Mod2_Capstone_Project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vsX0L1sG1iZj"
   },
   "source": [
    "# Capstone Project\n",
    "## Neural translation model\n",
    "### Instructions\n",
    "\n",
    "In this notebook, you will create a neural network that translates from English to German. You will use concepts from throughout this course, including building more flexible model architectures, freezing layers, data processing pipeline and sequence modelling.\n",
    "\n",
    "This project is peer-assessed. Within this notebook you will find instructions in each section for how to complete the project. Pay close attention to the instructions as the peer review will be carried out according to a grading rubric that checks key parts of the project instructions. Feel free to add extra cells into the notebook as required.\n",
    "\n",
    "### How to submit\n",
    "\n",
    "When you have completed the Capstone project notebook, you will submit a pdf of the notebook for peer review. First ensure that the notebook has been fully executed from beginning to end, and all of the cell outputs are visible. This is important, as the grading rubric depends on the reviewer being able to view the outputs of your notebook. Save the notebook as a pdf (you could download the notebook with File -> Download .ipynb, open the notebook locally, and then File -> Download as -> PDF via LaTeX), and then submit this pdf for review.\n",
    "\n",
    "### Let's get started!\n",
    "\n",
    "We'll start by running some imports, and loading the dataset. For this project you are free to make further imports throughout the notebook as you wish."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "2VyTvxPN1iZn"
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import tensorflow_hub as hub\n",
    "import unicodedata\n",
    "import re\n",
    "from IPython.display import Image\n",
    "\n",
    "import random\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tensorflow.keras.layers import Layer, concatenate, Input, Masking, LSTM, Embedding, Dense\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras.preprocessing.text import Tokenizer\n",
    "from tensorflow.keras.preprocessing.sequence import pad_sequences"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wuIi4lAR1iZt"
   },
   "source": [
    "For the capstone project, you will use a language dataset from http://www.manythings.org/anki/ to build a neural translation model. This dataset consists of over 200,000 pairs of sentences in English and German. In order to make the training quicker, we will restrict to our dataset to 20,000 pairs. Feel free to change this if you wish - the size of the dataset used is not part of the grading rubric.\n",
    "\n",
    "Your goal is to develop a neural translation model from English to German, making use of a pre-trained English word embedding module."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "pi9Dq6vv3FVO"
   },
   "source": [
    "#### Import the data\n",
    "\n",
    "The dataset is available for download as a zip file at the following link:\n",
    "\n",
    "https://drive.google.com/open?id=1KczOciG7sYY7SB9UlBeRP1T9659b121Q\n",
    "\n",
    "You should store the unzipped folder in Drive for use in this Colab notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "Nw99tEEQ3bKL",
    "outputId": "ca8bb90f-3cff-46b8-db36-86f2bc523a68"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mounted at /content/gdrive\n"
     ]
    }
   ],
   "source": [
    "# Run this cell to connect to your Drive folder\n",
    "\n",
    "from google.colab import drive\n",
    "drive.mount('/content/gdrive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "id": "o8PetPpw1iZu"
   },
   "outputs": [],
   "source": [
    "# Run this cell to load the dataset\n",
    "\n",
    "NUM_EXAMPLES = 20000\n",
    "data_examples = []\n",
    "with open('/content/deu.txt', 'r', encoding='utf8') as f:\n",
    "    for line in f.readlines():\n",
    "        if len(data_examples) < NUM_EXAMPLES:\n",
    "            data_examples.append(line)\n",
    "        else:\n",
    "            break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "JumLjJ631iZy"
   },
   "outputs": [],
   "source": [
    "# These functions preprocess English and German sentences\n",
    "\n",
    "def unicode_to_ascii(s):\n",
    "    return ''.join(c for c in unicodedata.normalize('NFD', s) if unicodedata.category(c) != 'Mn')\n",
    "\n",
    "def preprocess_sentence(sentence):\n",
    "    sentence = sentence.lower().strip()\n",
    "    sentence = re.sub(r\"ü\", 'ue', sentence)\n",
    "    sentence = re.sub(r\"ä\", 'ae', sentence)\n",
    "    sentence = re.sub(r\"ö\", 'oe', sentence)\n",
    "    sentence = re.sub(r'ß', 'ss', sentence)\n",
    "\n",
    "    sentence = unicode_to_ascii(sentence)\n",
    "    sentence = re.sub(r\"([?.!,])\", r\" \\1 \", sentence)\n",
    "    sentence = re.sub(r\"[^a-z?.!,']+\", \" \", sentence)\n",
    "    sentence = re.sub(r'[\" \"]+', \" \", sentence)\n",
    "\n",
    "    return sentence.strip()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "XFJap-TW1iZ2"
   },
   "source": [
    "#### The custom translation model\n",
    "The following is a schematic of the custom translation model architecture you will develop in this project."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 648
    },
    "id": "gBF1K2JN4RFJ",
    "outputId": "17a5f335-161e-4cf5-966b-5ba89f35affa"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Run this cell to download and view a schematic diagram for the neural translation model\n",
    "\n",
    "!wget -q -O neural_translation_model.png --no-check-certificate \"https://docs.google.com/uc?export=download&id=1XsS1VlXoaEo-RbYNilJ9jcscNZvsSPmd\"\n",
    "Image(\"neural_translation_model.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "0fP7P-yK4RS7"
   },
   "source": [
    "The custom model consists of an encoder RNN and a decoder RNN. The encoder takes words of an English sentence as input, and uses a pre-trained word embedding to embed the words into a 128-dimensional space. To indicate the end of the input sentence, a special end token (in the same 128-dimensional space) is passed in as an input. This token is a TensorFlow Variable that is learned in the training phase (unlike the pre-trained word embedding, which is frozen).\n",
    "\n",
    "The decoder RNN takes the internal state of the encoder network as its initial state. A start token is passed in as the first input, which is embedded using a learned German word embedding. The decoder RNN then makes a prediction for the next German word, which during inference is then passed in as the following input, and this process is repeated until the special `<end>` token is emitted from the decoder."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "z70nu6_01iZ3"
   },
   "source": [
    "## 1. Text preprocessing\n",
    "* Create separate lists of English and German sentences, and preprocess them using the `preprocess_sentence` function provided for you above.\n",
    "* Add a special `\"<start>\"` and `\"<end>\"` token to the beginning and end of every German sentence.\n",
    "* Use the Tokenizer class from the `tf.keras.preprocessing.text` module to tokenize the German sentences, ensuring that no character filters are applied. _Hint: use the Tokenizer's \"filter\" keyword argument._\n",
    "* Print out at least 5 randomly chosen examples of (preprocessed) English and German sentence pairs. For the German sentence, print out the text (with start and end tokens) as well as the tokenized sequence.\n",
    "* Pad the end of the tokenized German sequences with zeros, and batch the complete set of sequences into one numpy array."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "9G20C4bk1iZ4",
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "english_sentence_list = []\n",
    "german_sentence_list = []\n",
    "\n",
    "for sentence in data_examples:\n",
    "\n",
    "    english_sentence_list.append(preprocess_sentence(sentence.split(\"\\t\")[0]))\n",
    "    german_sentences = \"<start>\" + preprocess_sentence(sentence.split(\"\\t\")[1]) + \"<end>\"\n",
    "    german_sentence_list.append(german_sentences)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "Sxwl-1rB1iZ8"
   },
   "outputs": [],
   "source": [
    "def max_length(input_list):\n",
    "\n",
    "    max_length = 0\n",
    "\n",
    "    for item in input_list:\n",
    "\n",
    "        current_length = len(item)\n",
    "        if max_length < current_length:\n",
    "            max_length = current_length\n",
    "\n",
    "    return max_length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "id": "WnIdqZFk1iaA"
   },
   "outputs": [],
   "source": [
    "def tokenise_german_sentences(german_sentence_list):\n",
    "\n",
    "    tokeniser = Tokenizer(filters='')\n",
    "    tokeniser.fit_on_texts(german_sentence_list)\n",
    "\n",
    "    german_sequences = tokeniser.texts_to_sequences(german_sentence_list)\n",
    "    vocab_size = max(tokeniser.word_index.values()) + 1\n",
    "    max_len = max_length(german_sequences)\n",
    "    padded_german_sequences = pad_sequences(german_sequences, maxlen = max_len,padding = 'post', truncating= 'post')\n",
    "    return german_sequences, padded_german_sequences, vocab_size, tokeniser"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "id": "5UlnBdIK1iaE"
   },
   "outputs": [],
   "source": [
    "\n",
    "min_size = min([len(english_sentence_list), len(german_sentence_list)])\n",
    "\n",
    "def print_random_samples(english_sentence_list, german_sentence_list, padded_german_sequences):\n",
    "\n",
    "    for i in range(5):\n",
    "        idx = random.randint(0, min_size-1)\n",
    "        print(\"English sentence: \", english_sentence_list[idx])\n",
    "        print(\"German sentence: \",german_sentence_list[idx])\n",
    "        print(\"German sequence: \",german_sequences[idx])\n",
    "        print(\"===================\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ZGTaqt1t1iaI",
    "outputId": "df8c3dd2-ef70-458e-994b-c459475f164c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "English sentence:  tom was shy .\n",
      "German sentence:  <start>tom war schuechtern .<end>\n",
      "German sequence:  [3, 24, 351, 1]\n",
      "===================\n",
      "English sentence:  tom lost control .\n",
      "German sentence:  <start>tom hat die kontrolle verloren .<end>\n",
      "German sequence:  [3, 14, 35, 2081, 153, 1]\n",
      "===================\n",
      "English sentence:  it won't matter .\n",
      "German sentence:  <start>es wird keine rolle spielen .<end>\n",
      "German sequence:  [22, 49, 112, 1517, 417, 1]\n",
      "===================\n",
      "English sentence:  that's strange .\n",
      "German sentence:  <start>das ist seltsam .<end>\n",
      "German sequence:  [16, 4, 639, 1]\n",
      "===================\n",
      "English sentence:  tom complained .\n",
      "German sentence:  <start>tom hat sich beklagt .<end>\n",
      "German sequence:  [3, 14, 33, 4498, 1]\n",
      "===================\n"
     ]
    }
   ],
   "source": [
    "# tokenised sequence\n",
    "#Print out at least 5 randomly chosen examples of (preprocessed) English and German sentence pairs.\n",
    "#For the German sentence, print out the text (with start and end tokens) as well as the tokenized sequence\n",
    "\n",
    "german_sequences, padded_german_sequences, vocab_size, tokeniser = tokenise_german_sentences(german_sentence_list)\n",
    "print_random_samples(english_sentence_list, german_sentence_list, german_sequences)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "fqiO7HuQ1iaL",
    "outputId": "5ffe9c23-a367-4d57-b99c-0e5f452a37f8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6460\n"
     ]
    }
   ],
   "source": [
    "print(vocab_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "foL7Ihs21iaP"
   },
   "source": [
    "## 2. Prepare the data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "-9rCEE4z1iaQ"
   },
   "source": [
    "#### Load the embedding layer\n",
    "As part of the dataset preproceessing for this project, you will use a pre-trained English word embedding module from TensorFlow Hub. The URL for the module is https://tfhub.dev/google/tf2-preview/nnlm-en-dim128-with-normalization/1.\n",
    "\n",
    "This embedding takes a batch of text tokens in a 1-D tensor of strings as input. It then embeds the separate tokens into a 128-dimensional space.\n",
    "\n",
    "The code to load and test the embedding layer is provided for you below.\n",
    "\n",
    "**NB:** this model can also be used as a sentence embedding module. The module will process each token by removing punctuation and splitting on spaces. It then averages the word embeddings over a sentence to give a single embedding vector. However, we will use it only as a word embedding module, and will pass each word in the input sentence as a separate token."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "id": "ywZgobCh1iaR"
   },
   "outputs": [],
   "source": [
    "# Load embedding module from Tensorflow Hub\n",
    "\n",
    "embedding_layer = hub.KerasLayer(\"https://tfhub.dev/google/tf2-preview/nnlm-en-dim128/1\",\n",
    "                                 output_shape=[128], input_shape=[], dtype=tf.string)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "TiY8QEDp1iaV",
    "outputId": "e8b0f2ff-ac36-4310-ba43-eb21f75897f2"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TensorShape([7, 128])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Test the layer\n",
    "\n",
    "embedding_layer(tf.constant([\"these\", \"aren't\", \"the\", \"droids\", \"you're\", \"looking\", \"for\"])).shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KjuzXc-z1iaY"
   },
   "source": [
    "You should now prepare the training and validation Datasets.\n",
    "\n",
    "* Create a random training and validation set split of the data, reserving e.g. 20% of the data for validation (NB: each English dataset example is a single sentence string, and each German dataset example is a sequence of padded integer tokens).\n",
    "* Load the training and validation sets into a tf.data.Dataset object, passing in a tuple of English and German data for both training and validation sets.\n",
    "* Create a function to map over the datasets that splits each English sentence at spaces. Apply this function to both Dataset objects using the map method. _Hint: look at the tf.strings.split function._\n",
    "* Create a function to map over the datasets that embeds each sequence of English words using the loaded embedding layer/model. Apply this function to both Dataset objects using the map method.\n",
    "* Create a function to filter out dataset examples where the English sentence is greater than or equal to than 13 (embedded) tokens in length. Apply this function to both Dataset objects using the filter method.\n",
    "* Create a function to map over the datasets that pads each English sequence of embeddings with some distinct padding value before the sequence, so that each sequence is length 13. Apply this function to both Dataset objects using the map method. _Hint: look at the tf.pad function. You can extract a Tensor shape using tf.shape; you might also find the tf.math.maximum function useful._\n",
    "* Batch both training and validation Datasets with a batch size of 16.\n",
    "* Print the `element_spec` property for the training and validation Datasets.\n",
    "* Using the Dataset `.take(1)` method, print the shape of the English data example from the training Dataset.\n",
    "* Using the Dataset `.take(1)` method, print the German data example Tensor from the validation Dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "Q-BUJOl_1iaZ",
    "outputId": "e1cf7675-b539-4a68-f669-6eaa48fc2f21"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(TensorSpec(shape=(None, None, 128), dtype=tf.float32, name=None), TensorSpec(shape=(None, 12), dtype=tf.int32, name=None))\n",
      "(TensorSpec(shape=(None, None, 128), dtype=tf.float32, name=None), TensorSpec(shape=(None, 12), dtype=tf.int32, name=None))\n"
     ]
    }
   ],
   "source": [
    "def preprocess_data(english_sentences, german_sequences):\n",
    "    dataset = tf.data.Dataset.from_tensor_slices((english_sentences, german_sequences))\n",
    "    dataset = dataset.map(lambda x, y: (tf.strings.split(x, sep=' '), y))\n",
    "    dataset = dataset.map(lambda x, y: (embedding_layer(x), y))\n",
    "    dataset = dataset.filter(lambda x, y: tf.shape(x)[0] <= 13)\n",
    "    dataset = dataset.map(lambda x, y: (tf.pad(x, tf.constant([[13, 0], [0, 0]]))[-13:, :], y))\n",
    "    dataset = dataset.batch(16)\n",
    "    return dataset\n",
    "\n",
    "# Splitting training data into training and validation sets\n",
    "split = int(len(english_sentence_list) * 0.8)\n",
    "train_english_sentences = english_sentence_list[:split]\n",
    "validation_english_sentences = english_sentence_list[split:]\n",
    "train_german_padded_sequences = padded_german_sequences[:split]\n",
    "validation_german_padded_sequences = padded_german_sequences[split:]\n",
    "\n",
    "# Preprocessing training and validation datasets\n",
    "train_dataset = preprocess_data(train_english_sentences, train_german_padded_sequences)\n",
    "validation_dataset = preprocess_data(validation_english_sentences, validation_german_padded_sequences)\n",
    "\n",
    "# Print element spec for both training and validation sets\n",
    "print(train_dataset.element_spec)\n",
    "print(validation_dataset.element_spec)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4w6rM8Bl1iad",
    "outputId": "8c6cd34a-35c0-4642-93e0-c4dd766b4670"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of English Training Example:  (16, 13, 128)\n",
      "tf.Tensor(\n",
      "[[ 553    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [3430 1114    6    0    0    0    0    0    0    0    0    0]\n",
      " [1115    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [3431    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [3432    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [2459    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [3433    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [1533 3434    6    0    0    0    0    0    0    0    0    0]\n",
      " [2460    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [ 394    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [  86  132    1    0    0    0    0    0    0    0    0    0]\n",
      " [ 553    6    0    0    0    0    0    0    0    0    0    0]\n",
      " [   2  374    1    0    0    0    0    0    0    0    0    0]\n",
      " [   2  350    1    0    0    0    0    0    0    0    0    0]\n",
      " [3435    1    0    0    0    0    0    0    0    0    0    0]\n",
      " [   2 3436   10    1    0    0    0    0    0    0    0    0]], shape=(16, 12), dtype=int32)\n",
      "Shape of German Validation Example:  (16, 12)\n",
      "tf.Tensor(\n",
      "[[  16    4  126  551    1    0    0    0    0    0    0    0]\n",
      " [  16    4   20 5747    1    0    0    0    0    0    0    0]\n",
      " [ 125  438    4  439    1    0    0    0    0    0    0    0]\n",
      " [  75 1512 2986 5748    1    0    0    0    0    0    0    0]\n",
      " [  75 1512    4  364    6    0    0    0    0    0    0    0]\n",
      " [  16 1454    4  395    1    0    0    0    0    0    0    0]\n",
      " [  75 5749    4  954    1    0    0    0    0    0    0    0]\n",
      " [  75  613  143    1    0    0    0    0    0    0    0    0]\n",
      " [  75  613   26  143    1    0    0    0    0    0    0    0]\n",
      " [ 125 1683    4  345    1    0    0    0    0    0    0    0]\n",
      " [  16  117    4   40    1    0    0    0    0    0    0    0]\n",
      " [  16  117    4  641    1    0    0    0    0    0    0    0]\n",
      " [  16  117    4   88    1    0    0    0    0    0    0    0]\n",
      " [  75 2423    4  954    1    0    0    0    0    0    0    0]\n",
      " [  75 2425    4  954    1    0    0    0    0    0    0    0]\n",
      " [  75 2425    4  516    1    0    0    0    0    0    0    0]], shape=(16, 12), dtype=int32)\n"
     ]
    }
   ],
   "source": [
    "#get the shape for a padded english example in the train dataset\n",
    "\n",
    "for english_train_example, german_train_example in train_dataset.take(1):\n",
    "    print(\"Shape of English Training Example: \", english_train_example.shape)\n",
    "    print(german_train_example)\n",
    "\n",
    "for english_validation_example, german_validation_example in validation_dataset.take(1):\n",
    "    print(\"Shape of German Validation Example: \", german_validation_example.shape)\n",
    "    print(german_validation_example)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "D7bn3mRs1iaj"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "q0Fdso381ian"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "bGqlxuqZ1ias"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "MdqSXEoX1iav"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "isIYhjq01iay"
   },
   "source": [
    "## 3. Create the custom layer\n",
    "You will now create a custom layer to add the learned end token embedding to the encoder model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 423
    },
    "id": "e22f1Xyh6xvE",
    "outputId": "91f8cde9-a749-474e-bbcf-654294a8e10b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Run this cell to download and view a schematic diagram for the encoder model\n",
    "\n",
    "!wget -q -O neural_translation_model.png --no-check-certificate \"https://docs.google.com/uc?export=download&id=1JrtNOzUJDaOWrK4C-xv-4wUuZaI12sQI\"\n",
    "Image(\"neural_translation_model.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "M6gLIHG81iaz"
   },
   "source": [
    "You should now build the custom layer.\n",
    "* Using layer subclassing, create a custom layer that takes a batch of English data examples from one of the Datasets, and adds a learned embedded ‘end’ token to the end of each sequence.\n",
    "* This layer should create a TensorFlow Variable (that will be learned during training) that is 128-dimensional (the size of the embedding space). _Hint: you may find it helpful in the call method to use the tf.tile function to replicate the end token embedding across every element in the batch._\n",
    "* Using the Dataset `.take(1)` method, extract a batch of English data examples from the training Dataset and print the shape. Test the custom layer by calling the layer on the English data batch Tensor and print the resulting Tensor shape (the layer should increase the sequence length by one)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "id": "yg9hjZz11ia0"
   },
   "outputs": [],
   "source": [
    "class custom_embedding_layer(Layer):\n",
    "    def __init__(self, **kwargs):\n",
    "        super(custom_embedding_layer, self).__init__(**kwargs)\n",
    "        self.token_embedding = tf.Variable(initial_value = tf.random.uniform(shape = (128,)), trainable = True)\n",
    "\n",
    "    def call(self, inputs):\n",
    "        token_embed_altered = tf.reshape(self.token_embedding, (1,1,128))\n",
    "        input_shape = tf.shape(inputs)\n",
    "        end_embed_token = tf.tile(token_embed_altered, [input_shape[0],1,1])\n",
    "        concat_input = concatenate([inputs, end_embed_token], axis=1)\n",
    "        return concat_input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4PwI32O11ia3",
    "outputId": "b126cb64-c89f-4fda-aa7f-924f1ae3b8c3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Earlier English Shape: (16, 13, 128)\n",
      "After Learned Embedded Layer: (16, 14, 128)\n"
     ]
    }
   ],
   "source": [
    "for english, german in train_dataset.take(1):\n",
    "    print(\"Earlier English Shape:\", english.shape)\n",
    "    result = custom_embedding_layer()(english)\n",
    "    print(\"After Learned Embedded Layer:\", result.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "ebjK8PmL1ia6"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "OAd3i4_y1ia-"
   },
   "source": [
    "## 4. Build the encoder network\n",
    "The encoder network follows the schematic diagram above. You should now build the RNN encoder model.\n",
    "* Using the functional API, build the encoder network according to the following spec:\n",
    "    * The model will take a batch of sequences of embedded English words as input, as given by the Dataset objects.\n",
    "    * The next layer in the encoder will be the custom layer you created previously, to add a learned end token embedding to the end of the English sequence.\n",
    "    * This is followed by a Masking layer, with the `mask_value` set to the distinct padding value you used when you padded the English sequences with the Dataset preprocessing above.\n",
    "    * The final layer is an LSTM layer with 512 units, which also returns the hidden and cell states.\n",
    "    * The encoder is a multi-output model. There should be two output Tensors of this model: the hidden state and cell states of the LSTM layer. The output of the LSTM layer is unused.\n",
    "* Using the Dataset `.take(1)` method, extract a batch of English data examples from the training Dataset and test the encoder model by calling it on the English data Tensor, and print the shape of the resulting Tensor outputs.\n",
    "* Print the model summary for the encoder network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "6R2LqbfV1ia_"
   },
   "outputs": [],
   "source": [
    "def encoder_net(input_shape):\n",
    "\n",
    "    input = Input(shape = input_shape)\n",
    "    end_embedding = custom_embedding_layer()(input)\n",
    "    masking_layer = Masking(mask_value = -1)(end_embedding)\n",
    "    lstm, hidden, cell_states = LSTM(512, return_state = True)(masking_layer)\n",
    "    model = Model(inputs = input, outputs = [hidden, cell_states])\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "e5XW6NxL1ibC",
    "outputId": "17c9dd50-21d3-4a28-e78e-febc4131a611"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_2 (InputLayer)        [(None, 13, 128)]         0         \n",
      "                                                                 \n",
      " custom_embedding_layer_2 (c  (None, 14, 128)          128       \n",
      " ustom_embedding_layer)                                          \n",
      "                                                                 \n",
      " masking_1 (Masking)         (None, 14, 128)           0         \n",
      "                                                                 \n",
      " lstm_1 (LSTM)               [(None, 512),             1312768   \n",
      "                              (None, 512),                       \n",
      "                              (None, 512)]                       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 1,312,896\n",
      "Trainable params: 1,312,896\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "encoder_model = encoder_net(input_shape = (13, 128))\n",
    "encoder_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "jEk9ikVh1ibL",
    "outputId": "b31caa99-f8b5-4dd7-8622-7672bd429dd0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(16, 512)\n",
      "(16, 512)\n"
     ]
    }
   ],
   "source": [
    "for english, german in train_dataset.take(1):\n",
    "\n",
    "    hidden, cell_state = encoder_model(english)\n",
    "    print(hidden.shape)\n",
    "    print(cell_state.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KvkzpCeZ1ibR"
   },
   "source": [
    "## 5. Build the decoder network\n",
    "The decoder network follows the schematic diagram below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 501
    },
    "id": "yOjEb7cH7Y4S",
    "outputId": "599916e2-2c63-4c7a-a973-b07b589eadec"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Run this cell to download and view a schematic diagram for the decoder model\n",
    "\n",
    "!wget -q -O neural_translation_model.png --no-check-certificate \"https://docs.google.com/uc?export=download&id=1DTeaXD8tA8RjkpVrB2mr9csSBOY4LQiW\"\n",
    "Image(\"neural_translation_model.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "VPBK8WGL1ibS"
   },
   "source": [
    "You should now build the RNN decoder model.\n",
    "* Using Model subclassing, build the decoder network according to the following spec:\n",
    "    * The initializer should create the following layers:\n",
    "        * An Embedding layer with vocabulary size set to the number of unique German tokens, embedding dimension 128, and set to mask zero values in the input.\n",
    "        * An LSTM layer with 512 units, that returns its hidden and cell states, and also returns sequences.\n",
    "        * A Dense layer with number of units equal to the number of unique German tokens, and no activation function.\n",
    "    * The call method should include the usual `inputs` argument, as well as the additional keyword arguments `hidden_state` and `cell_state`. The default value for these keyword arguments should be `None`.\n",
    "    * The call method should pass the inputs through the Embedding layer, and then through the LSTM layer. If the `hidden_state` and `cell_state` arguments are provided, these should be used for the initial state of the LSTM layer. _Hint: use the_ `initial_state` _keyword argument when calling the LSTM layer on its input._\n",
    "    * The call method should pass the LSTM output sequence through the Dense layer, and return the resulting Tensor, along with the hidden and cell states of the LSTM layer.\n",
    "* Using the Dataset `.take(1)` method, extract a batch of English and German data examples from the training Dataset. Test the decoder model by first calling the encoder model on the English data Tensor to get the hidden and cell states, and then call the decoder model on the German data Tensor and hidden and cell states, and print the shape of the resulting decoder Tensor outputs.\n",
    "* Print the model summary for the decoder network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "id": "l50qhnXD1ibT"
   },
   "outputs": [],
   "source": [
    "class DecoderModel(Model):\n",
    "\n",
    "    def __init__(self):\n",
    "        super(DecoderModel, self).__init__()\n",
    "\n",
    "        self.embedding = Embedding(vocab_size, 128, mask_zero = True)\n",
    "        self.lstm_layer = LSTM(512, return_sequences = True, return_state = True)\n",
    "        self.dense = Dense(vocab_size)\n",
    "\n",
    "    def call(self, inputs, hidden_state = None, cell_state = None):\n",
    "        embedding_output = self.embedding(inputs)\n",
    "        lstm_out, hidden, cell_state = self.lstm_layer(inputs = embedding_output, initial_state = [hidden_state, cell_state])\n",
    "        dense_out = self.dense(lstm_out)\n",
    "\n",
    "        return (dense_out, hidden, cell_state)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "vG5CRIXV1ibi",
    "outputId": "ae8f6389-fa18-4521-9a79-89b7cbc459b3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Output Shape: (16, 12, 6460)\n",
      "Hidden Shape: (16, 512)\n",
      "Cell State Shape: (16, 512)\n"
     ]
    }
   ],
   "source": [
    "decoder_model = DecoderModel()\n",
    "\n",
    "\n",
    "for english, german in train_dataset.take(1):\n",
    "\n",
    "    hidden, cell_state = encoder_model(english)\n",
    "    output, hidden, cell_state = decoder_model(german, hidden, cell_state)\n",
    "    print(\"Output Shape:\", output.shape)\n",
    "    print(\"Hidden Shape:\", hidden.shape)\n",
    "    print(\"Cell State Shape:\", cell_state.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "SvqqT_ET1ibl",
    "outputId": "09b167ad-e114-4852-d411-29585fcc495c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"decoder_model\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " embedding (Embedding)       multiple                  826880    \n",
      "                                                                 \n",
      " lstm_2 (LSTM)               multiple                  1312768   \n",
      "                                                                 \n",
      " dense (Dense)               multiple                  3313980   \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 5,453,628\n",
      "Trainable params: 5,453,628\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "decoder_model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "pST9XGJ81ibo"
   },
   "source": [
    "## 6. Make a custom training loop\n",
    "You should now write a custom training loop to train your custom neural translation model.\n",
    "* Define a function that takes a Tensor batch of German data (as extracted from the training Dataset), and returns a tuple containing German inputs and outputs for the decoder model (refer to schematic diagram above).\n",
    "* Define a function that computes the forward and backward pass for your translation model. This function should take an English input, German input and German output as arguments, and should do the following:\n",
    "    * Pass the English input into the encoder, to get the hidden and cell states of the encoder LSTM.\n",
    "    * These hidden and cell states are then passed into the decoder, along with the German inputs, which returns a sequence of outputs (the hidden and cell state outputs of the decoder LSTM are unused in this function).\n",
    "    * The loss should then be computed between the decoder outputs and the German output function argument.\n",
    "    * The function returns the loss and gradients with respect to the encoder and decoder’s trainable variables.\n",
    "    * Decorate the function with `@tf.function`\n",
    "* Define and run a custom training loop for a number of epochs (for you to choose) that does the following:\n",
    "    * Iterates through the training dataset, and creates decoder inputs and outputs from the German sequences.\n",
    "    * Updates the parameters of the translation model using the gradients of the function above and an optimizer object.\n",
    "    * Every epoch, compute the validation loss on a number of batches from the validation and save the epoch training and validation losses.\n",
    "* Plot the learning curves for loss vs epoch for both training and validation sets.\n",
    "\n",
    "_Hint: This model is computationally demanding to train. The quality of the model or length of training is not a factor in the grading rubric. However, to obtain a better model we recommend using the GPU accelerator hardware on Colab._"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "id": "7hJHbWqs1ibr"
   },
   "outputs": [],
   "source": [
    "def german_data(german_data):\n",
    "\n",
    "    german_input = german_data[:,:-1]\n",
    "    german_output = german_data[:,1:]\n",
    "\n",
    "    return (german_input, german_output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "id": "Jvu4J4-u1ibu"
   },
   "outputs": [],
   "source": [
    "@tf.function\n",
    "\n",
    "def grad_fn(english, german_input, german_output):\n",
    "\n",
    "    with tf.GradientTape() as tape:\n",
    "\n",
    "        hidden, cell_state = encoder_model(english)\n",
    "        german_from_model, hidden, cell_state = decoder_model(german_input, hidden, cell_state)\n",
    "\n",
    "        loss = tf.keras.losses.sparse_categorical_crossentropy(german_output, german_from_model, from_logits = True)\n",
    "\n",
    "        grad_encoder, grad_decoder = tape.gradient(loss, [encoder_model.trainable_variables,decoder_model.trainable_variables])\n",
    "\n",
    "        return (loss, grad_encoder, grad_decoder)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "id": "Vl5H06811ibx"
   },
   "outputs": [],
   "source": [
    "optimizer = tf.keras.optimizers.legacy.Adam(learning_rate=0.001)\n",
    "num_epochs = 3\n",
    "epoch_train_loss = []\n",
    "epoch_val_loss = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "MVWOYJ3l1ib1",
    "outputId": "4c2302d3-7771-4a8d-f077-7b3a456a4603"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Completed epoch 1\n",
      "Completed epoch 2\n",
      "Completed epoch 3\n"
     ]
    }
   ],
   "source": [
    "\n",
    "epoch_train_loss = []\n",
    "epoch_val_loss = []\n",
    "\n",
    "for epoch in range(num_epochs):\n",
    "    train_loss = tf.keras.metrics.Mean()\n",
    "    val_loss = tf.keras.metrics.Mean()\n",
    "\n",
    "    # Training loop\n",
    "    for english, german in train_dataset:\n",
    "        german_in, german_out = german_data(german)\n",
    "\n",
    "        with tf.GradientTape() as tape:\n",
    "            loss_value, grad_encoder, grad_decoder = grad_fn(english, german_in, german_out)\n",
    "        optimizer.apply_gradients(zip(grad_encoder, encoder_model.trainable_variables))\n",
    "        optimizer.apply_gradients(zip(grad_decoder, decoder_model.trainable_variables))\n",
    "\n",
    "        train_loss(loss_value)\n",
    "\n",
    "    epoch_train_loss.append(train_loss.result())\n",
    "\n",
    "    # Validation loop\n",
    "    for english, german in validation_dataset:\n",
    "        german_in, german_out = german_data(german)\n",
    "\n",
    "        hidden, cell_state = encoder_model(english)\n",
    "        german_from_model, hidden, cell_state = decoder_model(german_in, hidden, cell_state)\n",
    "        loss = tf.keras.losses.sparse_categorical_crossentropy(german_out, german_from_model, from_logits=True)\n",
    "        val_loss(loss)\n",
    "\n",
    "    epoch_val_loss.append(val_loss.result())\n",
    "    print(f\"Completed epoch {epoch+1}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 506
    },
    "id": "6VSmBdwT1ib4",
    "outputId": "0ee4059d-6bde-4c5d-9d4d-2ac628ae897b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1500x800 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Plotting Epoch vs loss for both training and validation set\n",
    "\n",
    "fig, axes = plt.subplots(1, 2, figsize=(15, 8), sharex=True)\n",
    "\n",
    "axes[0].set_xlabel(\"Epochs\")\n",
    "axes[0].set_ylabel(\"Loss\")\n",
    "axes[0].set_title('Training Loss vs Epochs')\n",
    "axes[0].plot(epoch_train_loss)\n",
    "\n",
    "axes[1].set_xlabel(\"Epochs\")\n",
    "axes[1].set_ylabel(\"Loss\")\n",
    "axes[1].set_title('Validation Loss vs Epochs')\n",
    "axes[1].plot(epoch_val_loss)\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "dpacst2F1ib7"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "xM2gvBM11ib-"
   },
   "source": [
    "## 7. Use the model to translate\n",
    "Now it's time to put your model into practice! You should run your translation for five randomly sampled English sentences from the dataset. For each sentence, the process is as follows:\n",
    "* Preprocess and embed the English sentence according to the model requirements.\n",
    "* Pass the embedded sentence through the encoder to get the encoder hidden and cell states.\n",
    "* Starting with the special  `\"<start>\"` token, use this token and the final encoder hidden and cell states to get the one-step prediction from the decoder, as well as the decoder’s updated hidden and cell states.\n",
    "* Create a loop to get the next step prediction and updated hidden and cell states from the decoder, using the most recent hidden and cell states. Terminate the loop when the `\"<end>\"` token is emitted, or when the sentence has reached a maximum length.\n",
    "* Decode the output token sequence into German text and print the English text and the model's German translation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "bGnQtE7L1icA",
    "outputId": "f746a433-7951-4fa6-b08d-cb67c997be51"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sample: 0\n",
      "English Sentence: you can't win .\n",
      "Original German: <start>es ist aussichtslos .<end>\n",
      "Trainslated German: ['du', 'hier', '?<end>']\n",
      "=================================\n",
      "Sample: 1\n",
      "English Sentence: stop grumbling .\n",
      "Original German: <start>hoer auf zu murren .<end>\n",
      "Trainslated German: ['du', 'hier', '?<end>']\n",
      "=================================\n",
      "Sample: 2\n",
      "English Sentence: i like fish .\n",
      "Original German: <start>ich mag fisch gern .<end>\n",
      "Trainslated German: ['du', 'hier', '?<end>']\n",
      "=================================\n",
      "Sample: 3\n",
      "English Sentence: who yelled ?\n",
      "Original German: <start>wer schrie ?<end>\n",
      "Trainslated German: ['du', 'hier', '?<end>']\n",
      "=================================\n",
      "Sample: 4\n",
      "English Sentence: tom talks funny .\n",
      "Original German: <start>tom redet lustig .<end>\n",
      "Trainslated German: ['du', 'hier', '?<end>']\n",
      "=================================\n"
     ]
    }
   ],
   "source": [
    "#define the tokens\n",
    "start_token = tf.constant([[1]])\n",
    "end_token = tf.constant([[2]])\n",
    "end_token_val = end_token.numpy()[0][0]\n",
    "#get index to word dictionary\n",
    "index_to_word = tokeniser.index_word\n",
    "\n",
    "random_indexes = np.random.choice(len(english_sentence_list), 5)\n",
    "for idx, e_idx in enumerate(random_indexes):\n",
    "  curr_eng_sentence = english_sentence_list[e_idx]\n",
    "  print(f\"Sample: {idx}\")\n",
    "  print(f\"English Sentence: {curr_eng_sentence}\")\n",
    "  split_eng_sentence = tf.strings.split(curr_eng_sentence, sep=' ')\n",
    "  embedding_out = embedding_layer(split_eng_sentence)\n",
    "  pad_value = tf.constant([[13,0], [0,0]])\n",
    "  embeded_english_padded = tf.pad(embedding_out, pad_value)\n",
    "  embeded_english_padded_out = embeded_english_padded[-13:,:]\n",
    "\n",
    "\n",
    "  embedding_exp = tf.expand_dims(embeded_english_padded_out, 0)\n",
    "  hidden, cell_state = encoder_model(embedding_exp)\n",
    "  translated_german = []\n",
    "  decoder_input = start_token\n",
    "  # keep making predictions until we find the end token\n",
    "  while decoder_input.numpy()[0][0] != end_token_val:\n",
    "    # run through the decoder\n",
    "    decoder_out, hidden, cell_state = decoder_model(decoder_input,\n",
    "                                                    hidden,\n",
    "                                                    cell_state)\n",
    "    # get the predicted index\n",
    "    pred_out_idx = ((np.argmax(decoder_out, axis=-1))[0][0])\n",
    "    if pred_out_idx not in index_to_word:\n",
    "      break\n",
    "    #translate to german word\n",
    "    german_word = index_to_word[pred_out_idx]\n",
    "    #add to list\n",
    "    translated_german.append(german_word)\n",
    "    #update decoder input\n",
    "    decoder_input = tf.constant([[pred_out_idx]])\n",
    "  print(f\"Original German: {german_sentence_list[e_idx]}\")\n",
    "  print(f\"Trainslated German: {translated_german}\")\n",
    "  print(\"=================================\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "ohleJRcJ1icD"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Unk60cEy1icI"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "gpuType": "T4",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}