqNv95m3bnYAT8VzhpOTmGuNAz672YTa8ac39s39axGMDQCR05CARuIICAABchwIFAAC4DgUKAABwHQoUAADgOhQoAADAdZKni8dj1lLe1ZsifsTq2AknN9uaOv6lnsY4Y9E6KyZU4V8O2edvxDNajn51sDWXs7HSGN/7p+esmOk9BsZ8rrA8ds3qK8g3xg17yuNzLgBJ4aXyfzfbuegYwglcQQEAAK5DgQIAAFyHAgUAALgOBQoAAHAdChQAAOA6SdPF4zuzhzF29lRYMUf+t6MxrlmQb8W0/e0aY9zw/g4rJiPMXChvRoYxDmzaYsUc/HaxMW6/wN4J79SazwJqs3CtFdMQMo5bx044YZ4NRNcOkJzoiEEy4woKAABwnZgKlJ/97GfyeDzGq0+fPsH3a2pqVFpaqvbt2yszM1NjxoxRZWXl5xwRwKmAtQNArGK+gtKvXz/t3bs3+Hr11VeD702aNEkvvPCCFi5cqFWrVqm8vFzXXXddXBMGkJxYOwDEIuY9KD6fT/n59t6OqqoqPfvss5o/f76GDRsmSZo7d6769u2rtWvXavBg+w6pAE4drB0AYhHzFZRt27apoKBAPXr00E033aRdu3ZJksrKylRfX6+SkpJgbJ8+fdStWzetWbPmZIdTbW2tqqurjReAloe1A0AsYrqCMmjQIM2bN0+9e/fW3r17NW3aNF1yySV66623VFFRodTUVOXm5hqfycvLU0WF3XFzwowZMzRt2rSI527Yut1MvGMHKybjcrP7JkN2N45vZYF53KGN61AJHD8eMSa0Y+jgN4utmJznQhbgMM/CSTnLfDaQ/+33rBhfty7WXMOujyLm2BielBRrbsf0Qcb49J++FpdzpZzd25oLvPuBMf5w6gVWTOG98Tk/4iORa8eprDmfoYP/oHsqPmIqUEaMGBH87/79+2vQoEEqLCzUn/70J2WEtN1Ga8qUKZo8eXJwXF1dra5duzbqWADcibUDQKy+UJtxbm6uzjzzTG3fvl35+fmqq6vToUOHjJjKysqwv3c+IS0tTdnZ2cYLQMvG2gEgki9UoBw5ckTvv/++OnfurKKiIrVq1UorVqwIvr9161bt2rVLxcX2rzYAnLpYOwBEEtOveH70ox9p1KhRKiwsVHl5uaZOnaqUlBSNGzdOOTk5uuWWWzR58mS1a9dO2dnZuv3221VcXMwufOAUx9oBIFYxFSgfffSRxo0bp/3796tjx44aMmSI1q5dq44dP73F/KxZs+T1ejVmzBjV1tZq+PDhevrpp+OSqDPkXGPc8OpmKyb01vKhm1QlqeFu83b43rQDVkygtsaMad3aivGfd6Yx9oTJx5uWboytDbFhHFnW3ZrLHLE9TKRp59e7WXOFvzI38gaOHLVijl5pbuZq/ZcNVkxKV3MDbsOHO62YeG2KDeV/a2vEGDbEul8i1w4kFzaY4oSYCpQFCxZ87vvp6el66qmn9NRTT32hpAC0LKwdAGLFs3gAAIDrUKAAAADXoUABAACuQ4ECAABcJ+aHBSZKaJeMr8tpVkyHxe8aY3+Y28bvH9DGGLdfW2PFeEPubBk4bsekvL7FGB8ZY7dDevyOMW69dKMVc+gvp5v5fOOgFdNgzdi6/P2wNdf9pSPGeNv5tVZMxqJ1xtixIsJ37URy+Eb7/hVZ8yN3MQE4tTXm9vx0/rRMXEEBAACuQ4ECAABchwIFAAC4DgUKAABwHQoUAADgOknTxaOQjpztPyi0Qk7/ScgzWcJ08bT/VeTntnz0w4HG+LRfrLdiAnV1xjjrH+9YMf6qKnN86XlWTM6VZmdPQ5ic5QSMoXfAWXY+G9605radbx+qudCxA7QcdMkgEbiCAgAAXIcCBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHWSp4snhNWxIynwJbP7xrt6kxXjSUkxxo7fb8VYXTshXTRhdc2350K6eLyrN0c+ThQ8H+wOM2nXmqHfNZyRb+wzxkv7tbeDQr7/tjkXWiF97/3AGDd8/EnkHKP5uUbR1RSNuhEXWHOpy+zuLAC2xjwf51RG11N8cAUFAAC4DgUKAABwHQoUAADgOhQoAADAdZJmk6w3Pc0YB44ft2L2/5c513G1fRzngrPNibVhNn9FsQnTV9DZGDe8vc2K+eCRYmPc40dR3P49inP7Dx+OfBxJTkPkYy09q23ITOTP9Pre69ZcQ1QJxb65tVGfCYMNsWjp2JiJloYrKAAAwHUoUAAAgOtQoAAAANehQAEAAK5DgQIAAFwnebp4Opq3YA/sLrdiOo561xj7evawYhrCde2ECL39fWjHjiQ1lO81Y04vtGLO+LHZ7eJEPDMARIeuHbR0XEEBAACuQ4ECAABchwIFAAC4DgUKAABwHQoUAADgOknTxdOw6yNj7Ovd047Zut0cb/8gLud2qsM8+8Zj1naBfR/bIWnm84Oc4zVWjK+wqzmRnmrFNGx93xh7z+1j5/P2dmsqUFdnx4V4qdzsaqIzoGnM/+g1Y3xjl4sSlAlaitC/u0CiVR/2q+2Z8TseV1AAAIDrUKAAAADXoUABAACuQ4ECAABchwIFAAC4TtJ08YRyUu3UvWnp5sRZ9rN4vAfMjpyGXXusmJS2OcbYf+CgFXNk7GBjnPnHtSfN9YSj1w+25to8b37O16mj/UEnYAy9R2qtkIYwHTuVd5idInmPv2bFNKprxxOmrg3J0ZOSYn8s1exQChw/Hvu5w0jJzLTm/EePmRMh+TU3unaAU8ep2g3Z4NRLik/3rMQVFAAA4EIUKAAAwHUoUAAAgOtQoAAAANehQAEAAK6TtF08/jffteZCO0ecTVusmNBeju3PnWfF9PzmRmPsO63Aisn6c5mZz6X2cfYMNbuKuk6zu2hCVV9yhjXXZvEBc6LW7uIJ11kT2rWze6rdSVI40/yu+75lf48Oc0LyjqIjxvH77bk4de2E8h850iTHBRBfp2p3CxrnC11BmTlzpjwejyZOnBicq6mpUWlpqdq3b6/MzEyNGTNGlZWVXzRPAC0E6waAaDS6QFm/fr1+9atfqX///sb8pEmT9MILL2jhwoVatWqVysvLdd11133hRAEkP9YNANFqVIFy5MgR3XTTTXrmmWfUtm3b4HxVVZWeffZZPfrooxo2bJiKioo0d+5cvfbaa1q7NvKNzAC0XKwbAGLRqAKltLRUI0eOVElJiTFfVlam+vp6Y75Pnz7q1q2b1qxZE/ZYtbW1qq6uNl4AWp54rhsSawfQ0sW8SXbBggXauHGj1q9fb71XUVGh1NRU5ebmGvN5eXmqqKgIe7wZM2Zo2rRpsaYRVt/XPcZ4S1Hkz/QcvzliTMOecnsyZFNq6p5DVkjXaeYtf7fNsxPqdbO52Va37bNinD+bG07fm2nfDr/HjR9Zcykhfw6BfvZm0kBtjTG2NsQCcRDvdUOK79oBwH1iuoKye/du3XHHHfp//+//KT09PfIHojBlyhRVVVUFX7t3747LcQG4Q1OsGxJrB9DSxVSglJWVad++fTrvvPPk8/nk8/m0atUqPfHEE/L5fMrLy1NdXZ0OHTpkfK6yslL5+flhj5mWlqbs7GzjBaDlaIp1Q2LtAFq6mH7Fc9lll+nNN9805r71rW+pT58++vGPf6yuXbuqVatWWrFihcaMGSNJ2rp1q3bt2qXi4uL4ZQ0gabBuAGiMmAqUrKwsnX322cZcmzZt1L59++D8LbfcosmTJ6tdu3bKzs7W7bffruLiYg0ePDh+WQNIGqwbABoj7neSnTVrlrxer8aMGaPa2loNHz5cTz/9dLxPA6AFYd0AEMrjOI6T6CQ+q7q6Wjk5ORqq0fJ5WgXnU9q1NeI8IWNJKv7fd4zxP/unWTHW7fADYb5+FLdyD+3iqRl1vhXS5tVtxth/sMqKOXr9heZnFnLfByReg1OvlVqiqqqqpNnbcWLtOPheD2VnpUT+AMLidvRorHivGzwsEAAAuA4FCgAAcB0KFAAA4DoUKAAAwHUoUAAAgOvEvc24qfgPHDTGu37dxY4Z2NoYX/LGMSvmHz+5xBinLX29UfnUXWE+Vyf9L/Zxjo0yO3TSXrBj6NoB0Fzo0EEy4QoKAABwHQoUAADgOhQoAADAdShQAACA61CgAAAA10maLh6Pr5Ux7nb9m3ZMRoYxXl3a34pJ+6fZSVNz9YVWTGhHTsrZva2Y1GXrT57siXOFdO14U1OtmEB9Q8TjRPVsoCjUXmV/V6uLyWPXrLUjzecMhet8sp5x5PdHzCf0+UqSFKg+Yozrv2x3HfiO1JsTa+3/LVji9DMEktlL5f9OdAotEt1RTYMrKAAAwHUoUAAAgOtQoAAAANehQAEAAK6TNJtko9l0GaipNcb12a2smNBtqukvbIh87lYpEWO8IRt0JSlw/Lg5DrMhNiU7yxj7qw9bMdN2lBnj6edcaucY5tiBujpjfKSz/cedFmZTrBUTsik2dEOsJHmzQr7HoUMRj+scr7HnQv6cfcvtPx9vWrqZT8jPUJL8VVURzw8AsWJDbPPhCgoAAHAdChQAAOA6FCgAAMB1KFAAAIDrUKAAAADXSZounqhuVR4Sk758sx1TdLYxDGzcYoX4Tiswxg2b37WPE9L9cuBr51oh7f/3LWPsP2x36ETTbTK1e1HIzJGwcZG0f+a1Rn0uVLiOqmi6dkKFdjlF/bnakO6f0DFwCqK7BC0NV1AAAIDrUKAAAADXoUABAACuQ4ECAABchwIFAAC4TvJ08UThyNjBxjjzT69bMfvPyzbG7cvs7qCGPeXG+KM/n23FZL5gPv9lyX8/YsUMPeO/jHHhffHpoknJybHmwj3DJ5rOp30TLjLGnZ6MnON7z1xgzZ156/qInwPQdF4q/3eiU4gLupFwAldQAACA61CgAAAA16FAAQAArkOBAgAAXIcCBQAAuE6L6uLJWb7VGPvDdLHk/a8Z02ltthVTPrjaGHcZ85YVE/jSQGM8vuvFVkyhZ60x/uiei6yYLg+YMdF03oR7fo93wFn2XPk+Y9zw8SdWTDRdO6Ho2AEQiu4bxBtXUAAAgOtQoAAAANehQAEAAK5DgQIAAFwnaTfJetPSrTn/gYPG+Pi1g6yYrH99YIw3LLJvY18gc+Oor6CzFdOwelNUeX5Wt0c3W3OO12OO/fbnfHmdzHNX7rODwgj9eTSax6xjvef2sUICm7bEfJxwfJ3zjHHD3korJqVNa2PsP3LEzmdFV2PsvWy3nc6g/sbYWfdGxPwAhNdSbrXfXNhUHBlXUAAAgOtQoAAAANehQAEAAK5DgQIAAFyHAgUAALhO0nTxpGRlGWOnrt4OCukSafOi3ZUR6NvDGBc8ZN/q/cAt5i3p2z0b5nbwoR0p4W5RHzIXLufL3zA7bV7qZ9963//JfvvYIQL/jqKLprFCvkdUHTtRHCechvK9EWPCde2ECte1Y6VD1w5ANwlciysoAADAdWIqUGbPnq3+/fsrOztb2dnZKi4u1rJly4Lv19TUqLS0VO3bt1dmZqbGjBmjykr7PhYATi2sHQBiFVOB0qVLF82cOVNlZWXasGGDhg0bptGjR+vtt9+WJE2aNEkvvPCCFi5cqFWrVqm8vFzXXXddkyQOIHmwdgCIVUx7UEaNGmWMH3jgAc2ePVtr165Vly5d9Oyzz2r+/PkaNmyYJGnu3Lnq27ev1q5dq8GDB8cvawBJhbUDQKwavQfF7/drwYIFOnr0qIqLi1VWVqb6+nqVlJQEY/r06aNu3bppzZo1Jz1ObW2tqqurjReAlou1A0A0Yu7iefPNN1VcXKyamhplZmZq0aJFOuuss7R582alpqYqNzfXiM/Ly1NFRcVJjzdjxgxNmzYt4nn9hw8b45T+fe2gN94xhjXDzrFCUv/6esRztf+f9cbYCRNzfPQFxjhj8TorZve9ZjdQ99/tsWJWjDCfPZNyTqYV43/rvZOl+p/Phfl5+EN+HuF4MzKMceD4cSvG16mjGdO5oxUTTRdRSr8zzfzetr+Xr08vY7zzGvtcp80M01UVIvR7edu1tWIa9pRHPA7iJ1FrBz4fz9CJPzqj4iPmKyi9e/fW5s2btW7dOn3/+9/X+PHjtWVL41tcp0yZoqqqquBr9+7I7aEAkg9rB4BYxHwFJTU1VT179pQkFRUVaf369Xr88cc1duxY1dXV6dChQ8a/hCorK5Wfn3/S46WlpSktLS32zAEkFdYOALH4wvdBCQQCqq2tVVFRkVq1aqUVK1YE39u6dat27dql4uLiL3oaAC0MaweAzxPTFZQpU6ZoxIgR6tatmw4fPqz58+dr5cqVeumll5STk6NbbrlFkydPVrt27ZSdna3bb79dxcXF7MIHTnGsHQBiFVOBsm/fPn3zm9/U3r17lZOTo/79++ull17SV77yFUnSrFmz5PV6NWbMGNXW1mr48OF6+umnmyRxAMmDtQNArDyO44RrUkmY6upq5eTkaKhGy+dp9Z83Qp594zuts/XZho/MLhlfrzOsmPo885k+nlc3Nz5ZoIVqcOq1UktUVVWl7Gz7+VBudGLtOPheD2VnpSQ6najR8YGWIt7rBs/iAQAArkOBAgAAXIcCBQAAuA4FCgAAcB0KFAAA4Dox30nWLUI7diRZnT7lV+RZIZ1+Gfk5LnETko+cQMQYX4d2VkjDx5/Efq6TnS+CbU8OsubO/J35fB5n3Rv26VPMrgnH77cPHsXPI5pnA4XyFdgdXVvvKjTGZ0xca8Xsm2A+K6ngj/azgaL52adkhXSGtQ/zZ/jhTjMmxe4ySSnsaoz9He1d8NbPPsyfe+ixU9rm2PlE878pNAuehYOWovqwX23PjBwXLa6gAAAA16FAAQAArkOBAgAAXIcCBQAAuE7y3Oq+Ee77YJM1N73HQHMizCbDlDatjbH/yBErxtezhzFu2P6BFVMxydyEeeaYbVbM/v8+3Ri3emm9FWOJ04ZY4GS41T1ixS37wa3uAQBAi0eBAgAAXIcCBQAAuA4FCgAAcB0KFAAA4DrJc6v7cJ0rIbytzK/z4JeuChMV5hb5oXzmcXY/f44V0vX6N82JMPmd9j/vGOPqWQetmFb6OHI+oejYARAFOmuQzLiCAgAAXIcCBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHWSp4snxP/s+qc1N77rxebEx/sjHyhMR4xTW2uMC2+yn6ETiKKryH8gpGsn3HN/srPMz1QftmMy2xjj3d+3u4oKHn4tYj4f/tHe0d/9xrcifs4JmI9r2v7YhVZMzz/WmBOvbY54XABN66Xyfyc6hWZBt1LLxBUUAADgOhQoAADAdShQAACA61CgAAAA16FAAQAArpM0XTzejHRjPL7bJXaQxxw6fr8Vsm/CRcY4b/a6MMcxDxSoPW6HpKQY46xV7ayY6iEhz9kJ9wydFLNGDH2ekCT5D5udPTk7GvcsHv9HrSPGhPuZhep5x9pGnR/AqYPOGnxRXEEBAACuQ4ECAABchwIFAAC4DgUKAABwnaTZJHv8srONcX1ru7bK/KO5efPD+wbbMR+Z43CbQp1jx8yJMLeor7myyPzMkNetmNDP7VtyphXS6ep3jXHtKPs28mkvmMdus7Bxm1TPuHONNeeEiQNw6mAzK9yKKygAAMB1KFAAAIDrUKAAAADXoUABAACuQ4ECAABcJ2m6eEI7WZbvKbNihqRPMMY9Fh6wDxQwbxMf7sbuvl5nGOOGbe9HzCec0Nvhd77xIysm9PytPzhkHygryxge/1JfKyT1r5HzuWrLQWtuab/25kSY2/F7W5u3yA+EdjkBSFovlf870Sl8LrqMTl1cQQEAAK5DgQIAAFyHAgUAALgOBQoAAHAdChQAAOA6SdPF401NNcZXnVZkxeR61hnjD/7bfhZP4T2vRTxXw/Yd5kSYZ/H4Cruan/lwpxXjNNQbY/+ReismlP/t9yLGRNOxE87Ss9qGmbW7dqwIunYA16G7BS0dV1AAAIDrxFSgzJgxQxdccIGysrLUqVMnXXPNNdq6dasRU1NTo9LSUrVv316ZmZkaM2aMKisr45o0gOTC2gEgVjEVKKtWrVJpaanWrl2r5cuXq76+XpdffrmOHj0ajJk0aZJeeOEFLVy4UKtWrVJ5ebmuu+66uCcOIHmwdgCIVUx7UF588UVjPG/ePHXq1EllZWX60pe+pKqqKj377LOaP3++hg0bJkmaO3eu+vbtq7Vr12rwYHtPCICWj7UDQKy+0B6UqqoqSVK7du0kSWVlZaqvr1dJSUkwpk+fPurWrZvWrFkT9hi1tbWqrq42XgBaNtYOAJE0uosnEAho4sSJuvjii3X22WdLkioqKpSamqrc3FwjNi8vTxUVFWGPM2PGDE2bNi3y+erqzPHQ86wY78qNxrj7g5utmIofXmTm9tQ6Kyb0GTqh55akY307GePUMF08Vn4hz7SR7A6ZlL5nWjH+d0I6e4rD7N53wpxwbeRnbOxf2tsYt78/3YrxrH/LzDHkz1eSGvbvNz8T8jOUpJTTOpuf2WU/mwgtX3OvHS2V25+hk2h0OSW/Rl9BKS0t1VtvvaUFCxZ8oQSmTJmiqqqq4Gv37t1f6HgA3I21A0A0GnUFZcKECVq6dKlWr16tLl26BOfz8/NVV1enQ4cOGf8SqqysVH5+fthjpaWlKS0trTFpAEgyrB0AohXTFRTHcTRhwgQtWrRIL7/8srp37268X1RUpFatWmnFihXBua1bt2rXrl0qLi6OT8YAkg5rB4BYxXQFpbS0VPPnz9eSJUuUlZUV/N1wTk6OMjIylJOTo1tuuUWTJ09Wu3btlJ2drdtvv13FxcXswgdOYawdAGIVU4Eye/ZsSdLQoUON+blz5+rmm2+WJM2aNUter1djxoxRbW2thg8frqeffjouyQJITqwdAGLlcRwnXP9HwlRXVysnJ0dDNVo+T6uTxoU+C0eSGnbGvklu/60XWXMd5q43ximnhznX9g+M8dHr7X/ltXl+bcz5+Drbv29v2Gt2MaTk5ER1LP//tXICsWpw6rVSS1RVVaXs7OxEpxOVE2vHwfd6KDvL7iJD/NAhg3DivW7wLB4AAOA6FCgAAMB1KFAAAIDrUKAAAADXafSt7hOtMRtiw2n/G3sjqyfdvPlT6IbYcMJtiA3dzBrNptXQDbGSNHCTOd40sOk2v4a7RX3OqrbG+NCQTyJ+zvH745sYANdw22322bTbMnEFBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHUoUAAAgOskbRfPVVsOWnNLz2obJtLk69TRGDfs+9iKCRw/boy9aelWTMV3zzPGnZ54zYqJ163mNw2My2GiEq77JlzXTjSfAxA7OlKAT3EFBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHUoUAAAgOskbRdPNB07788qtubOmLTGGC/4aI0Vc0PXi41xoLbGignXtRNq++ODjfGZU96wYj6861xjnGE3FanjU5HPFY71fJyAY8d4PZ/7GUmqv/gcY5zySpkVc3ic+V2z/hDm2URZWcbYf/iwFfPJ9y8yxh1m2989mk6s0M6rcH+GgBu57Tk3iIzOq6bBFRQAAOA6FCgAAMB1KFAAAIDrUKAAAADXoUABAACu43Ecx27tSKDq6mrl5ORoqEbL52l18kCPXVvVXVFkjFOXrbdijo0xu01a/9nuNomGr7CrMW7YubtRxwn9Ht4M+7k/3sw25rnCdK0A8dTg1GullqiqqkrZ2dmJTicqJ9aOg+/1UHaW3Y2G6NCRgsaK97rBFRQAAOA6FCgAAMB1KFAAAIDrUKAAAADXSdpb3fvahbnVfcimWG9GhhUSzaZYb2qqMQ7U1VkxjdoUG2Zj78CNAWO8aeAx+2OtPmezMADEUTS32mcjLZoDV1AAAIDrUKAAAADXoUABAACuQ4ECAABchwIFAAC4TtJ28Sgr05ryeT3G2Dle06hDO4HY7/6/Y+ZF1lz3u18LPbAVs2lg5GP7q6pizgcAmko0nT7RoBsIn4crKAAAwHUoUAAAgOtQoAAAANehQAEAAK5DgQIAAFwnabt4dn3tNGuu4OHXwkSGCH0eTpjOmrx/tjbGn1xnP9OnYW+FMe56/keRzw0ACUTXDJIJV1AAAIDrUKAAAADXoUABAACuQ4ECAABchwIFAAC4TtJ28UTVsROGd2BfYxzY9I4VU1FsPvumcmI/KybvMbOLZ1CHnVZMmTzWHADEiu4bnIq4ggIAAFwn5gJl9erVGjVqlAoKCuTxeLR48WLjfcdxdN9996lz587KyMhQSUmJtm3bFq98ASQh1g0AsYq5QDl69KgGDBigp556Kuz7Dz/8sJ544gnNmTNH69atU5s2bTR8+HDV1NR84WQBJCfWDQCxinkPyogRIzRixIiw7zmOo8cee0z33HOPRo8eLUl67rnnlJeXp8WLF+uGG274YtkCSEqsGwBiFdc9KDt27FBFRYVKSkqCczk5ORo0aJDWrFkT9jO1tbWqrq42XgBOHY1ZNyTWDqCli2sXT0XFp50teXl5xnxeXl7wvVAzZszQtGnTIh47pV1bY+w/cDDiZyqX9LXm8ka/bYx9vc6wYhq2vW9+5jG7Y2jvjy4yxmXnRu4qSsnJseac48eNcaCuzorxZpjPAgqEfEaSjl87yP5cvWOM05a+HjFHoLk1Zt2Qol87WoKXyv+d6BQMdBWhOSS8i2fKlCmqqqoKvnbv3p3olAAkAdYOoGWLa4GSn58vSaqsrDTmKysrg++FSktLU3Z2tvECcOpozLohsXYALV1cC5Tu3bsrPz9fK1asCM5VV1dr3bp1Ki4ujuepALQQrBsAwol5D8qRI0e0ffv24HjHjh3avHmz2rVrp27dumnixIm6//771atXL3Xv3l333nuvCgoKdM0118QzbwBJhHUDQKxiLlA2bNigL3/5y8Hx5MmTJUnjx4/XvHnzdNddd+no0aO67bbbdOjQIQ0ZMkQvvvii0tPTv1CioZti99x9kRVz2kNrjXHbOZlWTOiG09ANsZLkTU01xoH6Bium8yPmptjdU+18uk4zY6pG2Jt2D/Y2L2LVtQ1YMWdMMje3HvjrmVZMu5HrrDlv69bG2D6yVP5jM++Ch+zNvp6UFGPs+P1hjgScXKLWDTSNeG3aZbMtPk/MBcrQoUPlOM5J3/d4PJo+fbqmT5/+hRID0HKwbgCIVcK7eAAAAEJRoAAAANehQAEAAK5DgQIAAFwnrre6b0qhHSmhHTuS5PF6jPHOsXbfSq9l5m3ifZ062ifzmT+WH6xebYU82bOXMQ7t2Aknc4Gds91nFFm7ke9FFRc4dixiTLiunVB07QDNh84W4FNcQQEAAK5DgQIAAFyHAgUAALgOBQoAAHAdChQAAOA6SdPFo4aQ5+E4doeONzPHGPe6uSzycf32cXZ+u7sxfrJXpRWj4nPM8Rr72RQ7ZprPuel+d5iOGY9ZIz6yw4750emD7c+F2DbvfGuu180bjLGvsKsVE9j3SciEfTvyQG2NMQ59No9Epw8QL/F6zk2yoosJJ3AFBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHUoUAAAgOt4HMex2zYSqLq6Wjk5ORqq0fJ5Wp080BOmtgrT2QMgdg1OvVZqiaqqqpSdnZ3odKJyYu04+F4PZWfZnWbxRrcJYIr3usEVFAAA4DoUKAAAwHUoUAAAgOtQoAAAANdJmlvdV95h3jY+73H7lvAVE82Y/MfC3FoeAAC4HldQAACA61CgAAAA16FAAQAArkOBAgAAXIcCBQAAuE7SdPGE69oJdbRrfO7a701NNcaBujorJnDpeeZnVm1s1Ll8HTsY44aPP7FiPCnmbbsdv79R54pG6HeXpEB9gzkR5pECKZmZZkiYHAM1tRGP4yvobE7U2j97p6CjMfa/+a4VE8rjsx+bYOUY7lEJoY9UCBNz3webjPEDo8dZMdHk+MEjxca4x3+ts2J87doa44b9+yMeF03jpfJ/JzoFoEkl+nEOXEEBAACuQ4ECAABchwIFAAC4DgUKAABwHQoUAADgOknTxRMqJSvLmjvzAbNTImyvS0hXxu57Blsh3R40uye8A86yjxPStePL62SFHDuv0BinLltvxVhdO6FdI2rarp3Q84XrWNp9r/mMo67/bXdU7ZrQ3xif9tBa+1whHTDhOmsayveeNNWgRnSuOA31MX/m0w+G6ewJMb3HwJCZyB074fT40ZqIMXTtAI2T6I4UxI4rKAAAwHUoUAAAgOtQoAAAANehQAEAAK5DgQIAAFwnabt4AseOWXPePmeYEwcORjxO6mF7LrRrxvn3Fium/C6zs6XgYbuzJfXFyB06VpdImK4R/2VFxjhlRZl9nGiEOb8v3+w+athbYcVYXTthjlP4vPm5BivCFq6zxtfJfM5Ow76PozgSAHy+cM9OorPH3biCAgAAXIcCBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHWSposnJTPTGHvatLZiAjs+MsZHr7efs5Oz+n1jnPd4mGfGRDi3JP3ye3OM8Yyfn2vF+HqbXUUN726zYjwpKZHP39iunVBhOoRCu3bC5WM9CyiK59NEFRMGXTsAIqH75tTQZFdQnnrqKZ1++ulKT0/XoEGD9PrrrzfVqQC0EKwbAE5okgLlj3/8oyZPnqypU6dq48aNGjBggIYPH659+/Y1xekAtACsGwA+q0kKlEcffVS33nqrvvWtb+mss87SnDlz1Lp1a/32t79titMBaAFYNwB8Vtz3oNTV1amsrExTpkwJznm9XpWUlGjNmjVWfG1trWpra4PjqqoqSVKD6iXnP3GOU2d8zhOwUw+ExDTU11gxDYGQGMe+m2no/onQc0vS0cPmvoywx/HXRozxRLFXw3H8EWPiJVw+UZ0/iu+K5NGgT//8HMeJEBkfsa4b0snXjuojjdv/hOTB+uJO8V434l6gfPLJJ/L7/crLyzPm8/Ly9O6771rxM2bM0LRp06z5V/U3c+KIPn8czuIoYqIR5lwr+ofOvG8HvRfFsZuv9ohOY/P5IK5ZwCX279+vnJycJj9PrOuGdPK1o/C8D5siRbgKC46bxWvdSHgXz5QpUzR58uTg+NChQyosLNSuXbuaZWGMh+rqanXt2lW7d+9WdnZ2otOJWjLmTc7No6qqSt26dVO7du0SncpJsXYkBjk3j2TMOd7rRtwLlA4dOiglJUWVlZXGfGVlpfLz8634tLQ0paWlWfM5OTlJ84dyQnZ2dtLlLCVn3uTcPLze5rlVUqzrhsTakWjk3DySMed4rRtxX31SU1NVVFSkFStWBOcCgYBWrFih4uLieJ8OQAvAugEgVJP8imfy5MkaP368zj//fF144YV67LHHdPToUX3rW99qitMBaAFYNwB8VpMUKGPHjtXHH3+s++67TxUVFTr33HP14osvWhvgwklLS9PUqVPDXrp1q2TMWUrOvMm5eSQi5y+ybkj8nJsLOTcPcpY8TnP1EQIAAESJhwUCAADXoUABAACuQ4ECAABchwIFAAC4DgUKAABwHdcVKE899ZROP/10paena9CgQXr99dcTnVLQ6tWrNWrUKBUUFMjj8Wjx4sXG+47j6L777lPnzp2VkZGhkpISbdu2LTHJ/p8ZM2boggsuUFZWljp16qRrrrlGW7duNWJqampUWlqq9u3bKzMzU2PGjLHu6NmcZs+erf79+wfvoFhcXKxly5a5Nt9wZs6cKY/Ho4kTJwbn3Jb3z372M3k8HuPVp08f1+b7edy8bkjJt3awbiRGMqwbUvOtHa4qUP74xz9q8uTJmjp1qjZu3KgBAwZo+PDh2rdvX6JTkyQdPXpUAwYM0FNPPRX2/YcfflhPPPGE5syZo3Xr1qlNmzYaPny4amrspyo3l1WrVqm0tFRr167V8uXLVV9fr8svv1xHjx4NxkyaNEkvvPCCFi5cqFWrVqm8vFzXXXddwnLu0qWLZs6cqbKyMm3YsEHDhg3T6NGj9fbbb7sy31Dr16/Xr371K/Xvbz5R0o159+vXT3v37g2+Xn311eB7bsw3HLevG1LyrR2sG80vmdYNqZnWDsdFLrzwQqe0tDQ49vv9TkFBgTNjxowEZhWeJGfRokXBcSAQcPLz852f//znwblDhw45aWlpzh/+8IcEZBjevn37HEnOqlWrHMf5NMdWrVo5CxcuDMa88847jiRnzZo1iUrT0rZtW+c3v/mN6/M9fPiw06tXL2f58uXOpZde6txxxx2O47jz5zx16lRnwIABYd9zY74nk0zrhuMk59rButG0kmndcJzmWztccwWlrq5OZWVlKikpCc55vV6VlJRozZo1CcwsOjt27FBFRYWRf05OjgYNGuSq/KuqqiQp+LTJsrIy1dfXG3n36dNH3bp1c0Xefr9fCxYs0NGjR1VcXOz6fEtLSzVy5EgjP8m9P+dt27apoKBAPXr00E033aRdu3ZJcm++oZJ93ZCSY+1g3WhaybZuSM2zdjTJre4b45NPPpHf77dua52Xl6d33303QVlFr6KiQpLC5n/ivUQLBAKaOHGiLr74Yp199tmSPs07NTVVubm5Rmyi837zzTdVXFysmpoaZWZmatGiRTrrrLO0efNmV+YrSQsWLNDGjRu1fv166z03/pwHDRqkefPmqXfv3tq7d6+mTZumSy65RG+99ZYr8w0n2dcNyf1rB+tG00q2dUNqvrXDNQUKml5paaneeust43eFbtW7d29t3rxZVVVVev755zV+/HitWrUq0Wmd1O7du3XHHXdo+fLlSk9PT3Q6URkxYkTwv/v3769BgwapsLBQf/rTn5SRkZHAzOAmrBtNJxnXDan51g7X/IqnQ4cOSklJsXb6VlZWKj8/P0FZRe9Ejm7Nf8KECVq6dKleeeUVdenSJTifn5+vuro6HTp0yIhPdN6pqanq2bOnioqKNGPGDA0YMECPP/64a/MtKyvTvn37dN5558nn88nn82nVqlV64okn5PP5lJeX58q8Pys3N1dnnnmmtm/f7tqfc6hkXzckd68drBtNqyWsG1LTrR2uKVBSU1NVVFSkFStWBOcCgYBWrFih4uLiBGYWne7duys/P9/Iv7q6WuvWrUto/o7jaMKECVq0aJFefvllde/e3Xi/qKhIrVq1MvLeunWrdu3a5aqfeyAQUG1trWvzveyyy/Tmm29q8+bNwdf555+vm266Kfjfbsz7s44cOaL3339fnTt3du3POVSyrxuSO9cO1o3m0RLWDakJ147G7+ONvwULFjhpaWnOvHnznC1btji33Xabk5ub61RUVCQ6NcdxPt1pvWnTJmfTpk2OJOfRRx91Nm3a5OzcudNxHMeZOXOmk5ub6yxZssR54403nNGjRzvdu3d3jh8/nrCcv//97zs5OTnOypUrnb179wZfx44dC8Z873vfc7p16+a8/PLLzoYNG5zi4mKnuLg4YTnffffdzqpVq5wdO3Y4b7zxhnP33Xc7Ho/H+fvf/+7KfE/ms7vxHcd9ed95553OypUrnR07djj/+te/nJKSEqdDhw7Ovn37XJnvybh93XCc5Fs7WDcSx+3rhuM039rhqgLFcRznl7/8pdOtWzcnNTXVufDCC521a9cmOqWgV155xZFkvcaPH+84zqftgvfee6+Tl5fnpKWlOZdddpmzdevWhOYcLl9Jzty5c4Mxx48fd37wgx84bdu2dVq3bu1ce+21zt69exOW87e//W2nsLDQSU1NdTp27OhcdtllwUXGjfmeTOhC47a8x44d63Tu3NlJTU11TjvtNGfs2LHO9u3bXZvv53HzuuE4ybd2sG4kjtvXDcdpvrXD4ziO08irOgAAAE3CNXtQAAAATqBAAQAArkOBAgAAXIcCBQAAuA4FCgAAcB0KFAAA4DoUKAAAwHUoUAAAgOtQoAAAANehQAEAAK5DgQIAAFzn/wNpnwaxmdFpjQAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.subplot(1, 2, 1)\n",
    "plt.pcolormesh(example_tokens.to_tensor())\n",
    "plt.title('Token IDs')\n",
    "\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.pcolormesh(example_tokens.to_tensor() != 0)\n",
    "plt.title('Mask')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "3O0B4XdFlRgc"
   },
   "source": [
    "### Process the dataset\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rVCuyuSp_whd"
   },
   "source": [
    "The `process_text` function below converts the `Datasets` of strings, into  0-padded tensors of token IDs. It also converts from a `(context, target)` pair to an `((context, target_in), target_out)` pair for training with `keras.Model.fit`. Keras expects `(inputs, labels)` pairs, the inputs are the `(context, target_in)` and the labels are `target_out`. The difference between `target_in` and `target_out` is that they are shifted by one step relative to eachother, so that at each location the label is the next token."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:27.911404Z",
     "iopub.status.busy": "2023-12-07T12:22:27.911139Z",
     "iopub.status.idle": "2023-12-07T12:22:28.115341Z",
     "shell.execute_reply": "2023-12-07T12:22:28.114646Z"
    },
    "id": "wk5tbZWQl5u1"
   },
   "outputs": [],
   "source": [
    "def process_text(context, target):\n",
    "  context = context_text_processor(context).to_tensor()\n",
    "  target = target_text_processor(target)\n",
    "  targ_in = target[:,:-1].to_tensor()\n",
    "  targ_out = target[:,1:].to_tensor()\n",
    "  return (context, targ_in), targ_out\n",
    "\n",
    "\n",
    "train_ds = train_raw.map(process_text, tf.data.AUTOTUNE)\n",
    "val_ds = val_raw.map(process_text, tf.data.AUTOTUNE)\n",
    "test_ds = test_raw.map(process_text, tf.data.AUTOTUNE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4iGi7X2m_tbM"
   },
   "source": [
    "Here is the first sequence of each, from the first batch:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:28.119272Z",
     "iopub.status.busy": "2023-12-07T12:22:28.118779Z",
     "iopub.status.idle": "2023-12-07T12:22:28.313215Z",
     "shell.execute_reply": "2023-12-07T12:22:28.312401Z"
    },
    "id": "woQBWAjLsJkr"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[  4 547 285   1   2   7 281  57 143   2]\n",
      "\n",
      "[  3  12  19 256   1  15  11 893 155   8]\n",
      "[ 12  19 256   1  15  11 893 155   8 946]\n"
     ]
    }
   ],
   "source": [
    "for (ex_context_tok, ex_tar_in), ex_tar_out in train_ds.take(1):\n",
    "  print(ex_context_tok[0, :10].numpy()) \n",
    "  print()\n",
    "  print(ex_tar_in[0, :10].numpy()) \n",
    "  print(ex_tar_out[0, :10].numpy()) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can decode the token IDs back to text by using the vocabulary lookup of each, the context and target "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "> French : [START] tout cela au nom de la legitime defense . [END]                                            \n",
      "> English Input : [START] all in the name of selfdefense .                                             \n"
     ]
    }
   ],
   "source": [
    "for (ex_context_tok, ex_tar_in), ex_tar_out in test_ds.take(1):\n",
    "    tokens = context_vocab[ex_context_tok[0].numpy()]\n",
    "    print('> French : ' + ' '.join(tokens))\n",
    "    tokens = target_vocab[ex_tar_in[0].numpy()]\n",
    "    print('> English Input : ' + ' '.join(tokens))\n",
    "    break"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TNfHIF71ulLu"
   },
   "source": [
    "## The encoder/decoder\n",
    "\n",
    "The following diagrams shows an overview of the model. In both the encoder is on the left, the decoder is on the right. At each time-step the decoder's output is combined with the encoder's output, to predict the next word. \n",
    "\n",
    "The original [left] contains a few extra connections that are intentionally omitted from this tutorial's model [right], as they are generally unnecessary, and difficult to implement. Those missing connections are:\n",
    "\n",
    "1. Feeding the state from the encoder's RNN to the decoder's RNN\n",
    "2. Feeding the attention output back to the RNN's input.\n",
    "\n",
    "<table>\n",
    "<tr>\n",
    "  <td>\n",
    "   <img width=500 src=\"https://www.tensorflow.org/images/seq2seq/attention_mechanism.jpg\"/>\n",
    "  </td>\n",
    "  <td>\n",
    "   <img width=380 src=\"https://www.tensorflow.org/images/tutorials/transformer/RNN+attention.png\"/>\n",
    "  </td>\n",
    "</tr>\n",
    "<tr>\n",
    "  <th colspan=1>The original from <a href=https://arxiv.org/abs/1508.04025v5>Effective Approaches to Attention-based Neural Machine Translation</a></th>\n",
    "  <th colspan=1>This tutorial's model</th>\n",
    "<tr>\n",
    "</table>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gzQWx2saImMV"
   },
   "source": [
    "Before getting into it define constants for the model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:28.317005Z",
     "iopub.status.busy": "2023-12-07T12:22:28.316322Z",
     "iopub.status.idle": "2023-12-07T12:22:28.319746Z",
     "shell.execute_reply": "2023-12-07T12:22:28.319160Z"
    },
    "id": "_a9uNz3-IrF-"
   },
   "outputs": [],
   "source": [
    "UNITS = 256"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "blNgVbLSzpsr"
   },
   "source": [
    "### The encoder\n",
    "\n",
    "The goal of the encoder is to process the context sequence into a sequence of vectors that are useful for the decoder as it attempts to predict the next output for each timestep. Since the context sequence is constant, there is no restriction on how information can flow in the encoder, so use a bidirectional-RNN to do the processing:\n",
    "\n",
    "<table>\n",
    "<tr>\n",
    "  <td>\n",
    "   <img width=500 src=\"https://tensorflow.org/images/tutorials/transformer/RNN-bidirectional.png\"/>\n",
    "  </td>\n",
    "</tr>\n",
    "<tr>\n",
    "  <th>A bidirectional RNN</th>\n",
    "<tr>\n",
    "</table>\n",
    "\n",
    "The encoder:\n",
    "\n",
    "1. Takes a list of token IDs (from `context_text_processor`).\n",
    "3. Looks up an embedding vector for each token (Using a `layers.Embedding`).\n",
    "4. Processes the embeddings into a new sequence (Using a bidirectional `layers.GRU`).\n",
    "5. Returns the processed sequence. This will be passed to the attention head."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:28.323345Z",
     "iopub.status.busy": "2023-12-07T12:22:28.322724Z",
     "iopub.status.idle": "2023-12-07T12:22:28.329737Z",
     "shell.execute_reply": "2023-12-07T12:22:28.329065Z"
    },
    "id": "nZ2rI24i3jFg"
   },
   "outputs": [],
   "source": [
    "class Encoder(tf.keras.layers.Layer):\n",
    "  def __init__(self, text_processor, units):\n",
    "    super(Encoder, self).__init__()\n",
    "    self.text_processor = text_processor\n",
    "    self.vocab_size = text_processor.vocabulary_size()\n",
    "    self.units = units\n",
    "    \n",
    "    # The embedding layer converts tokens to vectors\n",
    "    self.embedding = tf.keras.layers.Embedding(self.vocab_size, units,\n",
    "                                               mask_zero=True)\n",
    "\n",
    "    # The RNN layer processes those vectors sequentially.\n",
    "    self.rnn = tf.keras.layers.Bidirectional(\n",
    "        merge_mode='sum',\n",
    "        layer=tf.keras.layers.GRU(units,\n",
    "                            # Return the sequence and state\n",
    "                            return_sequences=True,\n",
    "                            recurrent_initializer='glorot_uniform'))\n",
    "\n",
    "  def call(self, x):\n",
    "    shape_checker = ShapeChecker()\n",
    "    shape_checker(x, 'batch s')\n",
    "\n",
    "    # 2. The embedding layer looks up the embedding vector for each token.\n",
    "    x = self.embedding(x)\n",
    "    shape_checker(x, 'batch s units')\n",
    "\n",
    "    # 3. The GRU processes the sequence of embeddings.\n",
    "    x = self.rnn(x)\n",
    "    shape_checker(x, 'batch s units')\n",
    "\n",
    "    # 4. Returns the new sequence of embeddings.\n",
    "    return x\n",
    "\n",
    "  def convert_input(self, texts):\n",
    "    texts = tf.convert_to_tensor(texts)\n",
    "    if len(texts.shape) == 0:\n",
    "      texts = tf.convert_to_tensor(texts)[tf.newaxis]\n",
    "    context = self.text_processor(texts).to_tensor()\n",
    "    context = self(context)\n",
    "    return context"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "D3SKkaQeGn-Q"
   },
   "source": [
    "Try it out:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:28.333093Z",
     "iopub.status.busy": "2023-12-07T12:22:28.332575Z",
     "iopub.status.idle": "2023-12-07T12:22:29.046111Z",
     "shell.execute_reply": "2023-12-07T12:22:29.045369Z"
    },
    "id": "60gSVh05Jl6l"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Context tokens, shape (batch, s): (64, 55)\n",
      "Encoder output, shape (batch, s, units): (64, 55, 256)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-23 17:55:54.535959: I external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:454] Loaded cuDNN version 8902\n"
     ]
    }
   ],
   "source": [
    "# Encode the input sequence.\n",
    "encoder = Encoder(context_text_processor, UNITS)\n",
    "ex_context = encoder(ex_context_tok)\n",
    "\n",
    "print(f'Context tokens, shape (batch, s): {ex_context_tok.shape}')\n",
    "print(f'Encoder output, shape (batch, s, units): {ex_context.shape}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "45xM_Gl1MgXY"
   },
   "source": [
    "### The attention layer\n",
    "\n",
    "The attention layer lets the decoder access the information extracted by the encoder. It computes a vector from the entire context sequence, and adds that to the decoder's output. \n",
    "\n",
    "The simplest way you could calculate a single vector from the entire sequence would be to take the average across the sequence (`layers.GlobalAveragePooling1D`). An attention layer is similar, but calculates a **weighted** average across the context sequence. Where the weights are calculated from the combination of context and \"query\" vectors.\n",
    "\n",
    "<table>\n",
    "<tr>\n",
    "  <td>\n",
    "   <img width=500 src=\"https://www.tensorflow.org/images/tutorials/transformer/CrossAttention-new-full.png\"/>\n",
    "  </td>\n",
    "</tr>\n",
    "<tr>\n",
    "  <th colspan=1>The attention layer</th>\n",
    "<tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.050088Z",
     "iopub.status.busy": "2023-12-07T12:22:29.049552Z",
     "iopub.status.idle": "2023-12-07T12:22:29.055725Z",
     "shell.execute_reply": "2023-12-07T12:22:29.055145Z"
    },
    "id": "-Ql3ymqwD8LS"
   },
   "outputs": [],
   "source": [
    "class CrossAttention(tf.keras.layers.Layer):\n",
    "  def __init__(self, units, **kwargs):\n",
    "    super().__init__()\n",
    "    self.mha = tf.keras.layers.MultiHeadAttention(key_dim=units, num_heads=1, **kwargs)\n",
    "    self.layernorm = tf.keras.layers.LayerNormalization()\n",
    "    self.add = tf.keras.layers.Add()\n",
    "\n",
    "  def call(self, x, context):\n",
    "    shape_checker = ShapeChecker()\n",
    " \n",
    "    shape_checker(x, 'batch t units')\n",
    "    shape_checker(context, 'batch s units')\n",
    "\n",
    "    attn_output, attn_scores = self.mha(\n",
    "        query=x,\n",
    "        value=context,\n",
    "        return_attention_scores=True)\n",
    "    \n",
    "    shape_checker(x, 'batch t units')\n",
    "    shape_checker(attn_scores, 'batch heads t s')\n",
    "    \n",
    "    # Cache the attention scores for plotting later.\n",
    "    attn_scores = tf.reduce_mean(attn_scores, axis=1)\n",
    "    shape_checker(attn_scores, 'batch t s')\n",
    "    self.last_attention_weights = attn_scores\n",
    "\n",
    "    x = self.add([x, attn_output])\n",
    "    x = self.layernorm(x)\n",
    "\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.058971Z",
     "iopub.status.busy": "2023-12-07T12:22:29.058489Z",
     "iopub.status.idle": "2023-12-07T12:22:29.356131Z",
     "shell.execute_reply": "2023-12-07T12:22:29.355469Z"
    },
    "id": "7y7hjPkNMmHh"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Context sequence, shape (batch, s, units): (64, 55, 256)\n",
      "Target sequence, shape (batch, t, units): (64, 53, 256)\n",
      "Attention result, shape (batch, t, units): (64, 53, 256)\n",
      "Attention weights, shape (batch, t, s):    (64, 53, 55)\n"
     ]
    }
   ],
   "source": [
    "attention_layer = CrossAttention(UNITS)\n",
    "\n",
    "# Attend to the encoded tokens\n",
    "embed = tf.keras.layers.Embedding(target_text_processor.vocabulary_size(),\n",
    "                                  output_dim=UNITS, mask_zero=True)\n",
    "ex_tar_embed = embed(ex_tar_in)\n",
    "\n",
    "result = attention_layer(ex_tar_embed, ex_context)\n",
    "\n",
    "print(f'Context sequence, shape (batch, s, units): {ex_context.shape}')\n",
    "print(f'Target sequence, shape (batch, t, units): {ex_tar_embed.shape}')\n",
    "print(f'Attention result, shape (batch, t, units): {result.shape}')\n",
    "print(f'Attention weights, shape (batch, t, s):    {attention_layer.last_attention_weights.shape}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vx9fUhi3Pmwp"
   },
   "source": [
    "The attention weights will sum to `1` over the context sequence, at each location in the target sequence."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.359942Z",
     "iopub.status.busy": "2023-12-07T12:22:29.359404Z",
     "iopub.status.idle": "2023-12-07T12:22:29.365050Z",
     "shell.execute_reply": "2023-12-07T12:22:29.364488Z"
    },
    "id": "zxyR7cmQPn9P"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.       , 1.       , 1.       , 1.       , 1.       , 0.9999999,\n",
       "       1.       , 1.       , 0.9999998, 0.9999998, 0.9999998, 0.9999998,\n",
       "       0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998,\n",
       "       0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998,\n",
       "       0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998,\n",
       "       0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998,\n",
       "       0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998,\n",
       "       0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998,\n",
       "       0.9999998, 0.9999998, 0.9999998, 0.9999998, 0.9999998],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "attention_layer.last_attention_weights[0].numpy().sum(axis=-1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "AagyXMH-Jhqt"
   },
   "source": [
    "\n",
    "\n",
    "Here are the attention weights across the context sequences at `t=0`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.368783Z",
     "iopub.status.busy": "2023-12-07T12:22:29.368214Z",
     "iopub.status.idle": "2023-12-07T12:22:29.629324Z",
     "shell.execute_reply": "2023-12-07T12:22:29.628459Z"
    },
    "id": "Rqr8XGsAJlf6"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "attention_weights = attention_layer.last_attention_weights\n",
    "mask=(ex_context_tok != 0).numpy()\n",
    "\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.pcolormesh(mask*attention_weights[:, 0, :])\n",
    "plt.title('Attention weights')\n",
    "\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.pcolormesh(mask)\n",
    "plt.title('Mask');\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6Eil-C_NN1rp"
   },
   "source": [
    "Because of the small-random initialization the attention weights are initially all close to `1/(sequence_length)`. The model will learn to make these less uniform as training progresses."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "aQ638eHN4iCK"
   },
   "source": [
    "### The decoder\n",
    "\n",
    "The decoder's job is to generate predictions for the next token at each location in the target sequence.\n",
    "\n",
    "1. It looks up embeddings for each token in the target sequence.\n",
    "2. It uses an RNN to process the target sequence, and keep track of what it has generated so far.\n",
    "3. It uses RNN output as the \"query\" to the attention layer, when attending to the encoder's output.\n",
    "4. At each location in the output it predicts the next token.\n",
    "\n",
    "When training, the model predicts the next word at each location. So it's important that the information only flows in one direction through the model. The decoder uses a unidirectional (not bidirectional) RNN to process the target sequence.\n",
    "\n",
    "When running inference with this model it produces one word at a time, and those are fed back into the model.\n",
    "\n",
    "<table>\n",
    "<tr>\n",
    "  <td>\n",
    "   <img width=500 src=\"https://tensorflow.org/images/tutorials/transformer/RNN.png\"/>\n",
    "  </td>\n",
    "</tr>\n",
    "<tr>\n",
    "  <th>A unidirectional RNN</th>\n",
    "<tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "pZsQJMqNmg_L"
   },
   "source": [
    "Here is the `Decoder` class' initializer. The initializer creates all the necessary layers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.633554Z",
     "iopub.status.busy": "2023-12-07T12:22:29.632980Z",
     "iopub.status.idle": "2023-12-07T12:22:29.639749Z",
     "shell.execute_reply": "2023-12-07T12:22:29.639078Z"
    },
    "id": "erYvHIgAl8kh"
   },
   "outputs": [],
   "source": [
    "class Decoder(tf.keras.layers.Layer):\n",
    "  @classmethod\n",
    "  def add_method(cls, fun):\n",
    "    setattr(cls, fun.__name__, fun)\n",
    "    return fun\n",
    "\n",
    "  def __init__(self, text_processor, units):\n",
    "    super(Decoder, self).__init__()\n",
    "    self.text_processor = text_processor\n",
    "    self.vocab_size = text_processor.vocabulary_size()\n",
    "    self.word_to_id = tf.keras.layers.StringLookup(\n",
    "        vocabulary=text_processor.get_vocabulary(),\n",
    "        mask_token='', oov_token='[UNK]')\n",
    "    self.id_to_word = tf.keras.layers.StringLookup(\n",
    "        vocabulary=text_processor.get_vocabulary(),\n",
    "        mask_token='', oov_token='[UNK]',\n",
    "        invert=True)\n",
    "    self.start_token = self.word_to_id('[START]')\n",
    "    self.end_token = self.word_to_id('[END]')\n",
    "\n",
    "    self.units = units\n",
    "\n",
    "\n",
    "    # 1. The embedding layer converts token IDs to vectors\n",
    "    self.embedding = tf.keras.layers.Embedding(self.vocab_size,\n",
    "                                               units, mask_zero=True)\n",
    "\n",
    "    # 2. The RNN keeps track of what's been generated so far.\n",
    "    self.rnn = tf.keras.layers.GRU(units,\n",
    "                                   return_sequences=True,\n",
    "                                   return_state=True,\n",
    "                                   recurrent_initializer='glorot_uniform')\n",
    "\n",
    "    # 3. The RNN output will be the query for the attention layer.\n",
    "    self.attention = CrossAttention(units)\n",
    "\n",
    "    # 4. This fully connected layer produces the logits for each\n",
    "    # output token.\n",
    "    self.output_layer = tf.keras.layers.Dense(self.vocab_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Sd8-nRNzFR8x"
   },
   "source": [
    "#### Training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "UPnaw583CpnY"
   },
   "source": [
    "Next, the `call` method, takes 3 arguments:\n",
    "\n",
    "* `inputs` -  a `context, x` pair where:\n",
    "  * `context` - is the context from the encoder's output.\n",
    "  * `x` - is the target sequence input.\n",
    "* `state` - Optional, the previous `state` output from the decoder (the internal state of the decoder's RNN). Pass the state from a previous run to continue generating text where you left off.\n",
    "* `return_state` - [Default: False] - Set this to `True` to return the RNN state. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.643561Z",
     "iopub.status.busy": "2023-12-07T12:22:29.642906Z",
     "iopub.status.idle": "2023-12-07T12:22:29.648348Z",
     "shell.execute_reply": "2023-12-07T12:22:29.647732Z"
    },
    "id": "PJOi5btHAPNK"
   },
   "outputs": [],
   "source": [
    "@Decoder.add_method\n",
    "def call(self,\n",
    "         context, x,\n",
    "         state=None,\n",
    "         return_state=False):  \n",
    "  shape_checker = ShapeChecker()\n",
    "  shape_checker(x, 'batch t')\n",
    "  shape_checker(context, 'batch s units')\n",
    "\n",
    "  # 1. Lookup the embeddings\n",
    "  x = self.embedding(x)\n",
    "  shape_checker(x, 'batch t units')\n",
    "\n",
    "  # 2. Process the target sequence.\n",
    "  x, state = self.rnn(x, initial_state=state)\n",
    "  shape_checker(x, 'batch t units')\n",
    "\n",
    "  # 3. Use the RNN output as the query for the attention over the context.\n",
    "  x = self.attention(x, context)\n",
    "  self.last_attention_weights = self.attention.last_attention_weights\n",
    "  shape_checker(x, 'batch t units')\n",
    "  shape_checker(self.last_attention_weights, 'batch t s')\n",
    "\n",
    "  # Step 4. Generate logit predictions for the next token.\n",
    "  logits = self.output_layer(x)\n",
    "  shape_checker(logits, 'batch t target_vocab_size')\n",
    "\n",
    "  if return_state:\n",
    "    return logits, state\n",
    "  else:\n",
    "    return logits"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "E1-mLAcUEXpK"
   },
   "source": [
    "That will be sufficient for training. Create an instance of the decoder to test out:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.651960Z",
     "iopub.status.busy": "2023-12-07T12:22:29.651328Z",
     "iopub.status.idle": "2023-12-07T12:22:29.754850Z",
     "shell.execute_reply": "2023-12-07T12:22:29.754189Z"
    },
    "id": "4ZUMbYXIEVeA"
   },
   "outputs": [],
   "source": [
    "decoder = Decoder(target_text_processor, UNITS)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "SFWaI4wqzt4t"
   },
   "source": [
    "In training you'll use the decoder like this:\n",
    "\n",
    "Given the context and target tokens, for each target token it predicts the next target token. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.758862Z",
     "iopub.status.busy": "2023-12-07T12:22:29.758328Z",
     "iopub.status.idle": "2023-12-07T12:22:29.833016Z",
     "shell.execute_reply": "2023-12-07T12:22:29.832287Z"
    },
    "id": "5YM-lD7bzx18"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "encoder output shape: (batch, s, units) (64, 55, 256)\n",
      "input target tokens shape: (batch, t) (64, 53)\n",
      "logits shape shape: (batch, target_vocabulary_size) (64, 53, 5000)\n"
     ]
    }
   ],
   "source": [
    "logits = decoder(ex_context, ex_tar_in)\n",
    "\n",
    "print(f'encoder output shape: (batch, s, units) {ex_context.shape}')\n",
    "print(f'input target tokens shape: (batch, t) {ex_tar_in.shape}')\n",
    "print(f'logits shape shape: (batch, target_vocabulary_size) {logits.shape}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "zhS_tbk7VQkX"
   },
   "source": [
    "#### Inference\n",
    "\n",
    "To use it for inference you'll need a couple more methods:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.836820Z",
     "iopub.status.busy": "2023-12-07T12:22:29.836126Z",
     "iopub.status.idle": "2023-12-07T12:22:29.840441Z",
     "shell.execute_reply": "2023-12-07T12:22:29.839810Z"
    },
    "id": "SPm12cnIVRQr"
   },
   "outputs": [],
   "source": [
    "@Decoder.add_method\n",
    "def get_initial_state(self, context):\n",
    "  batch_size = tf.shape(context)[0]\n",
    "  start_tokens = tf.fill([batch_size, 1], self.start_token)\n",
    "  done = tf.zeros([batch_size, 1], dtype=tf.bool)\n",
    "  embedded = self.embedding(start_tokens)\n",
    "  return start_tokens, done, self.rnn.get_initial_state(embedded)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.843716Z",
     "iopub.status.busy": "2023-12-07T12:22:29.843193Z",
     "iopub.status.idle": "2023-12-07T12:22:29.847217Z",
     "shell.execute_reply": "2023-12-07T12:22:29.846553Z"
    },
    "id": "TzeOhpBvVS5L"
   },
   "outputs": [],
   "source": [
    "@Decoder.add_method\n",
    "def tokens_to_text(self, tokens):\n",
    "  words = self.id_to_word(tokens)\n",
    "  result = tf.strings.reduce_join(words, axis=-1, separator=' ')\n",
    "  result = tf.strings.regex_replace(result, '^ *\\[START\\] *', '')\n",
    "  result = tf.strings.regex_replace(result, ' *\\[END\\] *$', '')\n",
    "  return result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.850498Z",
     "iopub.status.busy": "2023-12-07T12:22:29.850032Z",
     "iopub.status.idle": "2023-12-07T12:22:29.854957Z",
     "shell.execute_reply": "2023-12-07T12:22:29.854372Z"
    },
    "id": "v6ildnz_V1MA"
   },
   "outputs": [],
   "source": [
    "@Decoder.add_method\n",
    "def get_next_token(self, context, next_token, done, state, temperature = 0.0):\n",
    "  logits, state = self(\n",
    "    context, next_token,\n",
    "    state = state,\n",
    "    return_state=True) \n",
    "  \n",
    "  if temperature == 0.0:\n",
    "    next_token = tf.argmax(logits, axis=-1)\n",
    "  else:\n",
    "    logits = logits[:, -1, :]/temperature\n",
    "    next_token = tf.random.categorical(logits, num_samples=1)\n",
    "\n",
    "  # If a sequence produces an `end_token`, set it `done`\n",
    "  done = done | (next_token == self.end_token)\n",
    "  # Once a sequence is done it only produces 0-padding.\n",
    "  next_token = tf.where(done, tf.constant(0, dtype=tf.int64), next_token)\n",
    "  \n",
    "  return next_token, done, state"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "9WiXLrVs-FTE"
   },
   "source": [
    "With those extra functions, you can write a generation loop:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:29.858446Z",
     "iopub.status.busy": "2023-12-07T12:22:29.857883Z",
     "iopub.status.idle": "2023-12-07T12:22:30.453703Z",
     "shell.execute_reply": "2023-12-07T12:22:30.452994Z"
    },
    "id": "SuehagxL-JBZ"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([b'seriously judges taste recommendations authorities fed ghana kurdish followers literature',\n",
       "       b'madagascar pm austerity han counting portugal ground solve distance palm',\n",
       "       b'lopez while freedoms hear downtown operations cambodia serbian accepted opposition'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Setup the loop variables.\n",
    "next_token, done, state = decoder.get_initial_state(ex_context)\n",
    "tokens = []\n",
    "\n",
    "for n in range(10):\n",
    "  # Run one step.\n",
    "  next_token, done, state = decoder.get_next_token(\n",
    "      ex_context, next_token, done, state, temperature=1.0)\n",
    "  # Add the token to the output.\n",
    "  tokens.append(next_token)\n",
    "\n",
    "# Stack all the tokens together.\n",
    "tokens = tf.concat(tokens, axis=-1) # (batch, t)\n",
    "\n",
    "# Convert the tokens back to a a string\n",
    "result = decoder.tokens_to_text(tokens)\n",
    "result[:3].numpy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5ALTdqCMLGSY"
   },
   "source": [
    "Since the model's untrained, it outputs items from the vocabulary almost uniformly at random."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "B6xyru86m914"
   },
   "source": [
    "## The model\n",
    "\n",
    "Now that you have all the model components, combine them to build the model for training:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:30.457619Z",
     "iopub.status.busy": "2023-12-07T12:22:30.457379Z",
     "iopub.status.idle": "2023-12-07T12:22:30.462724Z",
     "shell.execute_reply": "2023-12-07T12:22:30.462054Z"
    },
    "id": "WWIyuy71TkJT"
   },
   "outputs": [],
   "source": [
    "class Translator(tf.keras.Model):\n",
    "  @classmethod\n",
    "  def add_method(cls, fun):\n",
    "    setattr(cls, fun.__name__, fun)\n",
    "    return fun\n",
    "\n",
    "  def __init__(self, units,\n",
    "               context_text_processor,\n",
    "               target_text_processor):\n",
    "    super().__init__()\n",
    "    # Build the encoder and decoder\n",
    "    encoder = Encoder(context_text_processor, units)\n",
    "    decoder = Decoder(target_text_processor, units)\n",
    "\n",
    "    self.encoder = encoder\n",
    "    self.decoder = decoder\n",
    "\n",
    "  def call(self, inputs):\n",
    "    context, x = inputs\n",
    "    context = self.encoder(context)\n",
    "    logits = self.decoder(context, x)\n",
    "\n",
    "    #TODO(b/250038731): remove this\n",
    "    try:\n",
    "      # Delete the keras mask, so keras doesn't scale the loss+accuracy. \n",
    "      del logits._keras_mask\n",
    "    except AttributeError:\n",
    "      pass\n",
    "\n",
    "    return logits"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5rPi0FkS2iA5"
   },
   "source": [
    "During training the model will be used like this:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:30.465902Z",
     "iopub.status.busy": "2023-12-07T12:22:30.465667Z",
     "iopub.status.idle": "2023-12-07T12:22:30.671420Z",
     "shell.execute_reply": "2023-12-07T12:22:30.670748Z"
    },
    "id": "8vhjTh84K6Mg"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Context tokens, shape: (batch, s, units) (64, 55)\n",
      "Target tokens, shape: (batch, t) (64, 53)\n",
      "logits, shape: (batch, t, target_vocabulary_size) (64, 53, 5000)\n"
     ]
    }
   ],
   "source": [
    "model = Translator(UNITS, context_text_processor, target_text_processor)\n",
    "\n",
    "logits = model((ex_context_tok, ex_tar_in))\n",
    "\n",
    "print(f'Context tokens, shape: (batch, s, units) {ex_context_tok.shape}')\n",
    "print(f'Target tokens, shape: (batch, t) {ex_tar_in.shape}')\n",
    "print(f'logits, shape: (batch, t, target_vocabulary_size) {logits.shape}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_ch_71VbIRfK"
   },
   "source": [
    "### Train"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "8FmzjGmprVmE"
   },
   "source": [
    "For training, you'll want to implement your own masked loss and accuracy functions:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:30.675550Z",
     "iopub.status.busy": "2023-12-07T12:22:30.674981Z",
     "iopub.status.idle": "2023-12-07T12:22:30.679248Z",
     "shell.execute_reply": "2023-12-07T12:22:30.678679Z"
    },
    "id": "WmTHr5iV3jFr"
   },
   "outputs": [],
   "source": [
    "def masked_loss(y_true, y_pred):\n",
    "    # Calculate the loss for each item in the batch.\n",
    "    loss_fn = tf.keras.losses.SparseCategoricalCrossentropy(\n",
    "        from_logits=True, reduction='none')\n",
    "    loss = loss_fn(y_true, y_pred)\n",
    "\n",
    "    # Mask off the losses on padding.\n",
    "    mask = tf.cast(y_true != 0, loss.dtype)\n",
    "    loss *= mask\n",
    "\n",
    "    # Return the total.\n",
    "    return tf.reduce_sum(loss)/tf.reduce_sum(mask)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:30.682724Z",
     "iopub.status.busy": "2023-12-07T12:22:30.682192Z",
     "iopub.status.idle": "2023-12-07T12:22:30.686290Z",
     "shell.execute_reply": "2023-12-07T12:22:30.685639Z"
    },
    "id": "nRB1CTmQWOIL"
   },
   "outputs": [],
   "source": [
    "def masked_acc(y_true, y_pred):\n",
    "    # Calculate the loss for each item in the batch.\n",
    "    y_pred = tf.argmax(y_pred, axis=-1)\n",
    "    y_pred = tf.cast(y_pred, y_true.dtype)\n",
    "    \n",
    "    match = tf.cast(y_true == y_pred, tf.float32)\n",
    "    mask = tf.cast(y_true != 0, tf.float32)\n",
    "    \n",
    "    return tf.reduce_sum(match)/tf.reduce_sum(mask)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "f32GuAhw2nXm"
   },
   "source": [
    "Configure the model for training:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:30.689447Z",
     "iopub.status.busy": "2023-12-07T12:22:30.689184Z",
     "iopub.status.idle": "2023-12-07T12:22:30.717258Z",
     "shell.execute_reply": "2023-12-07T12:22:30.716655Z"
    },
    "id": "9g0DRRvm3l9X"
   },
   "outputs": [],
   "source": [
    "model.compile(optimizer='adam',\n",
    "              loss=masked_loss, \n",
    "              metrics=[masked_acc, masked_loss])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5DWLI3pssjnx"
   },
   "source": [
    "The model is randomly initialized, and should give roughly uniform output probabilities. So it's easy to predict what the initial values of the metrics should be:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:30.720596Z",
     "iopub.status.busy": "2023-12-07T12:22:30.720355Z",
     "iopub.status.idle": "2023-12-07T12:22:30.824151Z",
     "shell.execute_reply": "2023-12-07T12:22:30.823460Z"
    },
    "id": "BuP3_LFENMJG"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'expected_loss': 8.517193, 'expected_acc': 0.0002}"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocab_size = 1.0 * target_text_processor.vocabulary_size()\n",
    "\n",
    "{\"expected_loss\": tf.math.log(vocab_size).numpy(),\n",
    " \"expected_acc\": 1/vocab_size}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "frVba49Usd0Z"
   },
   "source": [
    "That should roughly match the values returned by running a few steps of evaluation:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:30.828119Z",
     "iopub.status.busy": "2023-12-07T12:22:30.827582Z",
     "iopub.status.idle": "2023-12-07T12:22:37.523128Z",
     "shell.execute_reply": "2023-12-07T12:22:37.522406Z"
    },
    "id": "8rJITfxEsHKR"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20/20 [==============================] - 5s 35ms/step - loss: 8.5274 - masked_acc: 1.4713e-04 - masked_loss: 8.5274\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 8.52737808227539,\n",
       " 'masked_acc': 0.00014713083510287106,\n",
       " 'masked_loss': 8.52737808227539}"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(val_ds, steps=20, return_dict=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:22:37.526873Z",
     "iopub.status.busy": "2023-12-07T12:22:37.526210Z",
     "iopub.status.idle": "2023-12-07T12:24:09.857567Z",
     "shell.execute_reply": "2023-12-07T12:24:09.856788Z"
    },
    "id": "BQd_esVVoSf3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-01-23 17:56:08.931068: W tensorflow/core/common_runtime/type_inference.cc:339] Type inference failed. This indicates an invalid graph that escaped type checking. Error message: INVALID_ARGUMENT: expected compatible input types, but input 1:\n",
      "type_id: TFT_OPTIONAL\n",
      "args {\n",
      "  type_id: TFT_PRODUCT\n",
      "  args {\n",
      "    type_id: TFT_TENSOR\n",
      "    args {\n",
      "      type_id: TFT_INT32\n",
      "    }\n",
      "  }\n",
      "}\n",
      " is neither a subtype nor a supertype of the combined inputs preceding it:\n",
      "type_id: TFT_OPTIONAL\n",
      "args {\n",
      "  type_id: TFT_PRODUCT\n",
      "  args {\n",
      "    type_id: TFT_TENSOR\n",
      "    args {\n",
      "      type_id: TFT_INT8\n",
      "    }\n",
      "  }\n",
      "}\n",
      "\n",
      "\tfor Tuple type infernce function 0\n",
      "\twhile inferring type of node 'cond_35/output/_22'\n",
      "2024-01-23 17:56:09.379607: I external/local_xla/xla/service/service.cc:168] XLA service 0x7fe2d0182e50 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:\n",
      "2024-01-23 17:56:09.379666: I external/local_xla/xla/service/service.cc:176]   StreamExecutor device (0): NVIDIA GeForce RTX 3060 Ti, Compute Capability 8.6\n",
      "2024-01-23 17:56:09.394673: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.\n",
      "WARNING: All log messages before absl::InitializeLog() is called are written to STDERR\n",
      "I0000 00:00:1706028969.498582  353183 device_compiler.h:186] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100/100 [==============================] - 21s 124ms/step - loss: 5.9198 - masked_acc: 0.1341 - masked_loss: 5.9198 - val_loss: 5.4798 - val_masked_acc: 0.1737 - val_masked_loss: 5.4798\n",
      "Epoch 2/100\n",
      "100/100 [==============================] - 10s 95ms/step - loss: 5.1102 - masked_acc: 0.2192 - masked_loss: 5.1102 - val_loss: 4.8916 - val_masked_acc: 0.2413 - val_masked_loss: 4.8916\n",
      "Epoch 3/100\n",
      "100/100 [==============================] - 8s 78ms/step - loss: 4.7668 - masked_acc: 0.2507 - masked_loss: 4.7668 - val_loss: 4.6684 - val_masked_acc: 0.2624 - val_masked_loss: 4.6684\n",
      "Epoch 4/100\n",
      "100/100 [==============================] - 8s 76ms/step - loss: 4.5463 - masked_acc: 0.2744 - masked_loss: 4.5463 - val_loss: 4.4802 - val_masked_acc: 0.2803 - val_masked_loss: 4.4802\n",
      "Epoch 5/100\n",
      "100/100 [==============================] - 6s 62ms/step - loss: 4.3470 - masked_acc: 0.2973 - masked_loss: 4.3470 - val_loss: 4.2438 - val_masked_acc: 0.3068 - val_masked_loss: 4.2438\n",
      "Epoch 6/100\n",
      "100/100 [==============================] - 6s 64ms/step - loss: 4.1409 - masked_acc: 0.3194 - masked_loss: 4.1409 - val_loss: 4.0503 - val_masked_acc: 0.3258 - val_masked_loss: 4.0503\n",
      "Epoch 7/100\n",
      "100/100 [==============================] - 7s 70ms/step - loss: 3.9856 - masked_acc: 0.3386 - masked_loss: 3.9856 - val_loss: 3.8586 - val_masked_acc: 0.3539 - val_masked_loss: 3.8586\n",
      "Epoch 8/100\n",
      "100/100 [==============================] - 7s 66ms/step - loss: 3.8201 - masked_acc: 0.3566 - masked_loss: 3.8201 - val_loss: 3.7197 - val_masked_acc: 0.3693 - val_masked_loss: 3.7197\n",
      "Epoch 9/100\n",
      "100/100 [==============================] - 7s 67ms/step - loss: 3.6881 - masked_acc: 0.3702 - masked_loss: 3.6881 - val_loss: 3.6284 - val_masked_acc: 0.3813 - val_masked_loss: 3.6284\n",
      "Epoch 10/100\n",
      "100/100 [==============================] - 6s 62ms/step - loss: 3.5684 - masked_acc: 0.3832 - masked_loss: 3.5684 - val_loss: 3.5041 - val_masked_acc: 0.3977 - val_masked_loss: 3.5041\n",
      "Epoch 11/100\n",
      "100/100 [==============================] - 6s 59ms/step - loss: 3.4332 - masked_acc: 0.4035 - masked_loss: 3.4332 - val_loss: 3.4146 - val_masked_acc: 0.4070 - val_masked_loss: 3.4146\n",
      "Epoch 12/100\n",
      "100/100 [==============================] - 6s 59ms/step - loss: 3.3801 - masked_acc: 0.4128 - masked_loss: 3.3801 - val_loss: 3.3015 - val_masked_acc: 0.4207 - val_masked_loss: 3.3015\n",
      "Epoch 13/100\n",
      "100/100 [==============================] - 6s 62ms/step - loss: 3.3113 - masked_acc: 0.4184 - masked_loss: 3.3113 - val_loss: 3.2552 - val_masked_acc: 0.4240 - val_masked_loss: 3.2552\n",
      "Epoch 14/100\n",
      "100/100 [==============================] - 6s 60ms/step - loss: 3.2611 - masked_acc: 0.4250 - masked_loss: 3.2611 - val_loss: 3.2662 - val_masked_acc: 0.4214 - val_masked_loss: 3.2662\n",
      "Epoch 15/100\n",
      "100/100 [==============================] - 6s 63ms/step - loss: 3.1933 - masked_acc: 0.4337 - masked_loss: 3.1933 - val_loss: 3.1295 - val_masked_acc: 0.4431 - val_masked_loss: 3.1295\n",
      "Epoch 16/100\n",
      "100/100 [==============================] - 6s 63ms/step - loss: 3.1650 - masked_acc: 0.4380 - masked_loss: 3.1650 - val_loss: 3.1459 - val_masked_acc: 0.4412 - val_masked_loss: 3.1459\n",
      "Epoch 17/100\n",
      "100/100 [==============================] - 6s 62ms/step - loss: 3.1014 - masked_acc: 0.4461 - masked_loss: 3.1014 - val_loss: 3.0819 - val_masked_acc: 0.4483 - val_masked_loss: 3.0819\n",
      "Epoch 18/100\n",
      "100/100 [==============================] - 6s 61ms/step - loss: 3.0658 - masked_acc: 0.4501 - masked_loss: 3.0658 - val_loss: 3.1155 - val_masked_acc: 0.4461 - val_masked_loss: 3.1155\n",
      "Epoch 19/100\n",
      "100/100 [==============================] - 6s 58ms/step - loss: 3.0612 - masked_acc: 0.4489 - masked_loss: 3.0612 - val_loss: 3.0175 - val_masked_acc: 0.4554 - val_masked_loss: 3.0175\n",
      "Epoch 20/100\n",
      "100/100 [==============================] - 6s 61ms/step - loss: 3.0045 - masked_acc: 0.4557 - masked_loss: 3.0045 - val_loss: 3.0178 - val_masked_acc: 0.4551 - val_masked_loss: 3.0178\n",
      "Epoch 21/100\n",
      "100/100 [==============================] - 6s 60ms/step - loss: 2.9853 - masked_acc: 0.4602 - masked_loss: 2.9853 - val_loss: 2.9886 - val_masked_acc: 0.4603 - val_masked_loss: 2.9886\n",
      "Epoch 22/100\n",
      "100/100 [==============================] - 6s 59ms/step - loss: 2.9539 - masked_acc: 0.4629 - masked_loss: 2.9539 - val_loss: 2.9670 - val_masked_acc: 0.4657 - val_masked_loss: 2.9670\n",
      "Epoch 23/100\n",
      "100/100 [==============================] - 6s 65ms/step - loss: 2.9321 - masked_acc: 0.4647 - masked_loss: 2.9321 - val_loss: 2.9653 - val_masked_acc: 0.4595 - val_masked_loss: 2.9653\n",
      "Epoch 24/100\n",
      "100/100 [==============================] - 6s 64ms/step - loss: 2.9295 - masked_acc: 0.4675 - masked_loss: 2.9295 - val_loss: 2.8912 - val_masked_acc: 0.4760 - val_masked_loss: 2.8912\n",
      "Epoch 25/100\n",
      "100/100 [==============================] - 6s 59ms/step - loss: 2.8973 - masked_acc: 0.4702 - masked_loss: 2.8973 - val_loss: 2.8793 - val_masked_acc: 0.4696 - val_masked_loss: 2.8793\n",
      "Epoch 26/100\n",
      "100/100 [==============================] - 6s 58ms/step - loss: 2.8830 - masked_acc: 0.4719 - masked_loss: 2.8830 - val_loss: 2.8501 - val_masked_acc: 0.4807 - val_masked_loss: 2.8501\n",
      "Epoch 27/100\n",
      "100/100 [==============================] - 6s 62ms/step - loss: 2.8618 - masked_acc: 0.4734 - masked_loss: 2.8618 - val_loss: 2.9230 - val_masked_acc: 0.4654 - val_masked_loss: 2.9230\n",
      "Epoch 28/100\n",
      "100/100 [==============================] - 6s 64ms/step - loss: 2.8424 - masked_acc: 0.4763 - masked_loss: 2.8424 - val_loss: 2.8141 - val_masked_acc: 0.4802 - val_masked_loss: 2.8141\n",
      "Epoch 29/100\n",
      "100/100 [==============================] - 6s 59ms/step - loss: 2.8358 - masked_acc: 0.4771 - masked_loss: 2.8358 - val_loss: 2.8561 - val_masked_acc: 0.4762 - val_masked_loss: 2.8561\n",
      "Epoch 30/100\n",
      "100/100 [==============================] - 6s 62ms/step - loss: 2.8107 - masked_acc: 0.4819 - masked_loss: 2.8107 - val_loss: 2.8408 - val_masked_acc: 0.4758 - val_masked_loss: 2.8408\n",
      "Epoch 31/100\n",
      "100/100 [==============================] - 6s 59ms/step - loss: 2.8029 - masked_acc: 0.4832 - masked_loss: 2.8029 - val_loss: 2.8448 - val_masked_acc: 0.4792 - val_masked_loss: 2.8448\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(\n",
    "    train_ds.repeat(), \n",
    "    epochs=100,\n",
    "    steps_per_epoch = 100,\n",
    "    validation_data=val_ds,\n",
    "    validation_steps = 20,\n",
    "    callbacks=[\n",
    "        tf.keras.callbacks.EarlyStopping(patience=3)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:09.861393Z",
     "iopub.status.busy": "2023-12-07T12:24:09.861108Z",
     "iopub.status.idle": "2023-12-07T12:24:10.028516Z",
     "shell.execute_reply": "2023-12-07T12:24:10.027895Z"
    },
    "id": "38rLdlmtQHCm"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fe676bb1d50>"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'], label='loss')\n",
    "plt.plot(history.history['val_loss'], label='val_loss')\n",
    "plt.ylim([0, max(plt.ylim())])\n",
    "plt.xlabel('Epoch #')\n",
    "plt.ylabel('CE/token')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:10.031621Z",
     "iopub.status.busy": "2023-12-07T12:24:10.031364Z",
     "iopub.status.idle": "2023-12-07T12:24:10.186521Z",
     "shell.execute_reply": "2023-12-07T12:24:10.185872Z"
    },
    "id": "KkhXRASNG80_"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fe4cc701480>"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['masked_acc'], label='accuracy')\n",
    "plt.plot(history.history['val_masked_acc'], label='val_accuracy')\n",
    "plt.ylim([0, max(plt.ylim())])\n",
    "plt.xlabel('Epoch #')\n",
    "plt.ylabel('CE/token')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20/20 [==============================] - 1s 26ms/step - loss: 2.7699 - masked_acc: 0.4848 - masked_loss: 2.7699\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 2.7698731422424316,\n",
       " 'masked_acc': 0.48475804924964905,\n",
       " 'masked_loss': 2.7698731422424316}"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# evaluate the model on the test set \n",
    "model.evaluate(test_ds, steps=20, return_dict=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mU3Ce8M6I3rz"
   },
   "source": [
    "### Translate\n",
    "\n",
    "Now that the model is trained, implement a function to execute the full `text => text` translation. This code is basically identical to the [inference example](#inference) in the [decoder section](#the_decoder), but this also captures the attention weights."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:10.189649Z",
     "iopub.status.busy": "2023-12-07T12:24:10.189411Z",
     "iopub.status.idle": "2023-12-07T12:24:10.195285Z",
     "shell.execute_reply": "2023-12-07T12:24:10.194684Z"
    },
    "id": "mmgYPCVgEwp_"
   },
   "outputs": [],
   "source": [
    "#@title\n",
    "@Translator.add_method\n",
    "def translate(self,\n",
    "              texts, *,\n",
    "              max_length=50,\n",
    "              temperature=0.0):\n",
    "  # Process the input texts\n",
    "  context = self.encoder.convert_input(texts)\n",
    "  batch_size = tf.shape(texts)[0]\n",
    "\n",
    "  # Setup the loop inputs\n",
    "  tokens = []\n",
    "  attention_weights = []\n",
    "  next_token, done, state = self.decoder.get_initial_state(context)\n",
    "\n",
    "  for _ in range(max_length):\n",
    "    # Generate the next token\n",
    "    next_token, done, state = self.decoder.get_next_token(\n",
    "        context, next_token, done,  state, temperature)\n",
    "        \n",
    "    # Collect the generated tokens\n",
    "    tokens.append(next_token)\n",
    "    attention_weights.append(self.decoder.last_attention_weights)\n",
    "    \n",
    "    if tf.executing_eagerly() and tf.reduce_all(done):\n",
    "      break\n",
    "\n",
    "  # Stack the lists of tokens and attention weights.\n",
    "  tokens = tf.concat(tokens, axis=-1)   # t*[(batch 1)] -> (batch, t)\n",
    "  self.last_attention_weights = tf.concat(attention_weights, axis=1)  # t*[(batch 1 s)] -> (batch, t s)\n",
    "\n",
    "  result = self.decoder.tokens_to_text(tokens)\n",
    "  return result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "U4XufRntbbva"
   },
   "source": [
    "Here are the two helper methods, used above, to convert tokens to text, and to get the next token:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:10.198316Z",
     "iopub.status.busy": "2023-12-07T12:24:10.198091Z",
     "iopub.status.idle": "2023-12-07T12:24:10.384098Z",
     "shell.execute_reply": "2023-12-07T12:24:10.383514Z"
    },
    "id": "E5hqvbR5FUCD"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'is you still at home ? '"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result = model.translate(['Est tu encore à la maison ?'])\n",
    "result[0].numpy().decode()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wQ1iU63cVgfs"
   },
   "source": [
    "Use that to generate the attention plot:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:10.387418Z",
     "iopub.status.busy": "2023-12-07T12:24:10.387171Z",
     "iopub.status.idle": "2023-12-07T12:24:10.393730Z",
     "shell.execute_reply": "2023-12-07T12:24:10.392990Z"
    },
    "id": "s5hQWlbN3jGF"
   },
   "outputs": [],
   "source": [
    "#@title\n",
    "@Translator.add_method\n",
    "def plot_attention(self, text, **kwargs):\n",
    "  assert isinstance(text, str)\n",
    "  output = self.translate([text], **kwargs)\n",
    "  output = output[0].numpy().decode()\n",
    "\n",
    "  attention = self.last_attention_weights[0]\n",
    "\n",
    "  context = tf_lower_and_split_punct(text)\n",
    "  context = context.numpy().decode().split()\n",
    "\n",
    "  output = tf_lower_and_split_punct(output)\n",
    "  output = output.numpy().decode().split()[1:]\n",
    "\n",
    "  fig = plt.figure(figsize=(10, 10))\n",
    "  ax = fig.add_subplot(1, 1, 1)\n",
    "\n",
    "  ax.matshow(attention, cmap='viridis', vmin=0.0)\n",
    "\n",
    "  fontdict = {'fontsize': 14}\n",
    "\n",
    "  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)\n",
    "  ax.set_yticklabels([''] + output, fontdict=fontdict)\n",
    "\n",
    "  ax.xaxis.set_major_locator(ticker.MultipleLocator(1))\n",
    "  ax.yaxis.set_major_locator(ticker.MultipleLocator(1))\n",
    "\n",
    "  ax.set_xlabel('Input text')\n",
    "  ax.set_ylabel('Output text')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:10.397174Z",
     "iopub.status.busy": "2023-12-07T12:24:10.396757Z",
     "iopub.status.idle": "2023-12-07T12:24:10.808896Z",
     "shell.execute_reply": "2023-12-07T12:24:10.808187Z"
    },
    "id": "rrGawQv2eiA4"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_353013/3355722706.py:23: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.\n",
      "  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)\n",
      "/tmp/ipykernel_353013/3355722706.py:24: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.\n",
      "  ax.set_yticklabels([''] + output, fontdict=fontdict)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model.plot_attention('Est tu encore à la maison ?')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JHBdOf9duumm"
   },
   "source": [
    "Translate a few more sentences and plot them:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:10.812417Z",
     "iopub.status.busy": "2023-12-07T12:24:10.812113Z",
     "iopub.status.idle": "2023-12-07T12:24:11.211255Z",
     "shell.execute_reply": "2023-12-07T12:24:11.210549Z"
    },
    "id": "flT0VlQZK11s"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 219 ms, sys: 36.7 ms, total: 255 ms\n",
      "Wall time: 229 ms\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_353013/3355722706.py:23: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.\n",
      "  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)\n",
      "/tmp/ipykernel_353013/3355722706.py:24: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.\n",
      "  ax.set_yticklabels([''] + output, fontdict=fontdict)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%time\n",
    "# This is my life.\n",
    "model.plot_attention('C\\'est ma vie.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:11.214924Z",
     "iopub.status.busy": "2023-12-07T12:24:11.214656Z",
     "iopub.status.idle": "2023-12-07T12:24:11.618813Z",
     "shell.execute_reply": "2023-12-07T12:24:11.618101Z"
    },
    "id": "t-fPYP_9K8xa"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 197 ms, sys: 49.2 ms, total: 246 ms\n",
      "Wall time: 217 ms\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_353013/3355722706.py:23: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.\n",
      "  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)\n",
      "/tmp/ipykernel_353013/3355722706.py:24: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.\n",
      "  ax.set_yticklabels([''] + output, fontdict=fontdict)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%time\n",
    " # Try to find out.'\n",
    "model.plot_attention('Essaye de trouver.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rA3xI3NzrRJt"
   },
   "source": [
    "The short sentences often work well, but if the input is too long the model literally loses focus and stops providing reasonable predictions. There are two main reasons for this:\n",
    "\n",
    "1. The model was trained with teacher-forcing feeding the correct token at each step, regardless of the model's predictions. The model could be made more robust if it were sometimes fed its own predictions.\n",
    "2. The model only has access to its previous output through the RNN state. If the RNN state looses track of where it was in the context sequence there's no way for the model to recover. [Transformers](transformer.ipynb) improve on this by letting the decoder look at what it has output so far."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Vtz6QBoGWqT2"
   },
   "source": [
    "The raw data is sorted by length, so try translating the longest sequence:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:11.622335Z",
     "iopub.status.busy": "2023-12-07T12:24:11.622091Z",
     "iopub.status.idle": "2023-12-07T12:24:11.626087Z",
     "shell.execute_reply": "2023-12-07T12:24:11.625436Z"
    },
    "id": "-FUHFLEvSMbG"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Expected output:\n",
      " But now she's not wearing bangles.\n"
     ]
    }
   ],
   "source": [
    "long_text = context_raw[-1]\n",
    "\n",
    "import textwrap\n",
    "print('Expected output:\\n', '\\n'.join(textwrap.wrap(target_raw[-1])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:11.629569Z",
     "iopub.status.busy": "2023-12-07T12:24:11.628778Z",
     "iopub.status.idle": "2023-12-07T12:24:12.814050Z",
     "shell.execute_reply": "2023-12-07T12:24:12.813276Z"
    },
    "id": "lDa_8NaN_RUy"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_353013/3355722706.py:23: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.\n",
      "  ax.set_xticklabels([''] + context, fontdict=fontdict, rotation=90)\n",
      "/tmp/ipykernel_353013/3355722706.py:24: UserWarning: set_ticklabels() should only be used with a fixed number of ticks, i.e. after set_ticks() or using a FixedLocator.\n",
      "  ax.set_yticklabels([''] + output, fontdict=fontdict)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model.plot_attention(long_text)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "PToqG3GiIUPM"
   },
   "source": [
    "The `translate` function works on batches, so if you have multiple texts to translate you can pass them all at once, which is much more efficient than translating them one at a time:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_content_after_end(array):\n",
    "    try:\n",
    "        index = np.where(array == '[END]')[0][0]  # Find the index of the first occurrence of [END]\n",
    "        return array[:index]\n",
    "    except IndexError:\n",
    "        return array"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we split the test_raw into candidates_to_translate and references to translate against. The candidates are translated and the references are kept as is. The BLEU score is then calculated between the candidates and references."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34206, 34206)"
      ]
     },
     "execution_count": 214,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def process_ids(context_ids, targ_in_ids, targ_out_ids=None):\n",
    "    context_tokens = context_vocab[context_ids.numpy()]\n",
    "    context = ' '.join(context_tokens)\n",
    "\n",
    "    target_in_tokens = target_vocab[targ_in_ids.numpy()]\n",
    "    target_in = ' '.join(target_in_tokens)\n",
    "\n",
    "    return context, target_in\n",
    "\n",
    "def process_batch(ex_context_tok, ex_tar_in, ex_tar_out):\n",
    "    batch_size = tf.shape(ex_context_tok)[0]\n",
    "    return tf.map_fn(lambda i: process_ids(ex_context_tok[i], ex_tar_in[i]), tf.range(batch_size), dtype=(tf.string, tf.string))\n",
    "\n",
    "references_to_keep = []\n",
    "candidates_to_translate = []\n",
    "\n",
    "for (ex_context_tok, ex_tar_in), ex_tar_out in test_ds:\n",
    "    batch_contexts, batch_targets = process_batch(ex_context_tok, ex_tar_in, ex_tar_out)\n",
    "    candidates_to_translate.extend(candidate.decode() for candidate in batch_contexts.numpy())\n",
    "    references_to_keep.extend(target.decode() for target in batch_targets.numpy())\n",
    "    \n",
    "len(references_to_keep), len(candidates_to_translate)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We create a function to batch translate sentences, to fasten up the process."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_special_tokens(text):\n",
    "    text = text.replace('[START]', '')\n",
    "    text = text.replace('[END]', '')\n",
    "    return text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "metadata": {},
   "outputs": [],
   "source": [
    "def translate_on_batch(model, candidates_to_translate, batch_size=128):\n",
    "    candidates = []\n",
    "    for i in range(0, len(candidates_to_translate), batch_size):\n",
    "        result = model.translate(candidates_to_translate[i:i+batch_size])\n",
    "        candidates.extend([decoded_sentence.numpy().decode('utf-8') for decoded_sentence in result])\n",
    "    return candidates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_values_to_test = len(references_to_keep)\n",
    "references = references_to_keep[:n_values_to_test]\n",
    "candidates = translate_on_batch(model, candidates_to_translate[:n_values_to_test])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we also compute the length of each sentence and the bleu score on each sentence for later use."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = []\n",
    "for i in range(n_values_to_test):\n",
    "    data.append({\n",
    "            'Context': candidates_to_translate[i],\n",
    "            'Reference': references[i], \n",
    "            'Candidate': candidates[i], \n",
    "            'length': len(references[i].split())\n",
    "        })"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we want to compute the bleu score on length buckets, we split the data based on the reference length."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dataset_1_5: 894 samples\n",
      "dataset_6_10: 5362 samples\n",
      "dataset_11_15: 6158 samples\n",
      "dataset_16_20: 5733 samples\n",
      "dataset_21_30: 8689 samples\n",
      "dataset_31_40: 4383 samples\n",
      "dataset_41_60: 2514 samples\n",
      "dataset_61_inf: 473 samples\n"
     ]
    }
   ],
   "source": [
    "# split into small dataset based on the sentences length\n",
    "length_ranges = [(1, 5), (6, 10), (11, 15), (16, 20), (21, 30), (31, 40), (41, 60), (61, float('inf'))]\n",
    "\n",
    "small_datasets = {}\n",
    "for min_len, max_len in length_ranges:\n",
    "    filtered_examples = [example for example in data if example['length'] >= min_len and example['length'] <= max_len]\n",
    "    small_datasets[f'dataset_{min_len}_{max_len}'] = filtered_examples\n",
    "\n",
    "# Access the smaller datasets\n",
    "samples_per_range = []\n",
    "for key, dataset in small_datasets.items():\n",
    "    samples_per_range.append(len(dataset))\n",
    "    print(f\"{key}: {len(dataset)} samples\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_corpus_bleu(references, candidates):\n",
    "    if len(references) != len(candidates):\n",
    "        raise ValueError('The number of references and candidates must be the same :', len(references), len(candidates))\n",
    "    \n",
    "    if len(references) == 0: return 0.0\n",
    "    \n",
    "    reference_tokens = [[ref] for ref in references]\n",
    "    return corpus_bleu(reference_tokens, candidates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dataset_1_5: 0.6323\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dataset_6_10: 0.5945\n",
      "dataset_11_15: 0.5718\n",
      "dataset_16_20: 0.5589\n",
      "dataset_21_30: 0.5403\n",
      "dataset_31_40: 0.5258\n",
      "dataset_41_60: 0.4944\n",
      "dataset_61_inf: 0.3706\n"
     ]
    }
   ],
   "source": [
    "bleu_scores = []\n",
    "for key, dataset in small_datasets.items():\n",
    "    refs = [example['Reference'] for example in dataset]\n",
    "    cands = [example['Candidate'] for example in dataset]\n",
    "    \n",
    "    corpus_bleu_score = compute_corpus_bleu(refs, cands)\n",
    "    bleu_scores.append(corpus_bleu_score)\n",
    "    \n",
    "    print(f\"{key}: {corpus_bleu_score:.4f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "metadata": {},
   "outputs": [],
   "source": [
    "overall_bleu_score = corpus_bleu([[ref] for ref in references], candidates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5415764479170317"
      ]
     },
     "execution_count": 233,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "overall_bleu_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'we used our [UNK] and our networks to obtain donations and [UNK] [UNK] '"
      ]
     },
     "execution_count": 234,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.translate([\"nous avons utilise nos connexions et nos reseaux pour obtenir des dons sous forme de vêtements ou d'aliments\"])[0].numpy().decode()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x700 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.patches as mpatches\n",
    "\n",
    "plt.figure(figsize=(15, 7))\n",
    "\n",
    "colors = ['b', 'g', 'r', 'c', 'm', 'y', 'k']  # List of colors for each bar\n",
    "bar_plot = plt.bar([f'{start}-{end}' for start, end in length_ranges], bleu_scores, color=colors, alpha=0.7, label='BLEU Score')\n",
    "\n",
    "# Add \"All\" bar with legend\n",
    "all_bar = plt.bar(\"All\", overall_bleu_score, color='k', alpha=0.7)\n",
    "\n",
    "# Create a dummy handle for the \"All\" bar\n",
    "all_patch = mpatches.Patch(color='k', label=f'Sample = {len(candidates)}')\n",
    "legend_labels = [f'Sample = {value}' for value in samples_per_range]\n",
    "\n",
    "# Include the dummy handle in the legend\n",
    "plt.legend(handles=[*bar_plot, all_patch], labels=legend_labels + [f'Sample = {len(candidates)}'], loc='upper right', title='Samples per range')\n",
    "\n",
    "plt.xlabel('Word Count Range')\n",
    "plt.ylabel('BLEU Score')\n",
    "\n",
    "plt.title('BLEU Score and Number of Samples Based on Word Count Range')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'it is very beautiful today '"
      ]
     },
     "execution_count": 242,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "translation = model.translate(['il fait très beau aujourd\\'hui'])\n",
    "translation[0].numpy().decode()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "uvhMqIw26Bwd"
   },
   "source": [
    "So overall this text generation function mostly gets the job done, but so you've only used it here in python with eager execution. Let's try to export it next:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "X4POAuUgLxLv"
   },
   "source": [
    "### Export"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "S-6cFyqeUPQm"
   },
   "source": [
    "If you want to export this model you'll need to wrap the `translate` method in a `tf.function`. That implementation will get the job done:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:14.342139Z",
     "iopub.status.busy": "2023-12-07T12:24:14.341897Z",
     "iopub.status.idle": "2023-12-07T12:24:14.346466Z",
     "shell.execute_reply": "2023-12-07T12:24:14.345818Z"
    },
    "id": "fNhGwQaVKIAy"
   },
   "outputs": [],
   "source": [
    "class Export(tf.Module):\n",
    "  def __init__(self, model):\n",
    "    self.model = model\n",
    "\n",
    "  @tf.function(input_signature=[tf.TensorSpec(dtype=tf.string, shape=[None])])\n",
    "  def translate(self, inputs):\n",
    "    return self.model.translate(inputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:14.349374Z",
     "iopub.status.busy": "2023-12-07T12:24:14.348916Z",
     "iopub.status.idle": "2023-12-07T12:24:14.352239Z",
     "shell.execute_reply": "2023-12-07T12:24:14.351614Z"
    },
    "id": "5Tjqs9FzNwW5"
   },
   "outputs": [],
   "source": [
    "export = Export(model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "fkccvHDvXCa8"
   },
   "source": [
    "Run the `tf.function` once to compile it:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "inputs = [\n",
    "    'Est tu encore à la maison ?',\n",
    "    'C\\'est ma vie.',\n",
    "    'Essaye de trouver.',\n",
    "    'Je suis un homme.',\n",
    "    'Je suis une femme.',\n",
    "    'Je suis un garçon.'\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:24:14.354869Z",
     "iopub.status.busy": "2023-12-07T12:24:14.354649Z",
     "iopub.status.idle": "2023-12-07T12:25:11.701084Z",
     "shell.execute_reply": "2023-12-07T12:25:11.700336Z"
    },
    "id": "_NzrixLvVBjQ"
   },
   "outputs": [],
   "source": [
    "%%time\n",
    "_ = export.translate(tf.constant(inputs))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:25:11.704503Z",
     "iopub.status.busy": "2023-12-07T12:25:11.704000Z",
     "iopub.status.idle": "2023-12-07T12:25:11.841253Z",
     "shell.execute_reply": "2023-12-07T12:25:11.840595Z"
    },
    "id": "USJdu00tVFbd"
   },
   "outputs": [],
   "source": [
    "%%time\n",
    "result = export.translate(tf.constant(inputs))\n",
    "\n",
    "print(result[0].numpy().decode())\n",
    "print(result[1].numpy().decode())\n",
    "print(result[2].numpy().decode())\n",
    "print(result[3].numpy().decode())\n",
    "print(result[4].numpy().decode())\n",
    "print(result[5].numpy().decode())\n",
    "print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "NP2dNtEXJPEL"
   },
   "source": [
    "Now that the function has been traced it can be exported using `saved_model.save`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:25:11.844840Z",
     "iopub.status.busy": "2023-12-07T12:25:11.844301Z",
     "iopub.status.idle": "2023-12-07T12:26:22.908655Z",
     "shell.execute_reply": "2023-12-07T12:26:22.907946Z"
    },
    "id": "OyvxT5V0_X5B"
   },
   "outputs": [],
   "source": [
    "%%time\n",
    "tf.saved_model.save(export, '../models/nmt_with_attention_tensorflow',\n",
    "                    signatures={'serving_default': export.translate})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:26:22.914045Z",
     "iopub.status.busy": "2023-12-07T12:26:22.913795Z",
     "iopub.status.idle": "2023-12-07T12:27:11.535146Z",
     "shell.execute_reply": "2023-12-07T12:27:11.534384Z"
    },
    "id": "-I0j3i3ekOba"
   },
   "outputs": [],
   "source": [
    "%%time\n",
    "reloaded = tf.saved_model.load('../models/nmt_with_attention_tensorflow')\n",
    "_ = reloaded.translate(tf.constant(inputs)) #warmup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-07T12:27:11.538688Z",
     "iopub.status.busy": "2023-12-07T12:27:11.538405Z",
     "iopub.status.idle": "2023-12-07T12:27:11.681515Z",
     "shell.execute_reply": "2023-12-07T12:27:11.680681Z"
    },
    "id": "GXZF__FZXJCm"
   },
   "outputs": [],
   "source": [
    "%%time\n",
    "result = reloaded.translate(tf.constant(inputs))\n",
    "\n",
    "print(result[0].numpy().decode())\n",
    "print(result[1].numpy().decode())\n",
    "print(result[2].numpy().decode())\n",
    "print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "RTe5P5ioMJwN"
   },
   "source": [
    "## Next steps\n",
    "\n",
    "* [Download a different dataset](http://www.manythings.org/anki/) to experiment with translations, for example, English to German, or English to French.\n",
    "* Experiment with training on a larger dataset, or using more epochs.\n",
    "* Try the [transformer tutorial](transformer.ipynb) which implements a similar translation task but uses transformer layers instead of RNNs. This version also uses a `text.BertTokenizer` to implement word-piece tokenization.\n",
    "* Visit the [`tensorflow_addons.seq2seq` tutorial](https://www.tensorflow.org/addons/tutorials/networks_seq2seq_nmt), which demonstrates a higher-level functionality for implementing this sort of sequence-to-sequence model, such as `seq2seq.BeamSearchDecoder`."
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "last_runtime": {
    "build_target": "//learning/deepmind/public/tools/ml_python:ml_notebook",
    "kind": "private"
   },
   "name": "nmt_with_attention.ipynb",
   "private_outputs": true,
   "provenance": []
  },
  "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}