Upload 54 files

language_models/gru/motivating example models/problem1/model_GRU_ME1_original.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:835c895c1ffe67e6ad103f31397609de12a0b8ce0a1e724c78f7f01804a151c6
+size 529368600

language_models/gru/motivating example models/problem1/model_GRU_ME1_soln1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b15628a07a04a0f825a4809fb2e45fa89a3f242d49b11b9f5b2916eea8944d1
+size 529368608

language_models/gru/motivating example models/problem1/model_GRU_ME1_soln2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca0f245c0351b15bbc4c70dbcaf8d2fb18016ef4af717ea795c994b320e8a1b7
+size 529368600

language_models/gru/motivating example models/problem2/model_GRU_ME2_original.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:419ae2ca52e6460016a7227ad98a2a2779f640d48e48d973934d678b18a42bfb
+size 232568608

language_models/gru/motivating example models/problem2/model_GRU_ME2_soln1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0553c417cdcf67805e5b31e6b70fc8a2bedd8b556e29bbda038457df06c8bf95
+size 529368600

language_models/gru/motivating example models/problem2/model_GRU_ME2_soln2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66e66c458c6044d1a390fbcfd7d9359f6bf8bea6cc649ba8251906818378ab91
+size 232568600

language_models/gru/reuse models/model4_de_fr.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cade5c3c03557d71ec2ce3b1a73c0ad4705b7d47549582d243516fd223e13f31
+size 529368600

language_models/gru/reuse models/model4_de_it.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:377e7812d71a8d177e65f7b8c0adca4811b4433a493ad60a27c118560cd34f7c
+size 529368600

language_models/gru/reuse models/model4_fr_it.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba7110430b5882b123ca948f7662b15590e79557b5062b60f57f443e6b266790
+size 529368608

language_models/gru/rq1 models/model_GRU_1layer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c0393a7878c072feef7df4e54a1bec8661931324509b748d1cbac7890907e38
+size 378266768

language_models/gru/rq1 models/model_GRU_2layer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cc5ac4f8836ad5bb2019f385e6474805e0acf066e52084bddca952097c650f1
+size 428633280

language_models/gru/rq1 models/model_GRU_3layer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a6614568323b11b536b05353909c1d1179700aa0fedb93c497876aa6feb773c0
+size 479002056

language_models/gru/rq1 models/model_GRU_4layer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77b4fa3b332d39bbd9bd1e2b93a9f92f6f1450b3d0e4fd8774396f32a3509711
+size 529368600

language_models/gru/training script/MNMT_GRU_Experiment.ipynb

@@ -0,0 +1,1271 @@
+{
+  "cells": [
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "SIK13ZpDhF_n",
+        "outputId": "04cc436b-3b58-4cfa-bd09-b683b7dd6218"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Mounted at /content/drive\n"
+          ]
+        }
+      ],
+      "source": [
+        "#connect drive\n",
+        "from google.colab import drive\n",
+        "drive.mount('/content/drive', force_remount=True)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "CB3hzllVhKVC"
+      },
+      "outputs": [],
+      "source": [
+        "import csv\n",
+        "import os\n",
+        "import numpy as np\n",
+        "from keras.preprocessing.text import Tokenizer\n",
+        "import pandas as pd\n",
+        "import tensorflow as tf\n",
+        "from keras.utils.np_utils import to_categorical\n",
+        "from keras_preprocessing.sequence import pad_sequences\n",
+        "import tensorflow_datasets as tfds\n",
+        "from nltk.translate.bleu_score import corpus_bleu\n",
+        "import os\n",
+        "from keras.callbacks import ReduceLROnPlateau\n",
+        "from keras.callbacks import ModelCheckpoint\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "OdmG4c70hxQs"
+      },
+      "outputs": [],
+      "source": [
+        "def load_data(path):\n",
+        "    \"\"\"\n",
+        "    Load dataset\n",
+        "    \"\"\"\n",
+        "    input_file = os.path.join(path)\n",
+        "    with open(input_file, \"r\") as f:\n",
+        "        data = f.read()\n",
+        "\n",
+        "    return data.split('\\n')\n",
+        "\n",
+        "def load_glob_embedding(num_words, embed_size=100, word_index=None):\n",
+        "    from numpy import asarray\n",
+        "    from numpy import zeros\n",
+        "\n",
+        "    embeddings_dictionary = dict()\n",
+        "    glove_file = open('/content/drive/MyDrive/NMT/glove.6B.'+str(embed_size)+'d.txt', encoding=\"utf8\")\n",
+        "\n",
+        "    for line in glove_file:\n",
+        "        records = line.split()\n",
+        "        word = records[0]\n",
+        "        vector_dimensions = asarray(records[1:], dtype='float32')\n",
+        "        embeddings_dictionary[word] = vector_dimensions\n",
+        "    glove_file.close()\n",
+        "\n",
+        "    embedding_matrix = zeros((num_words, embed_size))\n",
+        "    for index, word in enumerate(word_index):\n",
+        "        embedding_vector = embeddings_dictionary.get(word)\n",
+        "        if embedding_vector is not None:\n",
+        "            embedding_matrix[index] = embedding_vector\n",
+        "\n",
+        "    return embedding_matrix\n",
+        "\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "oTEQAFkOK8B4"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/en_train.txt')\n",
+        "french_sentences = load_data('/content/drive/MyDrive/NMT/fr_train.txt')\n",
+        "german_sentences = load_data('/content/drive/MyDrive/NMT/de_train.txt')\n",
+        "italian_sentences = load_data('/content/drive/MyDrive/NMT/it_train.txt')"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "jNH6q4fk63Uc"
+      },
+      "source": [
+        "# Run only for the original model, not for inter-reuse experiment"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "b4BL1OgOiVFF"
+      },
+      "outputs": [],
+      "source": [
+        "text_pairs = []\n",
+        "for english,french,german,italian in zip(english_sentences, french_sentences,german_sentences,italian_sentences):\n",
+        "    # english = \"[starten] \" + english + \" [enden]\"\n",
+        "    french = \"[startfr] \" + french + \" [endfr]\"\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((english, french))\n",
+        "    text_pairs.append((english, german))\n",
+        "    text_pairs.append((english, italian))\n",
+        "\n",
+        "    # text_pairs.append((french, english))\n",
+        "    # text_pairs.append((french, german))\n",
+        "    # text_pairs.append((french, italian))\n",
+        "\n",
+        "    # text_pairs.append((german, english))\n",
+        "    # text_pairs.append((german, french))\n",
+        "    # text_pairs.append((german, italian))\n",
+        "\n",
+        "    # text_pairs.append((italian, english))\n",
+        "    # text_pairs.append((italian, french))\n",
+        "    # text_pairs.append((italian, german))\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "fXgeKJYJ2F6r"
+      },
+      "source": [
+        "\n",
+        "\n",
+        "# InterReuse (1st scenario) En to {de, it}"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "3UsmQC_159vK"
+      },
+      "outputs": [],
+      "source": [
+        "text_pairs = []\n",
+        "for english,french,german,italian in zip(english_sentences, french_sentences,german_sentences,italian_sentences):\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((english, german))\n",
+        "    text_pairs.append((english, italian))\n",
+        "    "
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "rKyB-Jtn2GRW"
+      },
+      "source": [
+        "# InterReuse (2nd scenario) En to {de, fr}"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "4NME4f-u6ee_"
+      },
+      "outputs": [],
+      "source": [
+        "text_pairs = []\n",
+        "for english,french,german,italian in zip(english_sentences, french_sentences,german_sentences,italian_sentences):\n",
+        "    french = \"[startfr] \" + french + \" [endfr]\"\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    \n",
+        "    text_pairs.append((english, french))\n",
+        "    text_pairs.append((english, german))\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "eGBLp-242GUM"
+      },
+      "source": [
+        "# InterReuse (3rd scenario) En to {it, fr}"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "kiES-xcO6rEf"
+      },
+      "outputs": [],
+      "source": [
+        "text_pairs = []\n",
+        "for english,french,german,italian in zip(english_sentences, french_sentences,german_sentences,italian_sentences):\n",
+        "    french = \"[startfr] \" + french + \" [endfr]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((english, french))\n",
+        "    text_pairs.append((english, italian))\n",
+        "    "
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "-bP39N4viCAV"
+      },
+      "source": [
+        "# Motivating Example 1 Original Model"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "VcIvoR2LiCoN"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/en_ua_train_original.txt')\n",
+        "french_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/fr_ua_train_original.txt')\n",
+        "german_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/de_ua_train_original.txt')\n",
+        "ukranian_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/ua_ua_train_original.txt')\n",
+        "italian_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/it_ua_train_original.txt')\n",
+        "\n",
+        "text_pairs = []\n",
+        "for english,french,german,italian,ukranian in zip(english_sentences, french_sentences,german_sentences,italian_sentences,ukranian_sentences):\n",
+        "    french = \"[startfr] \" + french + \" [endfr]\"\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "    ukranian = \"[startua] \" + ukranian + \" [endua]\"\n",
+        "\n",
+        "    text_pairs.append((english, french))\n",
+        "    text_pairs.append((english, german))\n",
+        "    text_pairs.append((english, italian))\n",
+        "    text_pairs.append((english, ukranian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "xFmSR42wE1GY"
+      },
+      "source": [
+        "# Motivating Example 1 Solution 1"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "ENunZM3FE23o"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/en_ua_train.txt')\n",
+        "german_sentences = load_data('/content/drive/MyDrive/NMT/de_ua_train.txt')\n",
+        "ukranian_sentences = load_data('/content/drive/MyDrive/NMT/ua_train.txt')\n",
+        "\n",
+        "text_pairs = []\n",
+        "for english, german, ukranian in zip(english_sentences,german_sentences, ukranian_sentences):\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    ukranian = \"[startua] \" + ukranian + \" [endua]\"\n",
+        "\n",
+        "    text_pairs.append((english, german))\n",
+        "    text_pairs.append((english, ukranian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "IUnNMpAVDpZG"
+      },
+      "source": [
+        "# Motivating Example 1 Solution 2"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "T5-vlFmeDrtQ"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/en_ua_train.txt')\n",
+        "ukranian_sentences = load_data('/content/drive/MyDrive/NMT/ua_train.txt')\n",
+        "\n",
+        "text_pairs = []\n",
+        "for english,ukranian in zip(english_sentences, ukranian_sentences):\n",
+        "    ukranian = \"[startua] \" + ukranian + \" [endua]\"\n",
+        "\n",
+        "    text_pairs.append((english, ukranian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "VtPspyGHKpYj"
+      },
+      "source": [
+        "# Motivating Example 2 Original\n",
+        "\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "BkgEgbc6Kteo"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/en_et_train.txt')\n",
+        "german_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/de_et_train.txt')\n",
+        "italian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/it_et_train.txt')\n",
+        "estonian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/et_et_train.txt')\n",
+        "text_pairs = []\n",
+        "for english, german, italian, estonian in zip(english_sentences,german_sentences, italian_sentences,estonian_sentences):\n",
+        "    english = \"[starten] \" + english + \" [enden]\"\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((estonian, english))\n",
+        "    text_pairs.append((estonian, german))\n",
+        "    text_pairs.append((estonian, italian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "bUyLIVUHn0Xu"
+      },
+      "source": [
+        "# Motivating Example 2 Solution 1\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "9-0JKeqWn4YW"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/en_et_train.txt')\n",
+        "italian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/it_et_train.txt')\n",
+        "estonian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/et_et_train.txt')\n",
+        "\n",
+        "text_pairs = []\n",
+        "for english, italian, estonian in zip(english_sentences, italian_sentences,estonian_sentences):\n",
+        "    english = \"[starten] \" + english + \" [enden]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((estonian, english))\n",
+        "    text_pairs.append((estonian, italian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "MI7Ti6Vnkf3v"
+      },
+      "source": [
+        "# Motivating Example 2 Solution 2\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "rxssKJSUkkJP"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/s2/en_et_train_s2.txt')\n",
+        "estonian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/s2/et_et_train_s2.txt')\n",
+        "text_pairs = []\n",
+        "for english, estonian in zip(english_sentences,estonian_sentences):\n",
+        "    english = \"[starten] \" + english + \" [enden]\"\n",
+        "\n",
+        "    text_pairs.append((estonian, english))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "w8044H00hfK-"
+      },
+      "source": [
+        "# Building the model"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "mrG_3rC8h01E",
+        "outputId": "70effc3f-048f-46de-a994-7ddb8d58a249"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "('Ma võtan vihmavarju kaasa.', \"[starten] I'm going to take an umbrella with me. [enden]\")\n"
+          ]
+        }
+      ],
+      "source": [
+        "import random\n",
+        "print(random.choice(text_pairs))"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "i4bKiSxSLKMI"
+      },
+      "outputs": [],
+      "source": [
+        "import random\n",
+        "random.shuffle(text_pairs)\n",
+        "num_val_samples = int(0.15 * len(text_pairs))\n",
+        "num_train_samples = len(text_pairs) - 2 * num_val_samples\n",
+        "train_pairs = text_pairs[:num_train_samples]\n",
+        "val_pairs = text_pairs[num_train_samples:num_train_samples + num_val_samples]\n",
+        "test_pairs = text_pairs[num_train_samples + num_val_samples:]"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "CB-X3EfULMec"
+      },
+      "outputs": [],
+      "source": [
+        "import tensorflow as tf\n",
+        "import string\n",
+        "import re\n",
+        "from tensorflow.keras import layers\n",
+        "\n",
+        "strip_chars = string.punctuation + \"¿\"\n",
+        "strip_chars = strip_chars.replace(\"[\", \"\")\n",
+        "strip_chars = strip_chars.replace(\"]\", \"\")\n",
+        "\n",
+        "def custom_standardization(input_string):\n",
+        "    lowercase = tf.strings.lower(input_string)\n",
+        "    return tf.strings.regex_replace(\n",
+        "        lowercase, f\"[{re.escape(strip_chars)}]\", \"\")\n",
+        "\n",
+        "vocab_size = 2000\n",
+        "# en_vocab_size = 11000\n",
+        "# fr_vocab_size = 18000\n",
+        "sequence_length = 20\n",
+        "\n",
+        "source_vectorization = layers.TextVectorization(\n",
+        "    max_tokens=vocab_size,\n",
+        "    output_mode=\"int\",\n",
+        "    output_sequence_length=sequence_length,\n",
+        ")\n",
+        "target_vectorization = layers.TextVectorization(\n",
+        "    max_tokens=vocab_size,\n",
+        "    output_mode=\"int\",\n",
+        "    output_sequence_length=sequence_length + 1,\n",
+        "    standardize=custom_standardization,\n",
+        ")\n",
+        "train_source_texts = [pair[0] for pair in train_pairs]\n",
+        "train_target_texts = [pair[1] for pair in train_pairs]\n",
+        "source_vectorization.adapt(train_source_texts)\n",
+        "target_vectorization.adapt(train_target_texts)\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "IhrYWMyHLPfr"
+      },
+      "outputs": [],
+      "source": [
+        "batch_size = 64\n",
+        "\n",
+        "def format_dataset(eng, spa):\n",
+        "    eng = source_vectorization(eng)\n",
+        "    spa = target_vectorization(spa)\n",
+        "    return ({\n",
+        "        \"source\": eng,\n",
+        "        \"target\": spa[:, :-1],\n",
+        "    }, spa[:, 1:])\n",
+        "\n",
+        "def make_dataset(pairs):\n",
+        "    eng_texts, spa_texts = zip(*pairs)\n",
+        "    eng_texts = list(eng_texts)\n",
+        "    spa_texts = list(spa_texts)\n",
+        "    dataset = tf.data.Dataset.from_tensor_slices((eng_texts, spa_texts))\n",
+        "    dataset = dataset.batch(batch_size)\n",
+        "    dataset = dataset.map(format_dataset, num_parallel_calls=4)\n",
+        "    return dataset.shuffle(2048).prefetch(16).cache()\n",
+        "\n",
+        "train_ds = make_dataset(train_pairs)\n",
+        "val_ds = make_dataset(val_pairs)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "tFFNByKXLRtV",
+        "outputId": "2e9a59bd-1a4d-4a67-c3f0-6ee124c81df3"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "inputs['source'].shape: (64, 20)\n",
+            "inputs['target'].shape: (64, 20)\n",
+            "targets.shape: (64, 20)\n"
+          ]
+        }
+      ],
+      "source": [
+        "for inputs, targets in train_ds.take(1):\n",
+        "    print(f\"inputs['source'].shape: {inputs['source'].shape}\")\n",
+        "    print(f\"inputs['target'].shape: {inputs['target'].shape}\")\n",
+        "    print(f\"targets.shape: {targets.shape}\")"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "kawQLBPHLTmr"
+      },
+      "outputs": [],
+      "source": [
+        "from tensorflow import keras\n",
+        "from tensorflow.keras import layers\n",
+        "\n",
+        "embed_dim = 200\n",
+        "latent_dim = 1024\n",
+        "\n",
+        "embedding_matrix = load_glob_embedding(vocab_size, 200, target_vectorization.get_vocabulary())\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "HUxTSb5ELV_S",
+        "outputId": "c8467ef6-02b2-4fa5-d52e-91c5d27bb637"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "WARNING:tensorflow:Layer rnn_encoder1 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
+            "WARNING:tensorflow:Layer rnn_encoder2 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
+            "WARNING:tensorflow:Layer rnn_encoder3 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
+            "WARNING:tensorflow:Layer rnn_encoder4 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
+            "WARNING:tensorflow:Layer rnn_decoder1 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
+            "WARNING:tensorflow:Layer rnn_decoder2 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
+            "WARNING:tensorflow:Layer rnn_decoder3 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
+            "WARNING:tensorflow:Layer rnn_decoder4 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n"
+          ]
+        }
+      ],
+      "source": [
+        "source = keras.Input(shape=(None,), dtype=\"int64\", name=\"source\")\n",
+        "\n",
+        "# x = layers.Embedding(vocab_size, embed_dim, mask_zero=True)(source)\n",
+        "x = layers.Embedding(vocab_size, embed_dim, weights=[embedding_matrix], mask_zero=True,\n",
+        "                      name='embed_encoder', trainable=False)(source)\n",
+        "\n",
+        "encoded_source = layers.GRU(latent_dim, return_sequences=True, reset_after=False, activation='tanh', name='rnn_encoder1')(x)\n",
+        "encoded_source = layers.GRU(latent_dim, return_sequences=True, reset_after=False, activation='tanh', name='rnn_encoder2')(encoded_source)\n",
+        "encoded_source = layers.GRU(latent_dim, return_sequences=True, reset_after=False, activation='tanh', name='rnn_encoder3')(encoded_source)\n",
+        "encoded_source, encoder_states = layers.GRU(latent_dim, reset_after=False, return_state=True, activation='tanh', name='rnn_encoder4')(x)\n",
+        "\n",
+        "past_target = keras.Input(shape=(None,), dtype=\"int64\", name=\"target\")\n",
+        "x = layers.Embedding(vocab_size, embed_dim, mask_zero=True, name='embed_decoder')(past_target)\n",
+        "\n",
+        "decoder_gru = layers.GRU(latent_dim, reset_after=False, return_sequences=True, activation='tanh', name='rnn_decoder1')\n",
+        "x = decoder_gru(x, initial_state=encoder_states)\n",
+        "x = layers.GRU(latent_dim, reset_after=False, return_sequences=True, activation='tanh', name='rnn_decoder2')(x)\n",
+        "x = layers.GRU(latent_dim, reset_after=False, return_sequences=True, activation='tanh', name='rnn_decoder3')(x)\n",
+        "x = layers.GRU(latent_dim, reset_after=False, return_sequences=True, activation='tanh', name='rnn_decoder4')(x)\n",
+        "\n",
+        "x = layers.Dropout(0.5)(x)\n",
+        "\n",
+        "target_next_step = layers.TimeDistributed(layers.Dense(vocab_size, activation=\"softmax\", name='output'))(x)\n",
+        "\n",
+        "seq2seq_rnn = keras.Model([source, past_target], target_next_step)\n",
+        "\n",
+        "seq2seq_rnn.compile(\n",
+        "    optimizer=\"rmsprop\",\n",
+        "    loss=\"sparse_categorical_crossentropy\",\n",
+        "    metrics=[\"accuracy\"])"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "twWGTTxHLX0-",
+        "outputId": "a994a363-7e6d-4353-a2e8-32a3421ac140"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Epoch 1/3\n",
+            "35/35 [==============================] - 17s 268ms/step - loss: 2.1299 - accuracy: 0.1399 - val_loss: 1.9851 - val_accuracy: 0.1586\n",
+            "Epoch 2/3\n",
+            "35/35 [==============================] - 8s 241ms/step - loss: 1.9150 - accuracy: 0.1603 - val_loss: 1.9674 - val_accuracy: 0.1561\n",
+            "Epoch 3/3\n",
+            "35/35 [==============================] - 8s 241ms/step - loss: 1.8826 - accuracy: 0.1795 - val_loss: 1.8760 - val_accuracy: 0.1873\n"
+          ]
+        }
+      ],
+      "source": [
+        "# checkpoint = ModelCheckpoint(filepath='model_1LSTM_original_chkpt.h5',\n",
+        "#                               monitor='val_loss', verbose=1, save_best_only=True,\n",
+        "#                               mode='min')\n",
+        "\n",
+        "seq2seq_rnn.fit(train_ds, epochs=3, validation_data=val_ds)\n",
+        "seq2seq_rnn.save('model_GRU_ME2_soln2.h5')"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "background_save": true
+        },
+        "id": "0SkImFO6os8J",
+        "outputId": "c2d98cb7-90f6-484e-d3c3-7b2f094abdf7"
+      },
+      "outputs": [
+        {
+          "ename": "KeyboardInterrupt",
+          "evalue": "ignored",
+          "output_type": "error",
+          "traceback": [
+            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
+            "\u001b[0;32m<ipython-input-31-9a163dae5324>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     33\u001b[0m     \u001b[0;31m# print(\"predicted: \", decode_sequence(test_eng_texts[i],acts[0], acts[-1]))\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     34\u001b[0m     \u001b[0mactual\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0macts\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 35\u001b[0;31m     \u001b[0mpredicted\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdecode_sequence\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtest_eng_texts\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0macts\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0macts\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     36\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     37\u001b[0m \u001b[0;31m#print(\"actual: \", actual, '\\n', \"predicted:\", predicted)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m<ipython-input-31-9a163dae5324>\u001b[0m in \u001b[0;36mdecode_sequence\u001b[0;34m(input_sentence, decoded_sentence, end_tag)\u001b[0m\n\u001b[1;32m     11\u001b[0m         \u001b[0mtokenized_target_sentence\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtarget_vectorization\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mdecoded_sentence\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m         next_token_predictions = seq2seq_rnn.predict(\n\u001b[0;32m---> 13\u001b[0;31m             [tokenized_input_sentence, tokenized_target_sentence])\n\u001b[0m\u001b[1;32m     14\u001b[0m         \u001b[0msampled_token_index\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0margmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnext_token_predictions\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mi\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m         \u001b[0msampled_token\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mspa_index_lookup\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0msampled_token_index\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     62\u001b[0m     \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     63\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 64\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     65\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint: disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, x, batch_size, verbose, steps, callbacks, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1976\u001b[0m       \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_predict_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1977\u001b[0m       \u001b[0mbatch_outputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1978\u001b[0;31m       \u001b[0;32mfor\u001b[0m \u001b[0m_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0miterator\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0menumerate_epochs\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# Single epoch.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1979\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcatch_stop_iteration\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1980\u001b[0m           \u001b[0;32mfor\u001b[0m \u001b[0mstep\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msteps\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/engine/data_adapter.py\u001b[0m in \u001b[0;36menumerate_epochs\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1189\u001b[0m     \u001b[0;34m\"\"\"Yields `(epoch, tf.data.Iterator)`.\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1190\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_truncate_execution_to_epoch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1191\u001b[0;31m       \u001b[0mdata_iterator\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0miter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dataset\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1192\u001b[0m       \u001b[0;32mfor\u001b[0m \u001b[0mepoch\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_initial_epoch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_epochs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1193\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_insufficient_data\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# Set by `catch_stop_iteration`.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/data/ops/dataset_ops.py\u001b[0m in \u001b[0;36m__iter__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    484\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mcontext\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexecuting_eagerly\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minside_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    485\u001b[0m       \u001b[0;32mwith\u001b[0m \u001b[0mops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcolocate_with\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_variant_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 486\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0miterator_ops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOwnedIterator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    487\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m       raise RuntimeError(\"`tf.data.Dataset` only supports Python-style \"\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/data/ops/iterator_ops.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, dataset, components, element_spec)\u001b[0m\n\u001b[1;32m    753\u001b[0m             \u001b[0;34m\"When `dataset` is provided, `element_spec` and `components` must \"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    754\u001b[0m             \"not be specified.\")\n\u001b[0;32m--> 755\u001b[0;31m       \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_create_iterator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    756\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    757\u001b[0m     \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_get_next_call_count\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/data/ops/iterator_ops.py\u001b[0m in \u001b[0;36m_create_iterator\u001b[0;34m(self, dataset)\u001b[0m\n\u001b[1;32m    785\u001b[0m                 \u001b[0moutput_types\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_flat_output_types\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    786\u001b[0m                 output_shapes=self._flat_output_shapes))\n\u001b[0;32m--> 787\u001b[0;31m         \u001b[0mgen_dataset_ops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmake_iterator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mds_variant\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_iterator_resource\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    788\u001b[0m         \u001b[0;31m# Delete the resource when this object is deleted\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    789\u001b[0m         self._resource_deleter = IteratorResourceDeleter(\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/gen_dataset_ops.py\u001b[0m in \u001b[0;36mmake_iterator\u001b[0;34m(dataset, iterator, name)\u001b[0m\n\u001b[1;32m   3314\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3315\u001b[0m       _result = pywrap_tfe.TFE_Py_FastPathExecute(\n\u001b[0;32m-> 3316\u001b[0;31m         _ctx, \"MakeIterator\", name, dataset, iterator)\n\u001b[0m\u001b[1;32m   3317\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0m_result\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3318\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0m_core\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
+          ]
+        }
+      ],
+      "source": [
+        "import numpy as np\n",
+        "spa_vocab = target_vectorization.get_vocabulary()\n",
+        "spa_index_lookup = dict(zip(range(len(spa_vocab)), spa_vocab))\n",
+        "max_decoded_sentence_length = 20\n",
+        "\n",
+        "def decode_sequence(input_sentence, decoded_sentence, end_tag):\n",
+        "    tokenized_input_sentence = source_vectorization([input_sentence])\n",
+        "    #decoded_sentence = \"[start]\"\n",
+        "\n",
+        "    for i in range(max_decoded_sentence_length):\n",
+        "        tokenized_target_sentence = target_vectorization([decoded_sentence])\n",
+        "        next_token_predictions = seq2seq_rnn.predict(\n",
+        "            [tokenized_input_sentence, tokenized_target_sentence])\n",
+        "        sampled_token_index = np.argmax(next_token_predictions[0, i, :])\n",
+        "        sampled_token = spa_index_lookup[sampled_token_index]\n",
+        "        decoded_sentence += \" \" + sampled_token\n",
+        "        if sampled_token == end_tag:\n",
+        "            break\n",
+        "    return decoded_sentence\n",
+        "\n",
+        "bleu_dic = {}\n",
+        "test_eng_texts = [pair[0] for pair in test_pairs]\n",
+        "test_fr_texts = [pair[1] for pair in test_pairs]\n",
+        "actual, predicted = [], []\n",
+        "for i in range(len(test_pairs)):\n",
+        "    # input_sentence = \n",
+        "    # target_sentence = random.choice(test_fr_texts)\n",
+        "    acts=test_fr_texts[i].split()\n",
+        "    # print(\"-\")\n",
+        "    # print(acts[0], acts[-1])\n",
+        "    # print(\"source: \", test_eng_texts[i])\n",
+        "    # print(\"actual target: \", test_fr_texts[i])\n",
+        "    # print(\"predicted: \", decode_sequence(test_eng_texts[i],acts[0], acts[-1]))\n",
+        "    actual.append([acts])\n",
+        "    predicted.append(decode_sequence(test_eng_texts[i],acts[0], acts[-1]).split())\n",
+        "\n",
+        "#print(\"actual: \", actual, '\\n', \"predicted:\", predicted)\n",
+        "bleu_dic['1-grams'] = corpus_bleu(actual, predicted, weights=(1.0, 0, 0, 0))\n",
+        "bleu_dic['1-2-grams'] = corpus_bleu(actual, predicted, weights=(0.5, 0.5, 0, 0))\n",
+        "bleu_dic['1-3-grams'] = corpus_bleu(actual, predicted, weights=(0.3, 0.3, 0.3, 0))\n",
+        "bleu_dic['1-4-grams'] = corpus_bleu(actual, predicted, weights=(0.25, 0.25, 0.25, 0.25))\n",
+        "print(bleu_dic)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 281
+        },
+        "id": "D_Qf1ZnRNual",
+        "outputId": "ef5196ee-1377-4041-cc41-1c0fdd71a911"
+      },
+      "outputs": [
+        {
+          "data": {
+            "image/png": "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\n",
+            "text/plain": [
+              "<Figure size 432x288 with 1 Axes>"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        }
+      ],
+      "source": [
+        "from nltk.translate.bleu_score import corpus_bleu\n",
+        "import matplotlib.pyplot as plt\n",
+        "\n",
+        "plt.bar(x = bleu_dic.keys(), height = bleu_dic.values())\n",
+        "plt.title(\"BLEU Score with the testing set\")\n",
+        "plt.ylim((0,1))\n",
+        "plt.show()"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "ThXu38YNo6LQ",
+        "outputId": "540b3456-f792-4b1e-d605-c7d320bf5729"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Mounted at /content/drive/\n"
+          ]
+        }
+      ],
+      "source": [
+        "from google.colab import drive\n",
+        "\n",
+        "drive.mount(\"/content/drive/\")\n",
+        "\n",
+        "#files.download(\"/content/model4_final_reuse_2.h5\")\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 425
+        },
+        "id": "6WX179UgqihS",
+        "outputId": "7ba02150-d3d8-4bbc-8186-015ad7c1e92e"
+      },
+      "outputs": [
+        {
+          "ename": "ValueError",
+          "evalue": "ignored",
+          "output_type": "error",
+          "traceback": [
+            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+            "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
+            "\u001b[0;32m<ipython-input-15-38f909bf98ba>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     14\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDropout\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0.5\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mSimpleRNN\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlatent_dim\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactivation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'relu'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'rnn_decoder2'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 16\u001b[0;31m \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mSimpleRNN\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlatent_dim\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactivation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'relu'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'rnn_decoder3'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     17\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m \u001b[0mtarget_next_step\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTimeDistributed\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDense\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvocab_size\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactivation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m\"softmax\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'output'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/layers/recurrent.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs, initial_state, constants, **kwargs)\u001b[0m\n\u001b[1;32m    677\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    678\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0minitial_state\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mconstants\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 679\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0msuper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mRNN\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__call__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    680\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    681\u001b[0m     \u001b[0;31m# If any of `initial_state` or `constants` are specified and are Keras\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     65\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint: disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 67\u001b[0;31m       \u001b[0;32mraise\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfiltered_tb\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     68\u001b[0m     \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     69\u001b[0m       \u001b[0;32mdel\u001b[0m \u001b[0mfiltered_tb\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/engine/input_spec.py\u001b[0m in \u001b[0;36massert_input_compatibility\u001b[0;34m(input_spec, inputs, layer_name)\u001b[0m\n\u001b[1;32m    212\u001b[0m       \u001b[0mndim\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mshape\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrank\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    213\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mndim\u001b[0m \u001b[0;34m!=\u001b[0m \u001b[0mspec\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndim\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 214\u001b[0;31m         raise ValueError(f'Input {input_index} of layer \"{layer_name}\" '\n\u001b[0m\u001b[1;32m    215\u001b[0m                          \u001b[0;34m'is incompatible with the layer: '\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    216\u001b[0m                          \u001b[0;34mf'expected ndim={spec.ndim}, found ndim={ndim}. '\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;31mValueError\u001b[0m: Input 0 of layer \"rnn_decoder3\" is incompatible with the layer: expected ndim=3, found ndim=2. Full shape received: (None, 1024)"
+          ]
+        }
+      ],
+      "source": [
+        "source = keras.Input(shape=(None,), dtype=\"int64\", name=\"source\")\n",
+        "\n",
+        "# x = layers.Embedding(vocab_size, embed_dim, mask_zero=True)(source)\n",
+        "x = layers.Embedding(vocab_size, embed_dim, weights=[embedding_matrix], mask_zero=True,\n",
+        "                      name='embed_encoder', trainable=False)(source)\n",
+        "\n",
+        "encoded_source = layers.SimpleRNN(latent_dim, activation='relu', name='rnn_encoder4')(x)\n",
+        "\n",
+        "past_target = keras.Input(shape=(None,), dtype=\"int64\", name=\"target\")\n",
+        "x = layers.Embedding(vocab_size, embed_dim, mask_zero=True, name='embed_decoder')(past_target)\n",
+        "\n",
+        "decoder_gru = layers.SimpleRNN(latent_dim, return_sequences=True, activation='relu', name='rnn_decoder1')\n",
+        "x = decoder_gru(x, initial_state=encoded_source)\n",
+        "x = layers.Dropout(0.5)(x)\n",
+        "x = layers.SimpleRNN(latent_dim, activation='relu', name='rnn_decoder2')(x)\n",
+        "x = layers.SimpleRNN(latent_dim, return_sequences=True, activation='relu', name='rnn_decoder3')(x)\n",
+        "\n",
+        "target_next_step = layers.TimeDistributed(layers.Dense(vocab_size, activation=\"softmax\", name='output'))(x)\n",
+        "\n",
+        "seq2seq_rnn = keras.Model([source, past_target], target_next_step)\n",
+        "\n",
+        "seq2seq_rnn.compile(\n",
+        "    optimizer=\"rmsprop\",\n",
+        "    loss=\"sparse_categorical_crossentropy\",\n",
+        "    metrics=[\"accuracy\"])"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 222
+        },
+        "id": "NvlZdKQ2qk6a",
+        "outputId": "c6dd1258-df31-4b5a-e79d-91e2d83f21af"
+      },
+      "outputs": [
+        {
+          "ename": "NameError",
+          "evalue": "ignored",
+          "output_type": "error",
+          "traceback": [
+            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
+            "\u001b[0;32m<ipython-input-1-26deb6ad3f93>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;31m#                               mode='min')\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m \u001b[0mseq2seq_rnn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_ds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalidation_data\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mval_ds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallbacks\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mcheckpoint\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m \u001b[0;31m# seq2seq_rnn.save('model4_final_ME_2.h5')\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;31mNameError\u001b[0m: name 'seq2seq_rnn' is not defined"
+          ]
+        }
+      ],
+      "source": [
+        "vocab_size = 30000\n",
+        "sequence_length = 20\n",
+        "\n",
+        "# checkpoint = ModelCheckpoint(filepath='model4_ME_2.h5',\n",
+        "#                               monitor='val_loss', verbose=1, save_best_only=True,\n",
+        "#                               mode='min')\n",
+        "\n",
+        "seq2seq_rnn.fit(train_ds, epochs=3, validation_data=val_ds, callbacks=[checkpoint])\n",
+        "# seq2seq_rnn.save('model4_final_ME_2.h5')\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "BRc0Z_ELHpdD"
+      },
+      "source": [
+        "### **The Codes after this are from prior experiment.**"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "stJYZc1eh2Kw"
+      },
+      "outputs": [],
+      "source": [
+        "def tokenize(x):\n",
+        "    \"\"\"\n",
+        "    Tokenize x\n",
+        "    :param x: List of sentences/strings to be tokenized\n",
+        "    :return: Tuple of (tokenized x data, tokenizer used to tokenize x)\n",
+        "    \"\"\"\n",
+        "    # TODO: Implement\n",
+        "    tokenizer = Tokenizer()\n",
+        "    tokenizer.fit_on_texts(x)\n",
+        "    return tokenizer.texts_to_sequences(x), tokenizer"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "CVZgVHkzh3tr"
+      },
+      "outputs": [],
+      "source": [
+        "def pad(x, length=None):\n",
+        "    \"\"\"\n",
+        "    Pad x\n",
+        "    :param x: List of sequences.\n",
+        "    :param length: Length to pad the sequence to.  If None, use length of longest sequence in x.\n",
+        "    :return: Padded numpy array of sequences\n",
+        "    \"\"\"\n",
+        "#     # TODO: Implement\n",
+        "#     if length is None:\n",
+        "#         length=max([len(sentence) for sentence in x])\n",
+        "#         print(length)\n",
+        "    \n",
+        "    return pad_sequences(x, maxlen=20, padding ='post')"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "msE26JGmh5TA"
+      },
+      "outputs": [],
+      "source": [
+        "import collections\n",
+        "\n",
+        "english_words_counter = collections.Counter([word for sentence in english_sentences for word in sentence.split()])\n",
+        "french_words_counter = collections.Counter([word for sentence in french_sentences for word in sentence.split()])\n",
+        "\n",
+        "print('{} English words.'.format(len([word for sentence in english_sentences for word in sentence.split()])))\n",
+        "print('{} unique English words.'.format(len(english_words_counter)))\n",
+        "print('10 Most common words in the English dataset:')\n",
+        "print('\"' + '\" \"'.join(list(zip(*english_words_counter.most_common(10)))[0]) + '\"')\n",
+        "print()\n",
+        "print('{} French words.'.format(len([word for sentence in french_sentences for word in sentence.split()])))\n",
+        "print('{} unique French words.'.format(len(french_words_counter)))\n",
+        "print('10 Most common words in the French dataset:')\n",
+        "print('\"' + '\" \"'.join(list(zip(*french_words_counter.most_common(10)))[0]) + '\"')"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "GHOxz_1Fh7Ha"
+      },
+      "outputs": [],
+      "source": [
+        "for sample_i in range(5):\n",
+        "    print('English sample {}:  {}'.format(sample_i + 1, english_sentences[sample_i+10000]))\n",
+        "    print('French sample {}:  {}\\n'.format(sample_i + 1, french_sentences[sample_i+10000]))\n",
+        "    print('German sample {}:  {}\\n'.format(sample_i + 1, german_sentences[sample_i+10000]))\n",
+        "    print('Italian sample {}:  {}\\n'.format(sample_i + 1, italian_sentences[sample_i+10000]))\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "ogGPGCf7h9Gw"
+      },
+      "outputs": [],
+      "source": [
+        "def preprocess(x, y1, y2, y3):\n",
+        "    \"\"\"\n",
+        "    Preprocess x and y\n",
+        "    :param x: Feature List of sentences\n",
+        "    :param y: Label List of sentences\n",
+        "    :return: Tuple of (Preprocessed x, Preprocessed y, x tokenizer, y tokenizer)\n",
+        "    \"\"\"\n",
+        "    preprocess_en, en_tk = tokenize(x)\n",
+        "    preprocess_fr, fr_tk = tokenize(y1)\n",
+        "    preprocess_de, de_tk = tokenize(y2)\n",
+        "    preprocess_it, it_tk = tokenize(y3)\n",
+        "    \n",
+        "    preprocess_en = pad(preprocess_en)\n",
+        "    preprocess_fr = pad(preprocess_fr)\n",
+        "    preprocess_de = pad(preprocess_de)\n",
+        "    preprocess_it = pad(preprocess_it)\n",
+        "\n",
+        "    \n",
+        "    # Keras's sparse_categorical_crossentropy function requires the labels to be in 3 dimensions\n",
+        "    preprocess_fr = preprocess_fr.reshape(*preprocess_fr.shape, 1)\n",
+        "    preprocess_de = preprocess_de.reshape(*preprocess_de.shape, 1)\n",
+        "    preprocess_it = preprocess_it.reshape(*preprocess_it.shape, 1)\n",
+        "\n",
+        "    return preprocess_en,preprocess_fr,preprocess_de,preprocess_it, en_tk, fr_tk, de_tk, it_tk\n",
+        "\n",
+        "inputTimestep = 30\n",
+        "outputTimestep = 30\n",
+        "\n",
+        "preproc_english_sentences,preproc_french_sentences, preproc_german_sentences, preproc_italian_sentences, en_tokenizer,fr_tokenizer, de_tokenizer, it_tokenizer =\\\n",
+        "    preprocess(english_sentences, french_sentences,german_sentences,italian_sentences )\n",
+        "\n",
+        "\n",
+        "max_english_sequence_length = preproc_english_sentences.shape[1]\n",
+        "max_french_sequence_length = preproc_french_sentences.shape[1]\n",
+        "max_german_sequence_length = preproc_german_sentences.shape[1]\n",
+        "max_italian_sequence_length = preproc_italian_sentences.shape[1]\n",
+        "\n",
+        "english_vocab_size = len(en_tokenizer.word_index)\n",
+        "french_vocab_size = len(fr_tokenizer.word_index)\n",
+        "german_vocab_size = len(de_tokenizer.word_index)\n",
+        "italian_vocab_size = len(it_tokenizer.word_index)\n",
+        "\n",
+        "print('Data Preprocessed')\n",
+        "\n",
+        "print(\"Max English sentence length:\", max_english_sequence_length)\n",
+        "print(\"Max French sentence length:\", max_french_sequence_length)\n",
+        "print(\"Max German sentence length:\", max_german_sequence_length)\n",
+        "print(\"Max Italian sentence length:\", max_italian_sequence_length)\n",
+        "\n",
+        "print(\"English vocabulary size:\", english_vocab_size)\n",
+        "print(\"French vocabulary size:\", french_vocab_size)\n",
+        "print(\"German vocabulary size:\", german_vocab_size)\n",
+        "print(\"Italian vocabulary size:\", italian_vocab_size)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "zdLpnbLHiEk3"
+      },
+      "outputs": [],
+      "source": [
+        "from keras.layers import GRU, Input, Dense, TimeDistributed, Activation, RepeatVector, Bidirectional, Dropout, LSTM\n",
+        "from keras.losses import sparse_categorical_crossentropy\n",
+        "from keras.models import Sequential\n",
+        "from keras.layers import Dense, Activation, TimeDistributed, RepeatVector, Flatten, Conv2D, Embedding\n",
+        "from keras.layers.recurrent import SimpleRNN, LSTM\n",
+        "from keras.utils import np_utils\n",
+        "from tensorflow.keras.models import Model\n",
+        "import keras\n",
+        "\n",
+        "\n",
+        "def many_many_tangled(input_shape, fr_output_sequence_length, english_vocab_size, french_vocab_size):\n",
+        "\n",
+        "\n",
+        "    # Hyperparameters\n",
+        "    opt = tf.keras.optimizers.Adam(learning_rate=1e-3)\n",
+        "    \n",
+        "    # Build the layers    \n",
+        "    model = Sequential()\n",
+        "    # Embedding\n",
+        "    model.add(Embedding(english_vocab_size, 256, input_length=input_shape[1],\n",
+        "                         input_shape=input_shape[1:]))\n",
+        "    # Encoder\n",
+        "    model.add(SimpleRNN(256))\n",
+        "    model.add(RepeatVector(fr_output_sequence_length))\n",
+        "    # Decoder\n",
+        "    model.add(SimpleRNN(256, return_sequences=True))\n",
+        "    model.add(TimeDistributed(Dense(512, activation='relu')))\n",
+        "    model.add(Dropout(0.5))\n",
+        "    model.add(TimeDistributed(Dense((french_vocab_size), activation='softmax')))\n",
+        "    model.compile(loss=sparse_categorical_crossentropy,\n",
+        "                  optimizer=opt,\n",
+        "                  metrics=['accuracy'])\n",
+        "    \n",
+        "    print(model.summary())\n",
+        "\n",
+        "    return model\n",
+        "\n",
+        "def many_many_functional(input_shape, output_sequence_length, english_vocab_size, french_vocab_size, german_vocab_size,italian_vocab_size):\n",
+        "    \n",
+        "    #input\n",
+        "    eng_input = Input(shape=(None,), dtype=\"int64\", name=\"english\")\n",
+        "\n",
+        "    #embedding\n",
+        "    embedding_layer = Embedding(english_vocab_size, 256)(eng_input)\n",
+        "\n",
+        "    rnn_layer_1 = SimpleRNN(256)(embedding_layer)\n",
+        "\n",
+        "    fr_input = Input(shape=(None,), dtype=\"int64\", name=\"spanish\")\n",
+        "\n",
+        "    embedding_layer = Embedding(french_vocab_size, 256)(fr_input)\n",
+        "\n",
+        "    #Encoder for two langauges\n",
+        "    rnn_layer_1 = SimpleRNN(256)(embedding_layer)\n",
+        "    repeat_vector = RepeatVector(output_sequence_length)(rnn_layer_1)\n",
+        "\n",
+        "    #Common decoder for all languages\n",
+        "    rnn_layer2 = SimpleRNN(256, return_sequences=True)(repeat_vector)\n",
+        "    time_distributed_1 = Dense(1024, activation='relu')(rnn_layer2)\n",
+        "    dropout_1 = Dropout(0.5)(time_distributed_1)\n",
+        "    \n",
+        "    output_fr = Dense(french_vocab_size, activation='softmax')(dropout_1)\n",
+        "    #output_de = TimeDistributed(Dense(german_vocab_size, activation='softmax'))(dropout_1)\n",
+        "    #output_it = TimeDistributed(Dense(italian_vocab_size, activation='softmax'))(dropout_1)\n",
+        "    \n",
+        "    #Create model\n",
+        "    #model = Model(inputs=eng_input, outputs=[output_fr,output_de,output_it])\n",
+        "    model = Model(inputs=[eng_input, fr_input], outputs=output_fr)\n",
+        "\n",
+        "\n",
+        "    model.compile(loss=sparse_categorical_crossentropy, optimizer='adam',metrics=['accuracy'])\n",
+        "\n",
+        "    print(model.summary())\n",
+        "    \n",
+        "    return model\n",
+        "\n",
+        "def embed_model(output_sequence_length, english_vocab_size, french_vocab_size):\n",
+        "\n",
+        "    # Hyperparameters\n",
+        "    opt = tf.keras.optimizers.Adam(learning_rate=1e-3)\n",
+        "    # Build the layers    \n",
+        "    model = Sequential()\n",
+        "    # Embedding\n",
+        "    model.add(Embedding(english_vocab_size, 256))\n",
+        "    # Encoder\n",
+        "    model.add(SimpleRNN(256))\n",
+        "    model.add(RepeatVector(output_sequence_length))\n",
+        "    # Decoder\n",
+        "    model.add(SimpleRNN(256, return_sequences=True))\n",
+        "    model.add(TimeDistributed(Dense(516, activation='relu')))\n",
+        "    model.add(Dropout(0.5))\n",
+        "    model.add(TimeDistributed(Dense(516, activation='relu')))\n",
+        "    model.add(Dropout(0.5))\n",
+        "    model.add(TimeDistributed(Dense(french_vocab_size, activation='softmax')))\n",
+        "    model.compile(loss=sparse_categorical_crossentropy,\n",
+        "                  optimizer=opt,\n",
+        "                  metrics=['accuracy'])\n",
+        "    return model\n",
+        "\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "VdDy5uL1t8Sy"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "CeQ4Z6j1spTP"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "Qr25F675iIyX"
+      },
+      "outputs": [],
+      "source": [
+        "# tmp_x = pad(preproc_english_sentences, preproc_french_sentences.shape[1])\n",
+        "# tmp_x = tmp_x.reshape((-1, preproc_french_sentences.shape[-2]))\n",
+        "\n",
+        "reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2,\n",
+        "                              patience=5, min_lr=0.001)\n",
+        "\n",
+        "callback = tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=5)\n",
+        "# TODO: Train the neural network\n",
+        "\n",
+        "many_many = many_many_functional(\n",
+        "    None,\n",
+        "    20,\n",
+        "    en_vocab_size+1,\n",
+        "    fr_vocab_size+1,\n",
+        "    None, None)\n",
+        "\n",
+        "many_many.summary()\n",
+        "\n",
+        "many_many.fit(train_ds, validation_data=val_ds, batch_size=64, epochs=30, callbacks=[callback, reduce_lr])"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "Zt2YCfXosq05"
+      },
+      "outputs": [],
+      "source": [
+        "def logits_to_text(logits, tokenizer):\n",
+        "    \"\"\"\n",
+        "    Turn logits from a neural network into text using the tokenizer\n",
+        "    :param logits: Logits from a neural network\n",
+        "    :param tokenizer: Keras Tokenizer fit on the labels\n",
+        "    :return: String that represents the text of the logits\n",
+        "    \"\"\"\n",
+        "    index_to_words = {id: word for word, id in tokenizer.word_index.items()}\n",
+        "    index_to_words[0] = '<PAD>'\n",
+        "\n",
+        "    return ' '.join([index_to_words[prediction] for prediction in np.argmax(logits, 1)])\n",
+        "\n",
+        "print('`logits_to_text` function loaded.')"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "AS88MAqtstIC"
+      },
+      "outputs": [],
+      "source": [
+        "# Print prediction(s)\n",
+        "print(\"Prediction:\")\n",
+        "\n",
+        "print(logits_to_text(many_many.predict(tmp_x[6:7])[0], fr_tokenizer))\n",
+        "\n",
+        "print(\"\\nCorrect Translation French:\")\n",
+        "print(french_sentences[6:7])\n",
+        "\n",
+        "print(\"\\nOriginal text:\")\n",
+        "print()"
+      ]
+    }
+  ],
+  "metadata": {
+    "accelerator": "GPU",
+    "colab": {
+      "machine_shape": "hm",
+      "provenance": []
+    },
+    "gpuClass": "standard",
+    "kernelspec": {
+      "display_name": "Python 3",
+      "name": "python3"
+    },
+    "language_info": {
+      "name": "python"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}

language_models/lstm/motivating example models/problem1/original_problem1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85158ef279de626273555fd37d4035e9125ba3e486205e353c69378aa0a88574
+size 801261536

language_models/lstm/motivating example models/problem1/solution1_problem1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6e0016869178720578a6fb4c3964426ca121c5919d0ba9e820d57f58aa37bca
+size 801261536

language_models/lstm/motivating example models/problem1/solution2_problem1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccd208ba23a1f099fe940d0a9e94141acf41b5017c85510a66df5d005932b970
+size 801261536

language_models/lstm/motivating example models/problem2/original_problem2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d9104b7d00bdb138f183dccefc022243ad0475c4d28680a619f44b31b333356
+size 801261536

language_models/lstm/motivating example models/problem2/solution1_problem2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3a1a3265df7eace0ca51e249a28526b3497e407f91e8a122ba38c48c4516c8a
+size 504461536

language_models/lstm/motivating example models/problem2/solution2_problem2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31d7d5e98dddccb4ce2c57388e2aff13981dc63afe94b6757d797761d69f74c9
+size 801261536

language_models/lstm/reuse models/model4_de_fr.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed88701982c60ffc1b66c7fafd749e88a4abbd3cf887267ec61fecd9344da293
+size 801261536

language_models/lstm/reuse models/model4_de_it.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1fff9de19afa7e8ebc88bcc1815084c6c68180c92027942ce2d53ac5856552a
+size 801261560

language_models/lstm/reuse models/model4_fr_it.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09992211a9e0a2e3a92f6a901c732beff9606472b43cc6f9dcb8deafcdbaa6ee
+size 801261536

language_models/lstm/rq1 models/model_LSTM_1layer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e37bf0e31b2a0de88b361c8863a5a7fd636c5e1971c0f9a84519fe407dd090d
+size 398337168

language_models/lstm/rq1 models/model_LSTM_2layer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee8fe0fe89df20dac87f6d9c42277ed86cc2805832147f0f7b82de8ba2d84f2e
+size 532645160

language_models/lstm/rq1 models/model_LSTM_3layer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c73a7407b3e824f07571ccbb94ec1f47ef3923c55312b31158a873c308cf959
+size 666952624

language_models/lstm/rq1 models/model_LSTM_4layer.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b67e5a14e3a14cfcde75627f69d54a1efec08a848f4a799af0783ac56265e02f
+size 801261560

language_models/lstm/training script/(LSTM)_NMT_Experiment.ipynb

@@ -0,0 +1,1317 @@
+{
+  "cells": [
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "SIK13ZpDhF_n",
+        "outputId": "3349e413-0325-4873-8022-5cf975e03643"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Mounted at /content/drive\n"
+          ]
+        }
+      ],
+      "source": [
+        "#connect drive\n",
+        "from google.colab import drive\n",
+        "drive.mount('/content/drive', force_remount=True)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "CB3hzllVhKVC"
+      },
+      "outputs": [],
+      "source": [
+        "import csv\n",
+        "import os\n",
+        "import numpy as np\n",
+        "from keras.preprocessing.text import Tokenizer\n",
+        "import pandas as pd\n",
+        "import tensorflow as tf\n",
+        "from keras.utils.np_utils import to_categorical\n",
+        "from keras_preprocessing.sequence import pad_sequences\n",
+        "import tensorflow_datasets as tfds\n",
+        "from nltk.translate.bleu_score import corpus_bleu\n",
+        "import os\n",
+        "from keras.callbacks import ReduceLROnPlateau\n",
+        "from keras.callbacks import ModelCheckpoint\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "OdmG4c70hxQs"
+      },
+      "outputs": [],
+      "source": [
+        "def load_data(path):\n",
+        "    \"\"\"\n",
+        "    Load dataset\n",
+        "    \"\"\"\n",
+        "    input_file = os.path.join(path)\n",
+        "    with open(input_file, \"r\") as f:\n",
+        "        data = f.read()\n",
+        "\n",
+        "    return data.split('\\n')\n",
+        "\n",
+        "def load_glob_embedding(num_words, embed_size=100, word_index=None):\n",
+        "    from numpy import asarray\n",
+        "    from numpy import zeros\n",
+        "\n",
+        "    embeddings_dictionary = dict()\n",
+        "    glove_file = open('/content/drive/MyDrive/NMT/glove.6B.'+str(embed_size)+'d.txt', encoding=\"utf8\")\n",
+        "\n",
+        "    for line in glove_file:\n",
+        "        records = line.split()\n",
+        "        word = records[0]\n",
+        "        vector_dimensions = asarray(records[1:], dtype='float32')\n",
+        "        embeddings_dictionary[word] = vector_dimensions\n",
+        "    glove_file.close()\n",
+        "\n",
+        "    embedding_matrix = zeros((num_words, embed_size))\n",
+        "    for index, word in enumerate(word_index):\n",
+        "        embedding_vector = embeddings_dictionary.get(word)\n",
+        "        if embedding_vector is not None:\n",
+        "            embedding_matrix[index] = embedding_vector\n",
+        "\n",
+        "    return embedding_matrix\n",
+        "\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "oTEQAFkOK8B4"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/en_train.txt')\n",
+        "french_sentences = load_data('/content/drive/MyDrive/NMT/fr_train.txt')\n",
+        "german_sentences = load_data('/content/drive/MyDrive/NMT/de_train.txt')\n",
+        "italian_sentences = load_data('/content/drive/MyDrive/NMT/it_train.txt')"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "jNH6q4fk63Uc"
+      },
+      "source": [
+        "# Run only for the original model, not for inter-reuse experiment"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "b4BL1OgOiVFF"
+      },
+      "outputs": [],
+      "source": [
+        "text_pairs = []\n",
+        "for english,french,german,italian in zip(english_sentences, french_sentences,german_sentences,italian_sentences):\n",
+        "    # english = \"[starten] \" + english + \" [enden]\"\n",
+        "    french = \"[startfr] \" + french + \" [endfr]\"\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((english, french))\n",
+        "    text_pairs.append((english, german))\n",
+        "    text_pairs.append((english, italian))\n",
+        "\n",
+        "    # text_pairs.append((french, english))\n",
+        "    # text_pairs.append((french, german))\n",
+        "    # text_pairs.append((french, italian))\n",
+        "\n",
+        "    # text_pairs.append((german, english))\n",
+        "    # text_pairs.append((german, french))\n",
+        "    # text_pairs.append((german, italian))\n",
+        "\n",
+        "    # text_pairs.append((italian, english))\n",
+        "    # text_pairs.append((italian, french))\n",
+        "    # text_pairs.append((italian, german))\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "fXgeKJYJ2F6r"
+      },
+      "source": [
+        "\n",
+        "\n",
+        "# InterReuse (1st scenario) En to {de, it}"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "3UsmQC_159vK"
+      },
+      "outputs": [],
+      "source": [
+        "text_pairs = []\n",
+        "for english,french,german,italian in zip(english_sentences, french_sentences,german_sentences,italian_sentences):\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((english, german))\n",
+        "    text_pairs.append((english, italian))\n",
+        "    "
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "rKyB-Jtn2GRW"
+      },
+      "source": [
+        "# InterReuse (2nd scenario) En to {de, fr}"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "4NME4f-u6ee_"
+      },
+      "outputs": [],
+      "source": [
+        "text_pairs = []\n",
+        "for english,french,german,italian in zip(english_sentences, french_sentences,german_sentences,italian_sentences):\n",
+        "    french = \"[startfr] \" + french + \" [endfr]\"\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    \n",
+        "    text_pairs.append((english, french))\n",
+        "    text_pairs.append((english, german))\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "eGBLp-242GUM"
+      },
+      "source": [
+        "# InterReuse (3rd scenario) En to {it, fr}"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "kiES-xcO6rEf"
+      },
+      "outputs": [],
+      "source": [
+        "text_pairs = []\n",
+        "for english,french,german,italian in zip(english_sentences, french_sentences,german_sentences,italian_sentences):\n",
+        "    french = \"[startfr] \" + french + \" [endfr]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((english, french))\n",
+        "    text_pairs.append((english, italian))\n",
+        "    "
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "-bP39N4viCAV"
+      },
+      "source": [
+        "# Motivating Example 1 Original Model"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "VcIvoR2LiCoN"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/en_ua_train_original.txt')\n",
+        "french_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/fr_ua_train_original.txt')\n",
+        "german_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/de_ua_train_original.txt')\n",
+        "ukranian_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/ua_ua_train_original.txt')\n",
+        "italian_sentences = load_data('/content/drive/MyDrive/NMT/MotivatingExample 1 (Original Model+dataset)/it_ua_train_original.txt')\n",
+        "\n",
+        "text_pairs = []\n",
+        "for english,french,german,italian,ukranian in zip(english_sentences, french_sentences,german_sentences,italian_sentences,ukranian_sentences):\n",
+        "    french = \"[startfr] \" + french + \" [endfr]\"\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "    ukranian = \"[startua] \" + ukranian + \" [endua]\"\n",
+        "\n",
+        "    text_pairs.append((english, french))\n",
+        "    text_pairs.append((english, german))\n",
+        "    text_pairs.append((english, italian))\n",
+        "    text_pairs.append((english, ukranian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "xFmSR42wE1GY"
+      },
+      "source": [
+        "# Motivating Example 1 Solution 1"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "ENunZM3FE23o"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/en_ua_train.txt')\n",
+        "german_sentences = load_data('/content/drive/MyDrive/NMT/de_ua_train.txt')\n",
+        "ukranian_sentences = load_data('/content/drive/MyDrive/NMT/ua_train.txt')\n",
+        "\n",
+        "text_pairs = []\n",
+        "for english, german, ukranian in zip(english_sentences,german_sentences, ukranian_sentences):\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    ukranian = \"[startua] \" + ukranian + \" [endua]\"\n",
+        "\n",
+        "    text_pairs.append((english, german))\n",
+        "    text_pairs.append((english, ukranian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "IUnNMpAVDpZG"
+      },
+      "source": [
+        "# Motivating Example 1 Solution 2"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "T5-vlFmeDrtQ"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/en_ua_train.txt')\n",
+        "ukranian_sentences = load_data('/content/drive/MyDrive/NMT/ua_train.txt')\n",
+        "\n",
+        "text_pairs = []\n",
+        "for english,ukranian in zip(english_sentences, ukranian_sentences):\n",
+        "    ukranian = \"[startua] \" + ukranian + \" [endua]\"\n",
+        "\n",
+        "    text_pairs.append((english, ukranian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "VtPspyGHKpYj"
+      },
+      "source": [
+        "# Motivating Example 2 Original\n",
+        "\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "BkgEgbc6Kteo"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/en_et_train.txt')\n",
+        "german_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/de_et_train.txt')\n",
+        "italian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/it_et_train.txt')\n",
+        "estonian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/et_et_train.txt')\n",
+        "text_pairs = []\n",
+        "for english, german, italian, estonian in zip(english_sentences,german_sentences, italian_sentences,estonian_sentences):\n",
+        "    english = \"[starten] \" + english + \" [enden]\"\n",
+        "    german = \"[startde] \" + german + \" [endde]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((estonian, english))\n",
+        "    text_pairs.append((estonian, german))\n",
+        "    text_pairs.append((estonian, italian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "bUyLIVUHn0Xu"
+      },
+      "source": [
+        "# Motivating Example 2 Solution 1\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "9-0JKeqWn4YW"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/en_et_train.txt')\n",
+        "italian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/it_et_train.txt')\n",
+        "estonian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/et_et_train.txt')\n",
+        "\n",
+        "text_pairs = []\n",
+        "for english, italian, estonian in zip(english_sentences, italian_sentences,estonian_sentences):\n",
+        "    english = \"[starten] \" + english + \" [enden]\"\n",
+        "    italian = \"[startit] \" + italian + \" [endit]\"\n",
+        "\n",
+        "    text_pairs.append((estonian, english))\n",
+        "    text_pairs.append((estonian, italian))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "MI7Ti6Vnkf3v"
+      },
+      "source": [
+        "# Motivating Example 2 Solution 2\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "rxssKJSUkkJP"
+      },
+      "outputs": [],
+      "source": [
+        "english_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/s2/en_et_train_s2.txt')\n",
+        "estonian_sentences = load_data('/content/drive/MyDrive/NMT/Dataset_ME2/s2/et_et_train_s2.txt')\n",
+        "text_pairs = []\n",
+        "for english, estonian in zip(english_sentences,estonian_sentences):\n",
+        "    english = \"[starten] \" + english + \" [enden]\"\n",
+        "\n",
+        "    text_pairs.append((estonian, english))"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "w8044H00hfK-"
+      },
+      "source": [
+        "# Building the model"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "mrG_3rC8h01E",
+        "outputId": "4e2ea5f3-444b-4386-d326-2c26d849f1cb"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "('Do you like peanut butter?', '[startde] Magst du Erdnussbutter? [endde]')\n"
+          ]
+        }
+      ],
+      "source": [
+        "import random\n",
+        "print(random.choice(text_pairs))"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "i4bKiSxSLKMI"
+      },
+      "outputs": [],
+      "source": [
+        "import random\n",
+        "random.shuffle(text_pairs)\n",
+        "num_val_samples = int(0.15 * len(text_pairs))\n",
+        "num_train_samples = len(text_pairs) - 2 * num_val_samples\n",
+        "train_pairs = text_pairs[:num_train_samples]\n",
+        "val_pairs = text_pairs[num_train_samples:num_train_samples + num_val_samples]\n",
+        "test_pairs = text_pairs[num_train_samples + num_val_samples:]"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "CB-X3EfULMec"
+      },
+      "outputs": [],
+      "source": [
+        "import tensorflow as tf\n",
+        "import string\n",
+        "import re\n",
+        "from tensorflow.keras import layers\n",
+        "\n",
+        "strip_chars = string.punctuation + \"¿\"\n",
+        "strip_chars = strip_chars.replace(\"[\", \"\")\n",
+        "strip_chars = strip_chars.replace(\"]\", \"\")\n",
+        "\n",
+        "def custom_standardization(input_string):\n",
+        "    lowercase = tf.strings.lower(input_string)\n",
+        "    return tf.strings.regex_replace(\n",
+        "        lowercase, f\"[{re.escape(strip_chars)}]\", \"\")\n",
+        "\n",
+        "vocab_size = 30000\n",
+        "\n",
+        "#change vocab size for ME2\n",
+        "# vocab_size = 2000\n",
+        "\n",
+        "# en_vocab_size = 11000\n",
+        "# fr_vocab_size = 18000\n",
+        "sequence_length = 20\n",
+        "\n",
+        "source_vectorization = layers.TextVectorization(\n",
+        "    max_tokens=vocab_size,\n",
+        "    output_mode=\"int\",\n",
+        "    output_sequence_length=sequence_length,\n",
+        ")\n",
+        "target_vectorization = layers.TextVectorization(\n",
+        "    max_tokens=vocab_size,\n",
+        "    output_mode=\"int\",\n",
+        "    output_sequence_length=sequence_length + 1,\n",
+        "    standardize=custom_standardization,\n",
+        ")\n",
+        "train_source_texts = [pair[0] for pair in train_pairs]\n",
+        "train_target_texts = [pair[1] for pair in train_pairs]\n",
+        "source_vectorization.adapt(train_source_texts)\n",
+        "target_vectorization.adapt(train_target_texts)\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "IhrYWMyHLPfr"
+      },
+      "outputs": [],
+      "source": [
+        "batch_size = 64\n",
+        "\n",
+        "def format_dataset(eng, spa):\n",
+        "    eng = source_vectorization(eng)\n",
+        "    spa = target_vectorization(spa)\n",
+        "    return ({\n",
+        "        \"source\": eng,\n",
+        "        \"target\": spa[:, :-1],\n",
+        "    }, spa[:, 1:])\n",
+        "\n",
+        "def make_dataset(pairs):\n",
+        "    eng_texts, spa_texts = zip(*pairs)\n",
+        "    eng_texts = list(eng_texts)\n",
+        "    spa_texts = list(spa_texts)\n",
+        "    dataset = tf.data.Dataset.from_tensor_slices((eng_texts, spa_texts))\n",
+        "    dataset = dataset.batch(batch_size)\n",
+        "    dataset = dataset.map(format_dataset, num_parallel_calls=4)\n",
+        "    return dataset.shuffle(2048).prefetch(16).cache()\n",
+        "\n",
+        "train_ds = make_dataset(train_pairs)\n",
+        "val_ds = make_dataset(val_pairs)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "tFFNByKXLRtV",
+        "outputId": "c11030c5-1d46-45b0-e2f1-ed45f27eb087"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "inputs['source'].shape: (64, 20)\n",
+            "inputs['target'].shape: (64, 20)\n",
+            "targets.shape: (64, 20)\n"
+          ]
+        }
+      ],
+      "source": [
+        "for inputs, targets in train_ds.take(1):\n",
+        "    print(f\"inputs['source'].shape: {inputs['source'].shape}\")\n",
+        "    print(f\"inputs['target'].shape: {inputs['target'].shape}\")\n",
+        "    print(f\"targets.shape: {targets.shape}\")"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "kawQLBPHLTmr"
+      },
+      "outputs": [],
+      "source": [
+        "\n",
+        "from tensorflow import keras\n",
+        "from tensorflow.keras import layers\n",
+        "\n",
+        "embed_dim = 200\n",
+        "latent_dim = 1024\n",
+        "\n",
+        "embedding_matrix = load_glob_embedding(vocab_size, 200, target_vectorization.get_vocabulary())\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "HUxTSb5ELV_S",
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "outputId": "a601a9f6-14f9-4634-de7d-e27846c88dc8"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Model: \"model_1\"\n",
+            "__________________________________________________________________________________________________\n",
+            " Layer (type)                   Output Shape         Param #     Connected to                     \n",
+            "==================================================================================================\n",
+            " source (InputLayer)            [(None, None)]       0           []                               \n",
+            "                                                                                                  \n",
+            " embed_encoder (Embedding)      (None, None, 200)    6000000     ['source[0][0]']                 \n",
+            "                                                                                                  \n",
+            " rnn_encoder1 (LSTM)            (None, None, 1024)   5017600     ['embed_encoder[0][0]']          \n",
+            "                                                                                                  \n",
+            " rnn_encoder2 (LSTM)            (None, None, 1024)   8392704     ['rnn_encoder1[0][0]']           \n",
+            "                                                                                                  \n",
+            " target (InputLayer)            [(None, None)]       0           []                               \n",
+            "                                                                                                  \n",
+            " rnn_encoder3 (LSTM)            (None, None, 1024)   8392704     ['rnn_encoder2[0][0]']           \n",
+            "                                                                                                  \n",
+            " embed_decoder (Embedding)      (None, None, 200)    6000000     ['target[0][0]']                 \n",
+            "                                                                                                  \n",
+            " rnn_encoder4 (LSTM)            [(None, 1024),       8392704     ['rnn_encoder3[0][0]']           \n",
+            "                                 (None, 1024),                                                    \n",
+            "                                 (None, 1024)]                                                    \n",
+            "                                                                                                  \n",
+            " rnn_decoder1 (LSTM)            (None, None, 1024)   5017600     ['embed_decoder[0][0]',          \n",
+            "                                                                  'rnn_encoder4[0][1]',           \n",
+            "                                                                  'rnn_encoder4[0][2]']           \n",
+            "                                                                                                  \n",
+            " rnn_decoder2 (LSTM)            (None, None, 1024)   8392704     ['rnn_decoder1[0][0]']           \n",
+            "                                                                                                  \n",
+            " rnn_decoder3 (LSTM)            (None, None, 1024)   8392704     ['rnn_decoder2[0][0]']           \n",
+            "                                                                                                  \n",
+            " rnn_decoder4 (LSTM)            (None, None, 1024)   8392704     ['rnn_decoder3[0][0]']           \n",
+            "                                                                                                  \n",
+            " dropout_1 (Dropout)            (None, None, 1024)   0           ['rnn_decoder4[0][0]']           \n",
+            "                                                                                                  \n",
+            " time_distributed_1 (TimeDistri  (None, None, 30000)  30750000   ['dropout_1[0][0]']              \n",
+            " buted)                                                                                           \n",
+            "                                                                                                  \n",
+            "==================================================================================================\n",
+            "Total params: 103,141,424\n",
+            "Trainable params: 97,141,424\n",
+            "Non-trainable params: 6,000,000\n",
+            "__________________________________________________________________________________________________\n"
+          ]
+        }
+      ],
+      "source": [
+        "source = keras.Input(shape=(None,), dtype=\"int64\", name=\"source\")\n",
+        "\n",
+        "# x = layers.Embedding(vocab_size, embed_dim, mask_zero=True)(source)\n",
+        "x = layers.Embedding(vocab_size, embed_dim, weights=[embedding_matrix], mask_zero=True,\n",
+        "                      name='embed_encoder', trainable=False)(source)\n",
+        "\n",
+        "encoded_source = layers.LSTM(latent_dim, return_sequences=True, activation='tanh', name='rnn_encoder1')(x)\n",
+        "encoded_source = layers.LSTM(latent_dim, return_sequences=True, activation='tanh', name='rnn_encoder2')(encoded_source)\n",
+        "encoded_source = layers.LSTM(latent_dim, return_sequences=True, activation='tanh', name='rnn_encoder3')(encoded_source)\n",
+        "encoded_source, state_h, state_c = layers.LSTM(latent_dim, return_state=True, activation='tanh', name='rnn_encoder4')(encoded_source)\n",
+        "\n",
+        "encoder_states = [state_h, state_c]\n",
+        "\n",
+        "past_target = keras.Input(shape=(None,), dtype=\"int64\", name=\"target\")\n",
+        "x = layers.Embedding(vocab_size, embed_dim, mask_zero=True, name='embed_decoder')(past_target)\n",
+        "\n",
+        "decoder_gru = layers.LSTM(latent_dim, return_sequences=True, activation='tanh', name='rnn_decoder1')\n",
+        "x = decoder_gru(x, initial_state=encoder_states)\n",
+        "x = layers.LSTM(latent_dim, return_sequences=True, activation='tanh', name='rnn_decoder2')(x)\n",
+        "x = layers.LSTM(latent_dim, return_sequences=True, activation='tanh', name='rnn_decoder3')(x)\n",
+        "x = layers.LSTM(latent_dim, return_sequences=True, activation='tanh', name='rnn_decoder4')(x)\n",
+        "\n",
+        "x = layers.Dropout(0.5)(x)\n",
+        "\n",
+        "target_next_step = layers.TimeDistributed(layers.Dense(vocab_size, activation=\"softmax\", name='output'))(x)\n",
+        "\n",
+        "seq2seq_rnn = keras.Model([source, past_target], target_next_step)\n",
+        "\n",
+        "seq2seq_rnn.compile(\n",
+        "    optimizer=\"rmsprop\",\n",
+        "    loss=\"sparse_categorical_crossentropy\",\n",
+        "    metrics=[\"accuracy\"])\n",
+        "\n",
+        "seq2seq_rnn.summary()"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "twWGTTxHLX0-",
+        "outputId": "565dea86-8562-4561-bb1b-ba936d14f7d9"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "Epoch 1/3\n",
+            "26/26 [==============================] - 38s 436ms/step - loss: 1.9507 - accuracy: 0.1266 - val_loss: 1.5340 - val_accuracy: 0.1645\n",
+            "Epoch 2/3\n",
+            "26/26 [==============================] - 2s 88ms/step - loss: 1.6646 - accuracy: 0.1742 - val_loss: 1.4814 - val_accuracy: 0.1751\n",
+            "Epoch 3/3\n",
+            "26/26 [==============================] - 2s 88ms/step - loss: 1.6099 - accuracy: 0.1869 - val_loss: 1.4740 - val_accuracy: 0.1777\n"
+          ]
+        }
+      ],
+      "source": [
+        "# vocab_size = 30000\n",
+        "# sequence_length = 20\n",
+        "\n",
+        "# checkpoint = ModelCheckpoint(filepath='model_1LSTM_original_chkpt.h5',\n",
+        "#                               monitor='val_loss', verbose=1, save_best_only=True,\n",
+        "#                               mode='min')\n",
+        "\n",
+        "seq2seq_rnn.fit(train_ds, epochs=3, validation_data=val_ds)\n",
+        "seq2seq_rnn.save('solution1_problem2.h5')"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "background_save": true
+        },
+        "id": "0SkImFO6os8J",
+        "outputId": "c2d98cb7-90f6-484e-d3c3-7b2f094abdf7"
+      },
+      "outputs": [
+        {
+          "ename": "KeyboardInterrupt",
+          "evalue": "ignored",
+          "output_type": "error",
+          "traceback": [
+            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
+            "\u001b[0;32m<ipython-input-31-9a163dae5324>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     33\u001b[0m     \u001b[0;31m# print(\"predicted: \", decode_sequence(test_eng_texts[i],acts[0], acts[-1]))\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     34\u001b[0m     \u001b[0mactual\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0macts\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 35\u001b[0;31m     \u001b[0mpredicted\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdecode_sequence\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtest_eng_texts\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0macts\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0macts\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msplit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     36\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     37\u001b[0m \u001b[0;31m#print(\"actual: \", actual, '\\n', \"predicted:\", predicted)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m<ipython-input-31-9a163dae5324>\u001b[0m in \u001b[0;36mdecode_sequence\u001b[0;34m(input_sentence, decoded_sentence, end_tag)\u001b[0m\n\u001b[1;32m     11\u001b[0m         \u001b[0mtokenized_target_sentence\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtarget_vectorization\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mdecoded_sentence\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m         next_token_predictions = seq2seq_rnn.predict(\n\u001b[0;32m---> 13\u001b[0;31m             [tokenized_input_sentence, tokenized_target_sentence])\n\u001b[0m\u001b[1;32m     14\u001b[0m         \u001b[0msampled_token_index\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0margmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnext_token_predictions\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mi\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m         \u001b[0msampled_token\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mspa_index_lookup\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0msampled_token_index\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     62\u001b[0m     \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     63\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 64\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     65\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint: disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, x, batch_size, verbose, steps, callbacks, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1976\u001b[0m       \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_predict_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1977\u001b[0m       \u001b[0mbatch_outputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1978\u001b[0;31m       \u001b[0;32mfor\u001b[0m \u001b[0m_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0miterator\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0menumerate_epochs\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# Single epoch.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1979\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcatch_stop_iteration\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1980\u001b[0m           \u001b[0;32mfor\u001b[0m \u001b[0mstep\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msteps\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/engine/data_adapter.py\u001b[0m in \u001b[0;36menumerate_epochs\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1189\u001b[0m     \u001b[0;34m\"\"\"Yields `(epoch, tf.data.Iterator)`.\"\"\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1190\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_truncate_execution_to_epoch\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1191\u001b[0;31m       \u001b[0mdata_iterator\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0miter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_dataset\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1192\u001b[0m       \u001b[0;32mfor\u001b[0m \u001b[0mepoch\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_initial_epoch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_epochs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1193\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_insufficient_data\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# Set by `catch_stop_iteration`.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/data/ops/dataset_ops.py\u001b[0m in \u001b[0;36m__iter__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    484\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mcontext\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexecuting_eagerly\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0minside_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    485\u001b[0m       \u001b[0;32mwith\u001b[0m \u001b[0mops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcolocate_with\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_variant_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 486\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0miterator_ops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOwnedIterator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    487\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m       raise RuntimeError(\"`tf.data.Dataset` only supports Python-style \"\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/data/ops/iterator_ops.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, dataset, components, element_spec)\u001b[0m\n\u001b[1;32m    753\u001b[0m             \u001b[0;34m\"When `dataset` is provided, `element_spec` and `components` must \"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    754\u001b[0m             \"not be specified.\")\n\u001b[0;32m--> 755\u001b[0;31m       \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_create_iterator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    756\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    757\u001b[0m     \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_get_next_call_count\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/data/ops/iterator_ops.py\u001b[0m in \u001b[0;36m_create_iterator\u001b[0;34m(self, dataset)\u001b[0m\n\u001b[1;32m    785\u001b[0m                 \u001b[0moutput_types\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_flat_output_types\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    786\u001b[0m                 output_shapes=self._flat_output_shapes))\n\u001b[0;32m--> 787\u001b[0;31m         \u001b[0mgen_dataset_ops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmake_iterator\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mds_variant\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_iterator_resource\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    788\u001b[0m         \u001b[0;31m# Delete the resource when this object is deleted\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    789\u001b[0m         self._resource_deleter = IteratorResourceDeleter(\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/gen_dataset_ops.py\u001b[0m in \u001b[0;36mmake_iterator\u001b[0;34m(dataset, iterator, name)\u001b[0m\n\u001b[1;32m   3314\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3315\u001b[0m       _result = pywrap_tfe.TFE_Py_FastPathExecute(\n\u001b[0;32m-> 3316\u001b[0;31m         _ctx, \"MakeIterator\", name, dataset, iterator)\n\u001b[0m\u001b[1;32m   3317\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0m_result\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3318\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0m_core\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
+          ]
+        }
+      ],
+      "source": [
+        "import numpy as np\n",
+        "spa_vocab = target_vectorization.get_vocabulary()\n",
+        "spa_index_lookup = dict(zip(range(len(spa_vocab)), spa_vocab))\n",
+        "max_decoded_sentence_length = 20\n",
+        "\n",
+        "def decode_sequence(input_sentence, decoded_sentence, end_tag):\n",
+        "    tokenized_input_sentence = source_vectorization([input_sentence])\n",
+        "    #decoded_sentence = \"[start]\"\n",
+        "\n",
+        "    for i in range(max_decoded_sentence_length):\n",
+        "        tokenized_target_sentence = target_vectorization([decoded_sentence])\n",
+        "        next_token_predictions = seq2seq_rnn.predict(\n",
+        "            [tokenized_input_sentence, tokenized_target_sentence])\n",
+        "        sampled_token_index = np.argmax(next_token_predictions[0, i, :])\n",
+        "        sampled_token = spa_index_lookup[sampled_token_index]\n",
+        "        decoded_sentence += \" \" + sampled_token\n",
+        "        if sampled_token == end_tag:\n",
+        "            break\n",
+        "    return decoded_sentence\n",
+        "\n",
+        "bleu_dic = {}\n",
+        "test_eng_texts = [pair[0] for pair in test_pairs]\n",
+        "test_fr_texts = [pair[1] for pair in test_pairs]\n",
+        "actual, predicted = [], []\n",
+        "for i in range(len(test_pairs)):\n",
+        "    # input_sentence = \n",
+        "    # target_sentence = random.choice(test_fr_texts)\n",
+        "    acts=test_fr_texts[i].split()\n",
+        "    # print(\"-\")\n",
+        "    # print(acts[0], acts[-1])\n",
+        "    # print(\"source: \", test_eng_texts[i])\n",
+        "    # print(\"actual target: \", test_fr_texts[i])\n",
+        "    # print(\"predicted: \", decode_sequence(test_eng_texts[i],acts[0], acts[-1]))\n",
+        "    actual.append([acts])\n",
+        "    predicted.append(decode_sequence(test_eng_texts[i],acts[0], acts[-1]).split())\n",
+        "\n",
+        "#print(\"actual: \", actual, '\\n', \"predicted:\", predicted)\n",
+        "bleu_dic['1-grams'] = corpus_bleu(actual, predicted, weights=(1.0, 0, 0, 0))\n",
+        "bleu_dic['1-2-grams'] = corpus_bleu(actual, predicted, weights=(0.5, 0.5, 0, 0))\n",
+        "bleu_dic['1-3-grams'] = corpus_bleu(actual, predicted, weights=(0.3, 0.3, 0.3, 0))\n",
+        "bleu_dic['1-4-grams'] = corpus_bleu(actual, predicted, weights=(0.25, 0.25, 0.25, 0.25))\n",
+        "print(bleu_dic)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 281
+        },
+        "id": "D_Qf1ZnRNual",
+        "outputId": "ef5196ee-1377-4041-cc41-1c0fdd71a911"
+      },
+      "outputs": [
+        {
+          "data": {
+            "image/png": "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\n",
+            "text/plain": [
+              "<Figure size 432x288 with 1 Axes>"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        }
+      ],
+      "source": [
+        "from nltk.translate.bleu_score import corpus_bleu\n",
+        "import matplotlib.pyplot as plt\n",
+        "\n",
+        "plt.bar(x = bleu_dic.keys(), height = bleu_dic.values())\n",
+        "plt.title(\"BLEU Score with the testing set\")\n",
+        "plt.ylim((0,1))\n",
+        "plt.show()"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "ThXu38YNo6LQ",
+        "outputId": "540b3456-f792-4b1e-d605-c7d320bf5729"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Mounted at /content/drive/\n"
+          ]
+        }
+      ],
+      "source": [
+        "from google.colab import drive\n",
+        "\n",
+        "drive.mount(\"/content/drive/\")\n",
+        "\n",
+        "#files.download(\"/content/model4_final_reuse_2.h5\")\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 425
+        },
+        "id": "6WX179UgqihS",
+        "outputId": "7ba02150-d3d8-4bbc-8186-015ad7c1e92e"
+      },
+      "outputs": [
+        {
+          "ename": "ValueError",
+          "evalue": "ignored",
+          "output_type": "error",
+          "traceback": [
+            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+            "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
+            "\u001b[0;32m<ipython-input-15-38f909bf98ba>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     14\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDropout\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0.5\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mSimpleRNN\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlatent_dim\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactivation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'relu'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'rnn_decoder2'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 16\u001b[0;31m \u001b[0mx\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mSimpleRNN\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlatent_dim\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactivation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'relu'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'rnn_decoder3'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     17\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m \u001b[0mtarget_next_step\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTimeDistributed\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlayers\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDense\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mvocab_size\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactivation\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m\"softmax\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'output'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/layers/recurrent.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs, initial_state, constants, **kwargs)\u001b[0m\n\u001b[1;32m    677\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    678\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0minitial_state\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mconstants\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 679\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0msuper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mRNN\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__call__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minputs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    680\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    681\u001b[0m     \u001b[0;31m# If any of `initial_state` or `constants` are specified and are Keras\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     65\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint: disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 67\u001b[0;31m       \u001b[0;32mraise\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfiltered_tb\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     68\u001b[0m     \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     69\u001b[0m       \u001b[0;32mdel\u001b[0m \u001b[0mfiltered_tb\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/engine/input_spec.py\u001b[0m in \u001b[0;36massert_input_compatibility\u001b[0;34m(input_spec, inputs, layer_name)\u001b[0m\n\u001b[1;32m    212\u001b[0m       \u001b[0mndim\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mshape\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrank\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    213\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mndim\u001b[0m \u001b[0;34m!=\u001b[0m \u001b[0mspec\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndim\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 214\u001b[0;31m         raise ValueError(f'Input {input_index} of layer \"{layer_name}\" '\n\u001b[0m\u001b[1;32m    215\u001b[0m                          \u001b[0;34m'is incompatible with the layer: '\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    216\u001b[0m                          \u001b[0;34mf'expected ndim={spec.ndim}, found ndim={ndim}. '\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;31mValueError\u001b[0m: Input 0 of layer \"rnn_decoder3\" is incompatible with the layer: expected ndim=3, found ndim=2. Full shape received: (None, 1024)"
+          ]
+        }
+      ],
+      "source": [
+        "source = keras.Input(shape=(None,), dtype=\"int64\", name=\"source\")\n",
+        "\n",
+        "# x = layers.Embedding(vocab_size, embed_dim, mask_zero=True)(source)\n",
+        "x = layers.Embedding(vocab_size, embed_dim, weights=[embedding_matrix], mask_zero=True,\n",
+        "                      name='embed_encoder', trainable=False)(source)\n",
+        "\n",
+        "encoded_source = layers.SimpleRNN(latent_dim, activation='relu', name='rnn_encoder4')(x)\n",
+        "\n",
+        "past_target = keras.Input(shape=(None,), dtype=\"int64\", name=\"target\")\n",
+        "x = layers.Embedding(vocab_size, embed_dim, mask_zero=True, name='embed_decoder')(past_target)\n",
+        "\n",
+        "decoder_gru = layers.SimpleRNN(latent_dim, return_sequences=True, activation='relu', name='rnn_decoder1')\n",
+        "x = decoder_gru(x, initial_state=encoded_source)\n",
+        "x = layers.Dropout(0.5)(x)\n",
+        "x = layers.SimpleRNN(latent_dim, activation='relu', name='rnn_decoder2')(x)\n",
+        "x = layers.SimpleRNN(latent_dim, return_sequences=True, activation='relu', name='rnn_decoder3')(x)\n",
+        "\n",
+        "target_next_step = layers.TimeDistributed(layers.Dense(vocab_size, activation=\"softmax\", name='output'))(x)\n",
+        "\n",
+        "seq2seq_rnn = keras.Model([source, past_target], target_next_step)\n",
+        "\n",
+        "seq2seq_rnn.compile(\n",
+        "    optimizer=\"rmsprop\",\n",
+        "    loss=\"sparse_categorical_crossentropy\",\n",
+        "    metrics=[\"accuracy\"])"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 222
+        },
+        "id": "NvlZdKQ2qk6a",
+        "outputId": "c6dd1258-df31-4b5a-e79d-91e2d83f21af"
+      },
+      "outputs": [
+        {
+          "ename": "NameError",
+          "evalue": "ignored",
+          "output_type": "error",
+          "traceback": [
+            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
+            "\u001b[0;32m<ipython-input-1-26deb6ad3f93>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;31m#                               mode='min')\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m \u001b[0mseq2seq_rnn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain_ds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalidation_data\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mval_ds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallbacks\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mcheckpoint\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m \u001b[0;31m# seq2seq_rnn.save('model4_final_ME_2.h5')\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+            "\u001b[0;31mNameError\u001b[0m: name 'seq2seq_rnn' is not defined"
+          ]
+        }
+      ],
+      "source": [
+        "vocab_size = 30000\n",
+        "sequence_length = 20\n",
+        "\n",
+        "# checkpoint = ModelCheckpoint(filepath='model4_ME_2.h5',\n",
+        "#                               monitor='val_loss', verbose=1, save_best_only=True,\n",
+        "#                               mode='min')\n",
+        "\n",
+        "seq2seq_rnn.fit(train_ds, epochs=3, validation_data=val_ds, callbacks=[checkpoint])\n",
+        "# seq2seq_rnn.save('model4_final_ME_2.h5')\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "BRc0Z_ELHpdD"
+      },
+      "source": [
+        "### **The Codes after this are from prior experiment.**"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "stJYZc1eh2Kw"
+      },
+      "outputs": [],
+      "source": [
+        "def tokenize(x):\n",
+        "    \"\"\"\n",
+        "    Tokenize x\n",
+        "    :param x: List of sentences/strings to be tokenized\n",
+        "    :return: Tuple of (tokenized x data, tokenizer used to tokenize x)\n",
+        "    \"\"\"\n",
+        "    # TODO: Implement\n",
+        "    tokenizer = Tokenizer()\n",
+        "    tokenizer.fit_on_texts(x)\n",
+        "    return tokenizer.texts_to_sequences(x), tokenizer"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "CVZgVHkzh3tr"
+      },
+      "outputs": [],
+      "source": [
+        "def pad(x, length=None):\n",
+        "    \"\"\"\n",
+        "    Pad x\n",
+        "    :param x: List of sequences.\n",
+        "    :param length: Length to pad the sequence to.  If None, use length of longest sequence in x.\n",
+        "    :return: Padded numpy array of sequences\n",
+        "    \"\"\"\n",
+        "#     # TODO: Implement\n",
+        "#     if length is None:\n",
+        "#         length=max([len(sentence) for sentence in x])\n",
+        "#         print(length)\n",
+        "    \n",
+        "    return pad_sequences(x, maxlen=20, padding ='post')"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "msE26JGmh5TA"
+      },
+      "outputs": [],
+      "source": [
+        "import collections\n",
+        "\n",
+        "english_words_counter = collections.Counter([word for sentence in english_sentences for word in sentence.split()])\n",
+        "french_words_counter = collections.Counter([word for sentence in french_sentences for word in sentence.split()])\n",
+        "\n",
+        "print('{} English words.'.format(len([word for sentence in english_sentences for word in sentence.split()])))\n",
+        "print('{} unique English words.'.format(len(english_words_counter)))\n",
+        "print('10 Most common words in the English dataset:')\n",
+        "print('\"' + '\" \"'.join(list(zip(*english_words_counter.most_common(10)))[0]) + '\"')\n",
+        "print()\n",
+        "print('{} French words.'.format(len([word for sentence in french_sentences for word in sentence.split()])))\n",
+        "print('{} unique French words.'.format(len(french_words_counter)))\n",
+        "print('10 Most common words in the French dataset:')\n",
+        "print('\"' + '\" \"'.join(list(zip(*french_words_counter.most_common(10)))[0]) + '\"')"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "GHOxz_1Fh7Ha"
+      },
+      "outputs": [],
+      "source": [
+        "for sample_i in range(5):\n",
+        "    print('English sample {}:  {}'.format(sample_i + 1, english_sentences[sample_i+10000]))\n",
+        "    print('French sample {}:  {}\\n'.format(sample_i + 1, french_sentences[sample_i+10000]))\n",
+        "    print('German sample {}:  {}\\n'.format(sample_i + 1, german_sentences[sample_i+10000]))\n",
+        "    print('Italian sample {}:  {}\\n'.format(sample_i + 1, italian_sentences[sample_i+10000]))\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "ogGPGCf7h9Gw"
+      },
+      "outputs": [],
+      "source": [
+        "def preprocess(x, y1, y2, y3):\n",
+        "    \"\"\"\n",
+        "    Preprocess x and y\n",
+        "    :param x: Feature List of sentences\n",
+        "    :param y: Label List of sentences\n",
+        "    :return: Tuple of (Preprocessed x, Preprocessed y, x tokenizer, y tokenizer)\n",
+        "    \"\"\"\n",
+        "    preprocess_en, en_tk = tokenize(x)\n",
+        "    preprocess_fr, fr_tk = tokenize(y1)\n",
+        "    preprocess_de, de_tk = tokenize(y2)\n",
+        "    preprocess_it, it_tk = tokenize(y3)\n",
+        "    \n",
+        "    preprocess_en = pad(preprocess_en)\n",
+        "    preprocess_fr = pad(preprocess_fr)\n",
+        "    preprocess_de = pad(preprocess_de)\n",
+        "    preprocess_it = pad(preprocess_it)\n",
+        "\n",
+        "    \n",
+        "    # Keras's sparse_categorical_crossentropy function requires the labels to be in 3 dimensions\n",
+        "    preprocess_fr = preprocess_fr.reshape(*preprocess_fr.shape, 1)\n",
+        "    preprocess_de = preprocess_de.reshape(*preprocess_de.shape, 1)\n",
+        "    preprocess_it = preprocess_it.reshape(*preprocess_it.shape, 1)\n",
+        "\n",
+        "    return preprocess_en,preprocess_fr,preprocess_de,preprocess_it, en_tk, fr_tk, de_tk, it_tk\n",
+        "\n",
+        "inputTimestep = 30\n",
+        "outputTimestep = 30\n",
+        "\n",
+        "preproc_english_sentences,preproc_french_sentences, preproc_german_sentences, preproc_italian_sentences, en_tokenizer,fr_tokenizer, de_tokenizer, it_tokenizer =\\\n",
+        "    preprocess(english_sentences, french_sentences,german_sentences,italian_sentences )\n",
+        "\n",
+        "\n",
+        "max_english_sequence_length = preproc_english_sentences.shape[1]\n",
+        "max_french_sequence_length = preproc_french_sentences.shape[1]\n",
+        "max_german_sequence_length = preproc_german_sentences.shape[1]\n",
+        "max_italian_sequence_length = preproc_italian_sentences.shape[1]\n",
+        "\n",
+        "english_vocab_size = len(en_tokenizer.word_index)\n",
+        "french_vocab_size = len(fr_tokenizer.word_index)\n",
+        "german_vocab_size = len(de_tokenizer.word_index)\n",
+        "italian_vocab_size = len(it_tokenizer.word_index)\n",
+        "\n",
+        "print('Data Preprocessed')\n",
+        "\n",
+        "print(\"Max English sentence length:\", max_english_sequence_length)\n",
+        "print(\"Max French sentence length:\", max_french_sequence_length)\n",
+        "print(\"Max German sentence length:\", max_german_sequence_length)\n",
+        "print(\"Max Italian sentence length:\", max_italian_sequence_length)\n",
+        "\n",
+        "print(\"English vocabulary size:\", english_vocab_size)\n",
+        "print(\"French vocabulary size:\", french_vocab_size)\n",
+        "print(\"German vocabulary size:\", german_vocab_size)\n",
+        "print(\"Italian vocabulary size:\", italian_vocab_size)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "zdLpnbLHiEk3"
+      },
+      "outputs": [],
+      "source": [
+        "from keras.layers import GRU, Input, Dense, TimeDistributed, Activation, RepeatVector, Bidirectional, Dropout, LSTM\n",
+        "from keras.losses import sparse_categorical_crossentropy\n",
+        "from keras.models import Sequential\n",
+        "from keras.layers import Dense, Activation, TimeDistributed, RepeatVector, Flatten, Conv2D, Embedding\n",
+        "from keras.layers.recurrent import SimpleRNN, LSTM\n",
+        "from keras.utils import np_utils\n",
+        "from tensorflow.keras.models import Model\n",
+        "import keras\n",
+        "\n",
+        "\n",
+        "def many_many_tangled(input_shape, fr_output_sequence_length, english_vocab_size, french_vocab_size):\n",
+        "\n",
+        "\n",
+        "    # Hyperparameters\n",
+        "    opt = tf.keras.optimizers.Adam(learning_rate=1e-3)\n",
+        "    \n",
+        "    # Build the layers    \n",
+        "    model = Sequential()\n",
+        "    # Embedding\n",
+        "    model.add(Embedding(english_vocab_size, 256, input_length=input_shape[1],\n",
+        "                         input_shape=input_shape[1:]))\n",
+        "    # Encoder\n",
+        "    model.add(SimpleRNN(256))\n",
+        "    model.add(RepeatVector(fr_output_sequence_length))\n",
+        "    # Decoder\n",
+        "    model.add(SimpleRNN(256, return_sequences=True))\n",
+        "    model.add(TimeDistributed(Dense(512, activation='relu')))\n",
+        "    model.add(Dropout(0.5))\n",
+        "    model.add(TimeDistributed(Dense((french_vocab_size), activation='softmax')))\n",
+        "    model.compile(loss=sparse_categorical_crossentropy,\n",
+        "                  optimizer=opt,\n",
+        "                  metrics=['accuracy'])\n",
+        "    \n",
+        "    print(model.summary())\n",
+        "\n",
+        "    return model\n",
+        "\n",
+        "def many_many_functional(input_shape, output_sequence_length, english_vocab_size, french_vocab_size, german_vocab_size,italian_vocab_size):\n",
+        "    \n",
+        "    #input\n",
+        "    eng_input = Input(shape=(None,), dtype=\"int64\", name=\"english\")\n",
+        "\n",
+        "    #embedding\n",
+        "    embedding_layer = Embedding(english_vocab_size, 256)(eng_input)\n",
+        "\n",
+        "    rnn_layer_1 = SimpleRNN(256)(embedding_layer)\n",
+        "\n",
+        "    fr_input = Input(shape=(None,), dtype=\"int64\", name=\"spanish\")\n",
+        "\n",
+        "    embedding_layer = Embedding(french_vocab_size, 256)(fr_input)\n",
+        "\n",
+        "    #Encoder for two langauges\n",
+        "    rnn_layer_1 = SimpleRNN(256)(embedding_layer)\n",
+        "    repeat_vector = RepeatVector(output_sequence_length)(rnn_layer_1)\n",
+        "\n",
+        "    #Common decoder for all languages\n",
+        "    rnn_layer2 = SimpleRNN(256, return_sequences=True)(repeat_vector)\n",
+        "    time_distributed_1 = Dense(1024, activation='relu')(rnn_layer2)\n",
+        "    dropout_1 = Dropout(0.5)(time_distributed_1)\n",
+        "    \n",
+        "    output_fr = Dense(french_vocab_size, activation='softmax')(dropout_1)\n",
+        "    #output_de = TimeDistributed(Dense(german_vocab_size, activation='softmax'))(dropout_1)\n",
+        "    #output_it = TimeDistributed(Dense(italian_vocab_size, activation='softmax'))(dropout_1)\n",
+        "    \n",
+        "    #Create model\n",
+        "    #model = Model(inputs=eng_input, outputs=[output_fr,output_de,output_it])\n",
+        "    model = Model(inputs=[eng_input, fr_input], outputs=output_fr)\n",
+        "\n",
+        "\n",
+        "    model.compile(loss=sparse_categorical_crossentropy, optimizer='adam',metrics=['accuracy'])\n",
+        "\n",
+        "    print(model.summary())\n",
+        "    \n",
+        "    return model\n",
+        "\n",
+        "def embed_model(output_sequence_length, english_vocab_size, french_vocab_size):\n",
+        "\n",
+        "    # Hyperparameters\n",
+        "    opt = tf.keras.optimizers.Adam(learning_rate=1e-3)\n",
+        "    # Build the layers    \n",
+        "    model = Sequential()\n",
+        "    # Embedding\n",
+        "    model.add(Embedding(english_vocab_size, 256))\n",
+        "    # Encoder\n",
+        "    model.add(SimpleRNN(256))\n",
+        "    model.add(RepeatVector(output_sequence_length))\n",
+        "    # Decoder\n",
+        "    model.add(SimpleRNN(256, return_sequences=True))\n",
+        "    model.add(TimeDistributed(Dense(516, activation='relu')))\n",
+        "    model.add(Dropout(0.5))\n",
+        "    model.add(TimeDistributed(Dense(516, activation='relu')))\n",
+        "    model.add(Dropout(0.5))\n",
+        "    model.add(TimeDistributed(Dense(french_vocab_size, activation='softmax')))\n",
+        "    model.compile(loss=sparse_categorical_crossentropy,\n",
+        "                  optimizer=opt,\n",
+        "                  metrics=['accuracy'])\n",
+        "    return model\n",
+        "\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "VdDy5uL1t8Sy"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "CeQ4Z6j1spTP"
+      },
+      "outputs": [],
+      "source": []
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "Qr25F675iIyX"
+      },
+      "outputs": [],
+      "source": [
+        "# tmp_x = pad(preproc_english_sentences, preproc_french_sentences.shape[1])\n",
+        "# tmp_x = tmp_x.reshape((-1, preproc_french_sentences.shape[-2]))\n",
+        "\n",
+        "reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2,\n",
+        "                              patience=5, min_lr=0.001)\n",
+        "\n",
+        "callback = tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=5)\n",
+        "# TODO: Train the neural network\n",
+        "\n",
+        "many_many = many_many_functional(\n",
+        "    None,\n",
+        "    20,\n",
+        "    en_vocab_size+1,\n",
+        "    fr_vocab_size+1,\n",
+        "    None, None)\n",
+        "\n",
+        "many_many.summary()\n",
+        "\n",
+        "many_many.fit(train_ds, validation_data=val_ds, batch_size=64, epochs=30, callbacks=[callback, reduce_lr])"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "Zt2YCfXosq05"
+      },
+      "outputs": [],
+      "source": [
+        "def logits_to_text(logits, tokenizer):\n",
+        "    \"\"\"\n",
+        "    Turn logits from a neural network into text using the tokenizer\n",
+        "    :param logits: Logits from a neural network\n",
+        "    :param tokenizer: Keras Tokenizer fit on the labels\n",
+        "    :return: String that represents the text of the logits\n",
+        "    \"\"\"\n",
+        "    index_to_words = {id: word for word, id in tokenizer.word_index.items()}\n",
+        "    index_to_words[0] = '<PAD>'\n",
+        "\n",
+        "    return ' '.join([index_to_words[prediction] for prediction in np.argmax(logits, 1)])\n",
+        "\n",
+        "print('`logits_to_text` function loaded.')"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "id": "AS88MAqtstIC"
+      },
+      "outputs": [],
+      "source": [
+        "# Print prediction(s)\n",
+        "print(\"Prediction:\")\n",
+        "\n",
+        "print(logits_to_text(many_many.predict(tmp_x[6:7])[0], fr_tokenizer))\n",
+        "\n",
+        "print(\"\\nCorrect Translation French:\")\n",
+        "print(french_sentences[6:7])\n",
+        "\n",
+        "print(\"\\nOriginal text:\")\n",
+        "print()"
+      ]
+    }
+  ],
+  "metadata": {
+    "accelerator": "GPU",
+    "colab": {
+      "machine_shape": "hm",
+      "provenance": []
+    },
+    "gpuClass": "standard",
+    "kernelspec": {
+      "display_name": "Python 3",
+      "name": "python3"
+    },
+    "language_info": {
+      "name": "python"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}

language_models/vanilla_rnn/motivating example models/problem1/original_problem1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33bff93b68beb81d69eaa91f563eead037f4279934e9c7a76a9f6d5aceb1a652
+size 438911736

language_models/vanilla_rnn/motivating example models/problem1/solution1_problem1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9389ca9eaeb58d3faf80fd6e263e7564bdda7fbc0e0ee29a2584b57e4d606fef
+size 438911712

language_models/vanilla_rnn/motivating example models/problem1/solution2_problem1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2e7ae72ab4ca1d09ee686d252ee560c89973a8e1fd4f6a5c6781f98441db458
+size 438911712

language_models/vanilla_rnn/motivating example models/problem2/original_problem2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4938d3a6be16e2dae5c2d9aa600ffbcb8c503a3a7bf7bce800b3bc91d0719864
+size 438911736

language_models/vanilla_rnn/motivating example models/problem2/solution1_problem2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e11189c864a9b76e8d1574de90d6651ce3b16513b89eb607a2bcdbc2703199c6
+size 438911712

language_models/vanilla_rnn/motivating example models/problem2/solution2_problem2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c539a9274240f3fa281bfb5f7b1ce68d9c08096e9b2f5dc712b2e62696df5fd
+size 438911712

language_models/vanilla_rnn/reuse models/model4_de_fr.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ee9e1beb8bb17f529b0fb1b0d5a78bcfc3974c0c9e1456f7a0dffabb46646fa
+size 438911712

language_models/vanilla_rnn/reuse models/model4_de_it.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72f7d12ea76cb2dafb7c2a62489f958007e663ded5eae27acf7dc9469f004025
+size 438911712

language_models/vanilla_rnn/reuse models/model4_fr_it.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e03ed8172d5f376e59583e97a82fd77d8bd27825e5091eac6e8d9e1987c89af
+size 438911712

language_models/vanilla_rnn/rq1 models/model1.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8637147ae3371c50cc1488264a567360d04c708a8d67ec0a56b2efe7d740c54e
+size 338125960

language_models/vanilla_rnn/rq1 models/model2.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fff06c0f77191431baca5ef487f4fd0b331b3c53e02cbe0e9a83e20cb3e0b348
+size 371721312