diff --git a/LICENSE.md b/LICENSE.md
new file mode 100644
index 0000000..2db9257
--- /dev/null
+++ b/LICENSE.md
@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 Tocard-inc
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
diff --git a/README.md b/README.md
new file mode 100644
index 0000000..dd102cf
--- /dev/null
+++ b/README.md
@@ -0,0 +1,25 @@
+# README
+
+Ce repository contient l'implémentation de deux réseaux de deep learning pour l'apprentissage semi-supervisé: simCLR et SGAN. Les deux réseaux ont été testés sur des jeux de données de classification d'images.
+
+## simCLR
+
+simCLR est une approche semi-supervisée pour l'apprentissage de représentations à partir d'un grand nombre d'images non étiquetées. Le modèle consiste en une chaîne de transformation qui transforme les entrées dans un espace d'apprentissage continu et en un classificateur qui prédit les labels des entrées transformées.
+
+## SGAN
+
+SGAN est un réseau GAN semi-supervisé qui utilise une générateur pour produire des images aléatoires et un discriminateur pour les classer en fonction des labels connus. Le but est de produire des images qui sont à la fois réalistes et informatives pour l'apprentissage de la classification.
+Comment utiliser ce repository
+
+## Comment utiliser ce repository
+
+Les notebooks (colab) se situent dans le dossier `src`.
+
+
+## Contacts
+
+- Laurent Fainsin <laurent@fainsin.bzh>
+- Damien Guillotin <damien.guillotin@etu.inp-n7.fr>
+- Pierre-Eliot Jourdan <pierreeliot.jourdan@etu.inp-n7.fr>
+
+## License MIT
diff --git a/src/notebook_simCLR.ipynb b/src/notebook_simCLR.ipynb
new file mode 100644
index 0000000..9f43662
--- /dev/null
+++ b/src/notebook_simCLR.ipynb
@@ -0,0 +1,5264 @@
+{
+  "cells": [
+    {
+      "cell_type": "code",
+      "execution_count": 1,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 2337,
+          "status": "ok",
+          "timestamp": 1672849277784,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "6Rh1k-zKS1i3"
+      },
+      "outputs": [],
+      "source": [
+        "import numpy as np\n",
+        "import os\n",
+        "from PIL import Image\n",
+        "\n",
+        "import math\n",
+        "import matplotlib.pyplot as plt\n",
+        "import tensorflow as tf\n",
+        "\n",
+        "from tensorflow import keras\n",
+        "from tensorflow.keras import layers\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 2,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 7,
+          "status": "ok",
+          "timestamp": 1672849277784,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "ADrwcAUsM3oy"
+      },
+      "outputs": [],
+      "source": [
+        "tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.ERROR)\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 62,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 2,
+          "status": "ok",
+          "timestamp": 1672861924874,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "oMm3tQbz_L9I"
+      },
+      "outputs": [],
+      "source": [
+        "\"\"\"Code to build custom dataset \"projsemisup\".\"\"\"\n",
+        "import tensorflow_datasets.public_api as tfds\n",
+        "import itertools\n",
+        "import pathlib\n",
+        "import os\n",
+        "\n",
+        "UNLABELLED = tfds.Split(\"unlabelled\")\n",
+        "\n",
+        "COMMIT = \"bbebb1e999f95086dc969217d52dc08f95f03f46\"\n",
+        "URL = f\"https://github.com/axelcarlier/projsemisup/archive/{COMMIT}.zip\"\n",
+        "IMG_SHAPE = (128, 128, 3)\n",
+        "CLASSES = [\n",
+        "  \"antilope\",\n",
+        "  \"autruche\",\n",
+        "  \"Buffle d'Afrique\",\n",
+        "  \"Chacal\",\n",
+        "  \"Chimpanzé\",\n",
+        "  \"Elephant\",\n",
+        "  \"Girafe\",\n",
+        "  \"Gnou\",\n",
+        "  \"Gorilla\",\n",
+        "  \"Guepard\",\n",
+        "  \"Hippopotame\",\n",
+        "  \"hyène\",\n",
+        "  \"Leopard\",\n",
+        "  \"Lion\",\n",
+        "  \"Phacochère\",\n",
+        "  \"Rhinoceros\",\n",
+        "  \"tigre\",\n",
+        "  \"Zèbre\",\n",
+        "]\n",
+        "\n",
+        "class Projsemisup(tfds.core.GeneratorBasedBuilder):\n",
+        "  \"\"\"projsemisup dataset.\"\"\"\n",
+        "\n",
+        "  VERSION = tfds.core.Version(\"1.1.0\")\n",
+        "\n",
+        "  def _info(self):\n",
+        "    return tfds.core.DatasetInfo(\n",
+        "        builder=self,\n",
+        "        features=tfds.features.FeaturesDict({\n",
+        "            \"image\": tfds.features.Image(shape=IMG_SHAPE),\n",
+        "            \"label\": tfds.features.ClassLabel(num_classes=len(CLASSES)),\n",
+        "        }),\n",
+        "        supervised_keys=(\"image\", \"label\"),\n",
+        "        homepage=\"https://github.com/axelcarlier/projsemisup/\",\n",
+        "    )\n",
+        "\n",
+        "  def _split_generators(self, dl_manager):\n",
+        "    \"\"\"Returns SplitGenerators.\"\"\"\n",
+        "    # download archive\n",
+        "    archive_path = dl_manager.download_and_extract(URL)\n",
+        "    archive_path = pathlib.Path(archive_path)\n",
+        "    archive_path /= f\"projsemisup-{COMMIT}\"\n",
+        "\n",
+        "    # define the splits\n",
+        "    splits = [\n",
+        "        tfds.core.SplitGenerator(\n",
+        "            name=tfds.Split.TRAIN,\n",
+        "            gen_kwargs={\"data_dir\": archive_path / \"Lab\", \"has_labels\": True},\n",
+        "        ),\n",
+        "        tfds.core.SplitGenerator(\n",
+        "            name=tfds.Split.TEST,\n",
+        "            gen_kwargs={\"data_dir\": archive_path / \"Test\", \"has_labels\": True},\n",
+        "        ),\n",
+        "        tfds.core.SplitGenerator(\n",
+        "            name=UNLABELLED,\n",
+        "            gen_kwargs={\"data_dir\": archive_path / \"Unlab\", \"has_labels\": False},\n",
+        "        ),\n",
+        "    ]\n",
+        "\n",
+        "    return splits\n",
+        "\n",
+        "  def _generate_examples(self, data_dir, has_labels):\n",
+        "    \"\"\"Yields examples.\n",
+        "\n",
+        "    Args:\n",
+        "        data_dir (Path): Path to the where images are stored \n",
+        "        has_labels (bool): Whether the dataset has labels or not\n",
+        "    \"\"\"\n",
+        "    # get all images recursively\n",
+        "    rglob_jpg = pathlib.Path(data_dir).rglob(\"*.jpg\")\n",
+        "    rglob_jpeg = pathlib.Path(data_dir).rglob(\"*.jpeg\")\n",
+        "    rglob = itertools.chain(rglob_jpg, rglob_jpeg)\n",
+        "\n",
+        "    for index, path in enumerate(rglob):\n",
+        "      # load the image\n",
+        "      image = tfds.core.lazy_imports.cv2.imread(str(path))\n",
+        "      # resize the image\n",
+        "      image = tfds.core.lazy_imports.cv2.resize(image, IMG_SHAPE[:2])\n",
+        "      # convert to RGB\n",
+        "      image = tfds.core.lazy_imports.cv2.cvtColor(image, tfds.core.lazy_imports.cv2.COLOR_BGR2RGB)\n",
+        "\n",
+        "      # get label\n",
+        "      label = CLASSES.index(path.parent.name) if has_labels else -1\n",
+        "\n",
+        "      # yield the image and the label\n",
+        "      yield index, {\n",
+        "          \"image\": image,\n",
+        "          \"label\": label, \n",
+        "      }\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 58,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 2,
+          "status": "ok",
+          "timestamp": 1672861656096,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "wXuUtSYn_r-s"
+      },
+      "outputs": [],
+      "source": [
+        "# Algorithm hyperparameters\n",
+        "num_epochs = 50\n",
+        "unlabeled_batch_size = 100\n",
+        "labeled_batch_size = 10\n",
+        "batch_size = unlabeled_batch_size + labeled_batch_size\n",
+        "\n",
+        "width = 128 # width of convolution in encoder\n",
+        "temperature = 0.1\n",
+        "\n",
+        "# Stronger augmentations for contrastive, weaker ones for supervised training\n",
+        "contrastive_augmentation = {\"min_area\": 0.25, \"brightness\": 0.6, \"jitter\": 0.2}\n",
+        "classification_augmentation = {\"min_area\": 0.75, \"brightness\": 0.3, \"jitter\": 0.1}\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 63,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 150,
+          "referenced_widgets": [
+            "f4d4b13251004dbfb7ba81f36289456b",
+            "92d5ca69eb10464cadd2ad8c6b92b5f1",
+            "cf39226fb80c417c990d43244daff8b4",
+            "fc17720d2e434619bf5396de7467590c",
+            "b073c9c0a0b54a12995ecefe0136e950",
+            "bf108424901946eb8b189fc726300472",
+            "0ccd60e561e94ceebd7fd9fb209f669b",
+            "a3a0c812bbc34c3c942d5372cc078b18",
+            "ed3859fe5b6f4c93855f0ef500d8bc91",
+            "b7f82bacd58940439544261b5e942048",
+            "b982bb2c7d784572aee1029aa217b969",
+            "f203872de5f94508a338390508651d0c",
+            "de6a8641f28d48d0b773ac5ad800eec6",
+            "74bfd61016d24ea7b677354077f04379",
+            "0bf4eaad630549a6b489172a93b8f2ee",
+            "4f13be67e1c343c6adec61bc58ff90f1",
+            "c3b6e4177f2e4b029b1666cf4c1a62a6",
+            "d18b6dd9bcd145ab9b4dcbd1bf217a39",
+            "6fc14bc422744eb29e13a1a0f31952e2",
+            "6d19543181544ddbac41e1357d448d34",
+            "db025fec91104f209b62774850ad33bb",
+            "eb50e6dc7ec945a9940e6f69e608a4fd",
+            "c1957bebe6f64eb0915d6e7a26157775",
+            "4fb533e568294997a4a4f359416c1b5b",
+            "3fdb7263acbe430ba27d42f46fbffdc0",
+            "88fdfea6ec91465ba7cda5e9547599d6",
+            "1bf1a0aa3dde41ceb1267c0fc9754986",
+            "a77dac4501f147f2bad2378c4300dafd",
+            "78f1b7c6540d4687ade023641b47c5a4",
+            "22f5687d1c514a26b1cff927b3243955",
+            "3c1c6edfd6994df38afb1ab8d21a757a",
+            "b17b9124cc884e2ea22011cd5e5b3636",
+            "bb1ca1dec70249f182852c32a06bb4c4",
+            "b7e79680e03b46e1b33510af665107fe",
+            "208d7f6179e347fe943a2ef9fb56d9a9",
+            "9901ea583eb54f5494dabb99cfc99483",
+            "1f58412868b448808175d63bcfb447da",
+            "49f5c7f0b5cc4e9999c6d905fdd74c1e",
+            "c80e7fc56adb400e907cd62b150b978d",
+            "ad2cd0ee331141fd8252cd600ccc5fff",
+            "71d8a3f083ef44bfbfa64f2bbb10c1f0",
+            "6b1184b6610a4660a8d6e3d8efa59bc8",
+            "0b9a6a3607be486cb44deab621476db3",
+            "625edd164dcf4932a36dec0851a79c89",
+            "3273f7f1d44b4741af5873ec8e411381",
+            "973841b307dd46b59175ca8545169581",
+            "51b36f07ba624efa93a48ee34cf906c5",
+            "d014273e04c7430d99b746883d71a85b",
+            "ae66b99545a84856af3793aa59273def",
+            "3b3435d568f344e5a284f663743f0690",
+            "5d4921be4b2244db90796202da5faf04",
+            "94ff6626e3b84758a2ffb12a9973ded8",
+            "edde9ce2a45344d6babb797dce13d486",
+            "47ae3e436a3548618f07a1a7a07e168e",
+            "182e7f0633264ae0aa5c97ea14f6b7df",
+            "e9eb28d864f34cf8bba737909172eb2e",
+            "37beb756d55345e39d41c42bd5f4dc16",
+            "80bf30467e6942bbb164bf90d6aca854",
+            "674c7bed061442718b1520380e88d522",
+            "022331d5cd3e43c58fac533918517d47",
+            "f3cbe23c356d43679cc4ff486fca3545",
+            "bf7519c8ab95488fa5f79cb35c8db1e1",
+            "282fb71f47174efc997f8d259b785d61",
+            "a8796bad61a34df89c1154a40bd814e9",
+            "266c3c3c39d147b2b4c95af49fb36f42",
+            "344ab71ee68b4af28cb2948cc3e2ef15",
+            "63af397c06c242b2a6d9d4acd0232cb3",
+            "e86e376d77c24506a3541e0ab9da1c32",
+            "f8774da5294944598574dfd2a8a35610",
+            "a6b8c7ab593148768205ad47b7fea634",
+            "b00733d89e474697873f02f7e60fa7d4",
+            "8bd20837baf947099b4fc9d6cb3e871e",
+            "34415bcdaaa54e089630854f1df6114c",
+            "b02f0e154835453fa47b161ec4cefcb1",
+            "811af7ec09af455785951ea0137f2e3a",
+            "e0c480f5b7254651a5af7e04410d9d43",
+            "69c4ff515c4e4b1f9f7c313374b3aa34",
+            "7ed215974c6b4311bb34d4ba721f790b",
+            "a274c06ad80347ec9a5077cb6ce5eba1",
+            "73d89e4492644d1a9e412b72d72a14de",
+            "f499165e66704145bfb38bf241613b87",
+            "40b74195618544d5a2c18e9307d8997f",
+            "16a63953514a422f9a7f8d385f115308",
+            "96b5945b44ff461387eed16bd3306a20",
+            "a299ad4d23e74945824a9deecedfc76f",
+            "eababa6c94554cd484df6dd55dc994db",
+            "ffd0376578b64811b35afde6c0cb63ab",
+            "72dfaac67f24431689781951c59d7feb",
+            "b5c568b161f24e418d38ecc1821267ac",
+            "2e58e17884b14d4e935bcf11fb01acf6",
+            "65053c140cfd402eb64fdbc1cc32cfc3",
+            "6bdb69e0423b4df68ae9408a01ab02cc",
+            "2c99fb33ca8a43f5a7efc29a975c2bd6",
+            "a1b346004e3c48a890b99bbfff63b83f",
+            "5f1c14872085484fa1e40e9f75bb1f40",
+            "69426376e67948f492cd8deded7efe2e",
+            "f71989288503489b847e8f629658798b",
+            "42f427596cf948268ec0ae6a784229d1",
+            "59f1ab66e02c4ae5a6a840a1da0d1e8e",
+            "f2596bfef90f4a4095713b88b1ac16c1",
+            "0bbc472ae27441dfb3ff3e9b58d56e71",
+            "7f1567f89ac1487eb70960c65b6816ec",
+            "dd5eb0967fb6431697bd5901e1bb4fba",
+            "45bccbfc63c14b308bbb6ad4c384ce74",
+            "d83d0f7418d04166975f056bb4c71f43",
+            "23ea228dcb214d6185701e9c7a33f9d7",
+            "5fc847f952624fba849142a2d01b4500",
+            "1c39c0bac3664b8c9eb97ba37d29bbf2",
+            "9f4b3d82a9bf41d49154d125ea0d548a",
+            "30093792a4584cea92bd8fe8e2c41687"
+          ]
+        },
+        "executionInfo": {
+          "elapsed": 173449,
+          "status": "ok",
+          "timestamp": 1672862101039,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "kzB0Ikfj_mZy",
+        "outputId": "773a6369-d470-4c2e-9e32-f9df11f396a7"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Downloading and preparing dataset Unknown size (download: Unknown size, generated: Unknown size, total: Unknown size) to ~/tensorflow_datasets/projsemisup/1.1.0...\n"
+          ]
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "f4d4b13251004dbfb7ba81f36289456b",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Dl Completed...: 0 url [00:00, ? url/s]"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "f203872de5f94508a338390508651d0c",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Dl Size...: 0 MiB [00:00, ? MiB/s]"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "c1957bebe6f64eb0915d6e7a26157775",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Extraction completed...: 0 file [00:00, ? file/s]"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "b7e79680e03b46e1b33510af665107fe",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Generating splits...:   0%|          | 0/3 [00:00<?, ? splits/s]"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "3273f7f1d44b4741af5873ec8e411381",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Generating train examples...: 0 examples [00:00, ? examples/s]"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "e9eb28d864f34cf8bba737909172eb2e",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Shuffling ~/tensorflow_datasets/projsemisup/1.1.0.incompleteUMKSYI/projsemisup-train.tfrecord*...:   0%|      …"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "63af397c06c242b2a6d9d4acd0232cb3",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Generating test examples...: 0 examples [00:00, ? examples/s]"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "7ed215974c6b4311bb34d4ba721f790b",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Shuffling ~/tensorflow_datasets/projsemisup/1.1.0.incompleteUMKSYI/projsemisup-test.tfrecord*...:   0%|       …"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "b5c568b161f24e418d38ecc1821267ac",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Generating unlabelled examples...: 0 examples [00:00, ? examples/s]"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "data": {
+            "application/vnd.jupyter.widget-view+json": {
+              "model_id": "f2596bfef90f4a4095713b88b1ac16c1",
+              "version_major": 2,
+              "version_minor": 0
+            },
+            "text/plain": [
+              "Shuffling ~/tensorflow_datasets/projsemisup/1.1.0.incompleteUMKSYI/projsemisup-unlabelled.tfrecord*...:   0%| …"
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        },
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Dataset projsemisup downloaded and prepared to ~/tensorflow_datasets/projsemisup/1.1.0. Subsequent calls will reuse this data.\n"
+          ]
+        }
+      ],
+      "source": [
+        "def prepare_dataset():\n",
+        "    unlabeled_train_dataset = (\n",
+        "        tfds.load(\"Projsemisup\", split=\"unlabelled\", as_supervised=True, shuffle_files=True)\n",
+        "        .shuffle(buffer_size=10 * unlabeled_batch_size)\n",
+        "        .batch(unlabeled_batch_size)\n",
+        "    )\n",
+        "    labeled_train_dataset = (\n",
+        "        tfds.load(\"Projsemisup\", split=\"train\", as_supervised=True, shuffle_files=True)\n",
+        "        .shuffle(buffer_size=10 * labeled_batch_size)\n",
+        "        .batch(labeled_batch_size)\n",
+        "    )\n",
+        "    test_dataset = (\n",
+        "        tfds.load(\"Projsemisup\", split=\"test\", as_supervised=True)\n",
+        "        .batch(batch_size)\n",
+        "        .prefetch(buffer_size=tf.data.AUTOTUNE)\n",
+        "    )\n",
+        "\n",
+        "    # Labeled and unlabeled datasets are zipped together\n",
+        "    train_dataset = tf.data.Dataset.zip(\n",
+        "        (unlabeled_train_dataset, labeled_train_dataset)\n",
+        "    ).prefetch(buffer_size=tf.data.AUTOTUNE)\n",
+        "\n",
+        "    return train_dataset, labeled_train_dataset, test_dataset\n",
+        "\n",
+        "# Load dataset\n",
+        "train_dataset, labeled_train_dataset, test_dataset = prepare_dataset()\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 7,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 24,
+          "status": "ok",
+          "timestamp": 1672849536052,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "xYCFy2QSUqut"
+      },
+      "outputs": [],
+      "source": [
+        "# Distorts the color distibutions of images\n",
+        "class RandomColorAffine(layers.Layer):\n",
+        "    def __init__(self, brightness=0, jitter=0, **kwargs):\n",
+        "        super().__init__(**kwargs)\n",
+        "\n",
+        "        self.brightness = brightness\n",
+        "        self.jitter = jitter\n",
+        "\n",
+        "    def get_config(self):\n",
+        "        config = super().get_config()\n",
+        "        config.update({\"brightness\": self.brightness, \"jitter\": self.jitter})\n",
+        "        return config\n",
+        "\n",
+        "    def call(self, images, training=True):\n",
+        "        if training:\n",
+        "            batch_size = tf.shape(images)[0]\n",
+        "\n",
+        "            # Same for all colors\n",
+        "            brightness_scales = 1 + tf.random.uniform(\n",
+        "                (batch_size, 1, 1, 1), minval=-self.brightness, maxval=self.brightness\n",
+        "            )\n",
+        "            # Different for all colors\n",
+        "            jitter_matrices = tf.random.uniform(\n",
+        "                (batch_size, 1, 3, 3), minval=-self.jitter, maxval=self.jitter\n",
+        "            )\n",
+        "\n",
+        "            color_transforms = (\n",
+        "                tf.eye(3, batch_shape=[batch_size, 1]) * brightness_scales\n",
+        "                + jitter_matrices\n",
+        "            )\n",
+        "            images = tf.clip_by_value(tf.matmul(images, color_transforms), 0, 1)\n",
+        "        return images\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 12,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 313,
+          "status": "ok",
+          "timestamp": 1672849737252,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "iTC1pcyVXdPe"
+      },
+      "outputs": [],
+      "source": [
+        "# Image augmentation module\n",
+        "def get_augmenter(min_area, brightness, jitter):\n",
+        "    zoom_factor = 1.0 - math.sqrt(min_area)\n",
+        "    return keras.Sequential(\n",
+        "        [\n",
+        "            keras.Input(shape=IMG_SHAPE),\n",
+        "            layers.Rescaling(1 / 255),\n",
+        "            layers.RandomFlip(\"horizontal\"),\n",
+        "            layers.RandomTranslation(zoom_factor / 2, zoom_factor / 2),\n",
+        "            layers.RandomZoom((-zoom_factor, 0.0), (-zoom_factor, 0.0)),\n",
+        "            RandomColorAffine(brightness, jitter),\n",
+        "            layers.Rescaling(2, offset=-1),\n",
+        "        ]\n",
+        "    )\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 13,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 2,
+          "status": "ok",
+          "timestamp": 1672849739077,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "WCJg9_PJYGgK"
+      },
+      "outputs": [],
+      "source": [
+        "def visualize_augmentations(num_images):\n",
+        "    # Sample a batch from a dataset\n",
+        "    images = next(iter(train_dataset))[0][0][:num_images]\n",
+        "    # Apply augmentations\n",
+        "    augmented_images = zip(\n",
+        "        images,\n",
+        "        get_augmenter(**classification_augmentation)(images),\n",
+        "        get_augmenter(**contrastive_augmentation)(images),\n",
+        "        get_augmenter(**contrastive_augmentation)(images),\n",
+        "    )\n",
+        "    row_titles = [\n",
+        "        \"Original:\",\n",
+        "        \"Weakly augmented:\",\n",
+        "        \"Strongly augmented:\",\n",
+        "        \"Strongly augmented:\",\n",
+        "    ]\n",
+        "    plt.figure(figsize=(num_images * 2.2, 4 * 2.2), dpi=100)\n",
+        "    for column, image_row in enumerate(augmented_images):\n",
+        "        for row, image in enumerate(image_row):\n",
+        "\n",
+        "            if row != 0:\n",
+        "              image = image / 2 + 0.5\n",
+        "\n",
+        "            plt.subplot(4, num_images, row * num_images + column + 1)\n",
+        "            plt.imshow(image)\n",
+        "            if column == 0:\n",
+        "                plt.title(row_titles[row], loc=\"left\")\n",
+        "            plt.axis(\"off\")\n",
+        "    plt.tight_layout()\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 14,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 865,
+          "output_embedded_package_id": "1QkcJsF3lxaTFp4R2uZkESALFuspdd7G3"
+        },
+        "executionInfo": {
+          "elapsed": 9294,
+          "status": "ok",
+          "timestamp": 1672849752622,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "kVdOLvwPBz0u",
+        "outputId": "3c5a2929-04a8-4cc8-9b05-4ac4bca1d179"
+      },
+      "outputs": [
+        {
+          "data": {
+            "text/plain": [
+              "Output hidden; open in https://colab.research.google.com to view."
+            ]
+          },
+          "metadata": {},
+          "output_type": "display_data"
+        }
+      ],
+      "source": [
+        "visualize_augmentations(num_images=8)\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 45,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 2,
+          "status": "ok",
+          "timestamp": 1672852743092,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "PdJ82_vNaegq"
+      },
+      "outputs": [],
+      "source": [
+        "# Define the encoder architecture\n",
+        "def get_encoder():\n",
+        "    return keras.Sequential(\n",
+        "        [\n",
+        "            keras.Input(shape=IMG_SHAPE),\n",
+        "            tf.keras.applications.mobilenet.MobileNet(input_shape=IMG_SHAPE, classes=len(CLASSES), include_top=False, weights=None),\n",
+        "            layers.Flatten(),\n",
+        "            layers.Dense(width, activation=\"relu\"),\n",
+        "        ],\n",
+        "        name=\"encoder\",\n",
+        "    )\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 41,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "executionInfo": {
+          "elapsed": 1734,
+          "status": "ok",
+          "timestamp": 1672852229074,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "3RMBciDNaixE",
+        "outputId": "04c80ca3-6f72-4dd0-b14c-ef2d5de9cdf7"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Model: \"baseline_model\"\n",
+            "_________________________________________________________________\n",
+            " Layer (type)                Output Shape              Param #   \n",
+            "=================================================================\n",
+            " sequential_22 (Sequential)  (None, 128, 128, 3)       0         \n",
+            "                                                                 \n",
+            " mobilenet_1.00_128 (Functio  (None, 4, 4, 1024)       3228864   \n",
+            " nal)                                                            \n",
+            "                                                                 \n",
+            " flatten_15 (Flatten)        (None, 16384)             0         \n",
+            "                                                                 \n",
+            " dense_15 (Dense)            (None, 18)                294930    \n",
+            "                                                                 \n",
+            "=================================================================\n",
+            "Total params: 3,523,794\n",
+            "Trainable params: 3,501,906\n",
+            "Non-trainable params: 21,888\n",
+            "_________________________________________________________________\n"
+          ]
+        }
+      ],
+      "source": [
+        "# Baseline supervised training with random initialization\n",
+        "baseline_model = keras.Sequential(\n",
+        "    [\n",
+        "        keras.Input(shape=IMG_SHAPE),\n",
+        "        get_augmenter(**classification_augmentation),\n",
+        "        tf.keras.applications.mobilenet.MobileNet(input_shape=IMG_SHAPE, classes=len(CLASSES), include_top=False, weights=None),\n",
+        "        layers.Flatten(),\n",
+        "        layers.Dense(len(CLASSES)),\n",
+        "    ],\n",
+        "    name=\"baseline_model\",\n",
+        ")\n",
+        "\n",
+        "baseline_model.compile(\n",
+        "    optimizer=keras.optimizers.Adam(),\n",
+        "    loss=keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
+        "    metrics=[\n",
+        "        keras.metrics.SparseCategoricalAccuracy(name=\"top1\"),\n",
+        "        tf.keras.metrics.SparseTopKCategoricalAccuracy(k=3, name=\"top3\"),\n",
+        "        tf.keras.metrics.SparseTopKCategoricalAccuracy(k=5, name=\"top5\")\n",
+        "    ],\n",
+        ")\n",
+        "\n",
+        "baseline_model.summary()\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 42,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "executionInfo": {
+          "elapsed": 419341,
+          "status": "ok",
+          "timestamp": 1672852648760,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "I_JJVq9xCR1B",
+        "outputId": "5842868a-4adc-441d-dbfa-f8cb56aba7b2"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Epoch 1/100\n",
+            "12/12 [==============================] - 8s 299ms/step - loss: 6.0856 - top1: 0.0667 - top3: 0.1750 - top5: 0.2972 - val_loss: 2.8917 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 2/100\n",
+            "12/12 [==============================] - 3s 259ms/step - loss: 5.0710 - top1: 0.0750 - top3: 0.1778 - top5: 0.2889 - val_loss: 2.8951 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 3/100\n",
+            "12/12 [==============================] - 3s 258ms/step - loss: 4.6516 - top1: 0.0722 - top3: 0.2083 - top5: 0.2889 - val_loss: 2.9139 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 4/100\n",
+            "12/12 [==============================] - 3s 274ms/step - loss: 4.2882 - top1: 0.0611 - top3: 0.2028 - top5: 0.3472 - val_loss: 2.9319 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 5/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 4.2393 - top1: 0.0500 - top3: 0.1750 - top5: 0.2861 - val_loss: 2.9575 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 6/100\n",
+            "12/12 [==============================] - 3s 274ms/step - loss: 4.0733 - top1: 0.0694 - top3: 0.1806 - top5: 0.2833 - val_loss: 2.9812 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 7/100\n",
+            "12/12 [==============================] - 3s 255ms/step - loss: 3.6540 - top1: 0.0806 - top3: 0.2111 - top5: 0.3389 - val_loss: 2.9831 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 8/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 3.7301 - top1: 0.0722 - top3: 0.2167 - top5: 0.3278 - val_loss: 2.9854 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 9/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 3.8372 - top1: 0.0944 - top3: 0.2444 - top5: 0.3556 - val_loss: 3.0430 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 10/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 4.1383 - top1: 0.1056 - top3: 0.2167 - top5: 0.3194 - val_loss: 3.0421 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 11/100\n",
+            "12/12 [==============================] - 3s 259ms/step - loss: 3.8502 - top1: 0.1056 - top3: 0.2583 - top5: 0.3861 - val_loss: 3.0284 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 12/100\n",
+            "12/12 [==============================] - 3s 264ms/step - loss: 3.6339 - top1: 0.0806 - top3: 0.2583 - top5: 0.3972 - val_loss: 3.0184 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 13/100\n",
+            "12/12 [==============================] - 3s 287ms/step - loss: 3.6809 - top1: 0.1028 - top3: 0.2500 - top5: 0.3806 - val_loss: 3.1172 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 14/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 3.6238 - top1: 0.0917 - top3: 0.2500 - top5: 0.3861 - val_loss: 3.4350 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 15/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 3.7031 - top1: 0.1250 - top3: 0.2861 - top5: 0.3833 - val_loss: 3.1098 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 16/100\n",
+            "12/12 [==============================] - 3s 264ms/step - loss: 3.1602 - top1: 0.1111 - top3: 0.3111 - top5: 0.4639 - val_loss: 3.1246 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 17/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 3.0242 - top1: 0.1389 - top3: 0.3444 - top5: 0.4778 - val_loss: 3.4144 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 18/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 2.9790 - top1: 0.1444 - top3: 0.3222 - top5: 0.4639 - val_loss: 3.3467 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 19/100\n",
+            "12/12 [==============================] - 3s 283ms/step - loss: 3.0180 - top1: 0.1667 - top3: 0.3194 - top5: 0.4750 - val_loss: 3.2752 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 20/100\n",
+            "12/12 [==============================] - 3s 268ms/step - loss: 2.8265 - top1: 0.1500 - top3: 0.3778 - top5: 0.5444 - val_loss: 3.3366 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 21/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 2.9435 - top1: 0.1639 - top3: 0.3944 - top5: 0.5417 - val_loss: 3.2390 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 22/100\n",
+            "12/12 [==============================] - 3s 287ms/step - loss: 2.9606 - top1: 0.1500 - top3: 0.3472 - top5: 0.5000 - val_loss: 3.1749 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 23/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 2.7517 - top1: 0.1444 - top3: 0.3722 - top5: 0.5611 - val_loss: 3.2657 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 24/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 2.7840 - top1: 0.2000 - top3: 0.4417 - top5: 0.5778 - val_loss: 3.2229 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 25/100\n",
+            "12/12 [==============================] - 3s 264ms/step - loss: 2.6860 - top1: 0.2111 - top3: 0.4333 - top5: 0.5806 - val_loss: 3.1691 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 26/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 2.6168 - top1: 0.2222 - top3: 0.4444 - top5: 0.6250 - val_loss: 3.2330 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 27/100\n",
+            "12/12 [==============================] - 3s 262ms/step - loss: 2.8648 - top1: 0.1778 - top3: 0.4361 - top5: 0.6056 - val_loss: 3.2809 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 28/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 2.6126 - top1: 0.2278 - top3: 0.4778 - top5: 0.6472 - val_loss: 3.2464 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 29/100\n",
+            "12/12 [==============================] - 3s 262ms/step - loss: 2.4680 - top1: 0.2111 - top3: 0.5000 - top5: 0.6556 - val_loss: 3.2316 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 30/100\n",
+            "12/12 [==============================] - 3s 283ms/step - loss: 2.5806 - top1: 0.2333 - top3: 0.4917 - top5: 0.6528 - val_loss: 3.3003 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 31/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 2.5545 - top1: 0.2194 - top3: 0.4722 - top5: 0.6278 - val_loss: 3.3889 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 32/100\n",
+            "12/12 [==============================] - 3s 276ms/step - loss: 2.5008 - top1: 0.2361 - top3: 0.5167 - top5: 0.6500 - val_loss: 3.2262 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 33/100\n",
+            "12/12 [==============================] - 3s 284ms/step - loss: 2.5381 - top1: 0.2556 - top3: 0.4917 - top5: 0.6444 - val_loss: 3.3344 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 34/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 2.2850 - top1: 0.2611 - top3: 0.5722 - top5: 0.7278 - val_loss: 3.3519 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 35/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 2.3399 - top1: 0.2694 - top3: 0.5583 - top5: 0.7056 - val_loss: 3.3309 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 36/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 2.3586 - top1: 0.2694 - top3: 0.5389 - top5: 0.7111 - val_loss: 3.5713 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 37/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 2.2336 - top1: 0.2722 - top3: 0.5583 - top5: 0.6972 - val_loss: 3.3591 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 38/100\n",
+            "12/12 [==============================] - 3s 277ms/step - loss: 2.2746 - top1: 0.2694 - top3: 0.5639 - top5: 0.7000 - val_loss: 3.3947 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 39/100\n",
+            "12/12 [==============================] - 3s 266ms/step - loss: 2.2941 - top1: 0.3139 - top3: 0.5778 - top5: 0.7111 - val_loss: 3.4380 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 40/100\n",
+            "12/12 [==============================] - 3s 277ms/step - loss: 2.1254 - top1: 0.3194 - top3: 0.6056 - top5: 0.7389 - val_loss: 3.4597 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 41/100\n",
+            "12/12 [==============================] - 3s 274ms/step - loss: 2.2076 - top1: 0.3000 - top3: 0.6194 - top5: 0.7472 - val_loss: 3.4901 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 42/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 2.1909 - top1: 0.3250 - top3: 0.5694 - top5: 0.7417 - val_loss: 3.3669 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2756\n",
+            "Epoch 43/100\n",
+            "12/12 [==============================] - 3s 276ms/step - loss: 2.0359 - top1: 0.3806 - top3: 0.6417 - top5: 0.7639 - val_loss: 3.5426 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 44/100\n",
+            "12/12 [==============================] - 3s 276ms/step - loss: 1.8790 - top1: 0.4194 - top3: 0.6806 - top5: 0.8194 - val_loss: 3.6248 - val_top1: 0.0556 - val_top3: 0.1672 - val_top5: 0.2906\n",
+            "Epoch 45/100\n",
+            "12/12 [==============================] - 3s 275ms/step - loss: 1.9642 - top1: 0.3861 - top3: 0.6694 - top5: 0.7833 - val_loss: 3.9569 - val_top1: 0.0583 - val_top3: 0.2000 - val_top5: 0.2778\n",
+            "Epoch 46/100\n",
+            "12/12 [==============================] - 3s 271ms/step - loss: 1.9059 - top1: 0.3694 - top3: 0.6611 - top5: 0.7917 - val_loss: 3.5539 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2944\n",
+            "Epoch 47/100\n",
+            "12/12 [==============================] - 3s 256ms/step - loss: 1.9633 - top1: 0.3694 - top3: 0.6472 - top5: 0.8056 - val_loss: 4.0311 - val_top1: 0.0561 - val_top3: 0.1661 - val_top5: 0.2839\n",
+            "Epoch 48/100\n",
+            "12/12 [==============================] - 3s 266ms/step - loss: 2.0153 - top1: 0.3889 - top3: 0.6722 - top5: 0.8083 - val_loss: 3.8323 - val_top1: 0.0567 - val_top3: 0.1800 - val_top5: 0.2794\n",
+            "Epoch 49/100\n",
+            "12/12 [==============================] - 3s 266ms/step - loss: 1.8955 - top1: 0.4028 - top3: 0.6611 - top5: 0.8000 - val_loss: 4.4012 - val_top1: 0.0567 - val_top3: 0.1728 - val_top5: 0.3011\n",
+            "Epoch 50/100\n",
+            "12/12 [==============================] - 3s 280ms/step - loss: 1.8424 - top1: 0.4083 - top3: 0.6861 - top5: 0.8111 - val_loss: 3.7789 - val_top1: 0.0667 - val_top3: 0.1728 - val_top5: 0.3061\n",
+            "Epoch 51/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 1.6980 - top1: 0.4472 - top3: 0.7528 - top5: 0.8611 - val_loss: 4.2862 - val_top1: 0.0639 - val_top3: 0.1789 - val_top5: 0.3011\n",
+            "Epoch 52/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 1.6796 - top1: 0.4750 - top3: 0.7333 - top5: 0.8694 - val_loss: 3.4254 - val_top1: 0.0867 - val_top3: 0.2300 - val_top5: 0.3550\n",
+            "Epoch 53/100\n",
+            "12/12 [==============================] - 3s 283ms/step - loss: 1.7892 - top1: 0.4111 - top3: 0.7306 - top5: 0.8639 - val_loss: 3.4443 - val_top1: 0.0994 - val_top3: 0.2683 - val_top5: 0.3967\n",
+            "Epoch 54/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 1.7414 - top1: 0.4611 - top3: 0.7167 - top5: 0.8444 - val_loss: 3.0180 - val_top1: 0.1350 - val_top3: 0.3344 - val_top5: 0.4983\n",
+            "Epoch 55/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 1.6590 - top1: 0.4667 - top3: 0.7611 - top5: 0.8750 - val_loss: 3.0000 - val_top1: 0.1589 - val_top3: 0.3594 - val_top5: 0.5133\n",
+            "Epoch 56/100\n",
+            "12/12 [==============================] - 3s 290ms/step - loss: 1.7276 - top1: 0.4917 - top3: 0.7722 - top5: 0.8444 - val_loss: 3.2188 - val_top1: 0.1489 - val_top3: 0.3756 - val_top5: 0.5117\n",
+            "Epoch 57/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 1.5034 - top1: 0.5417 - top3: 0.7889 - top5: 0.8750 - val_loss: 3.1225 - val_top1: 0.1667 - val_top3: 0.3767 - val_top5: 0.5356\n",
+            "Epoch 58/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 1.3995 - top1: 0.5278 - top3: 0.8278 - top5: 0.9194 - val_loss: 3.3075 - val_top1: 0.1633 - val_top3: 0.3783 - val_top5: 0.5378\n",
+            "Epoch 59/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 1.4013 - top1: 0.5667 - top3: 0.8250 - top5: 0.9056 - val_loss: 3.3536 - val_top1: 0.1550 - val_top3: 0.3867 - val_top5: 0.5317\n",
+            "Epoch 60/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 1.5395 - top1: 0.4972 - top3: 0.7889 - top5: 0.8917 - val_loss: 3.4875 - val_top1: 0.1483 - val_top3: 0.3667 - val_top5: 0.5050\n",
+            "Epoch 61/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 1.4433 - top1: 0.5389 - top3: 0.8222 - top5: 0.9083 - val_loss: 3.4089 - val_top1: 0.1628 - val_top3: 0.3739 - val_top5: 0.5356\n",
+            "Epoch 62/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 1.3400 - top1: 0.6028 - top3: 0.8222 - top5: 0.9000 - val_loss: 3.9708 - val_top1: 0.1500 - val_top3: 0.3511 - val_top5: 0.5061\n",
+            "Epoch 63/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 1.4233 - top1: 0.5528 - top3: 0.8028 - top5: 0.8944 - val_loss: 3.5423 - val_top1: 0.1533 - val_top3: 0.3761 - val_top5: 0.5406\n",
+            "Epoch 64/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 1.1232 - top1: 0.6417 - top3: 0.8667 - top5: 0.9444 - val_loss: 3.9507 - val_top1: 0.1600 - val_top3: 0.3767 - val_top5: 0.5350\n",
+            "Epoch 65/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 1.3086 - top1: 0.5917 - top3: 0.8306 - top5: 0.9167 - val_loss: 4.4946 - val_top1: 0.1439 - val_top3: 0.3500 - val_top5: 0.5133\n",
+            "Epoch 66/100\n",
+            "12/12 [==============================] - 3s 289ms/step - loss: 1.1497 - top1: 0.6167 - top3: 0.8667 - top5: 0.9472 - val_loss: 5.0533 - val_top1: 0.1278 - val_top3: 0.3306 - val_top5: 0.4817\n",
+            "Epoch 67/100\n",
+            "12/12 [==============================] - 3s 279ms/step - loss: 1.1579 - top1: 0.6139 - top3: 0.8833 - top5: 0.9528 - val_loss: 5.2465 - val_top1: 0.1206 - val_top3: 0.3172 - val_top5: 0.4567\n",
+            "Epoch 68/100\n",
+            "12/12 [==============================] - 3s 287ms/step - loss: 1.1025 - top1: 0.6361 - top3: 0.8861 - top5: 0.9472 - val_loss: 5.4059 - val_top1: 0.1439 - val_top3: 0.3417 - val_top5: 0.4822\n",
+            "Epoch 69/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 1.1948 - top1: 0.6222 - top3: 0.8778 - top5: 0.9333 - val_loss: 5.7314 - val_top1: 0.1239 - val_top3: 0.3083 - val_top5: 0.4506\n",
+            "Epoch 70/100\n",
+            "12/12 [==============================] - 3s 292ms/step - loss: 1.0875 - top1: 0.6583 - top3: 0.8778 - top5: 0.9417 - val_loss: 6.4302 - val_top1: 0.1278 - val_top3: 0.3161 - val_top5: 0.4767\n",
+            "Epoch 71/100\n",
+            "12/12 [==============================] - 3s 291ms/step - loss: 1.0140 - top1: 0.6694 - top3: 0.8972 - top5: 0.9611 - val_loss: 5.3594 - val_top1: 0.1361 - val_top3: 0.3444 - val_top5: 0.4861\n",
+            "Epoch 72/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 0.9031 - top1: 0.7167 - top3: 0.9111 - top5: 0.9583 - val_loss: 5.7482 - val_top1: 0.1139 - val_top3: 0.3006 - val_top5: 0.4533\n",
+            "Epoch 73/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 0.7967 - top1: 0.7389 - top3: 0.9250 - top5: 0.9806 - val_loss: 5.8548 - val_top1: 0.1256 - val_top3: 0.3172 - val_top5: 0.4578\n",
+            "Epoch 74/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 0.9731 - top1: 0.6722 - top3: 0.9194 - top5: 0.9722 - val_loss: 7.7033 - val_top1: 0.1017 - val_top3: 0.3000 - val_top5: 0.4450\n",
+            "Epoch 75/100\n",
+            "12/12 [==============================] - 3s 262ms/step - loss: 0.8453 - top1: 0.7389 - top3: 0.9194 - top5: 0.9833 - val_loss: 6.3192 - val_top1: 0.1056 - val_top3: 0.3233 - val_top5: 0.4689\n",
+            "Epoch 76/100\n",
+            "12/12 [==============================] - 3s 288ms/step - loss: 0.9650 - top1: 0.7111 - top3: 0.9333 - top5: 0.9722 - val_loss: 8.0213 - val_top1: 0.0844 - val_top3: 0.2594 - val_top5: 0.4106\n",
+            "Epoch 77/100\n",
+            "12/12 [==============================] - 3s 259ms/step - loss: 0.9453 - top1: 0.6889 - top3: 0.9000 - top5: 0.9500 - val_loss: 8.6821 - val_top1: 0.0739 - val_top3: 0.2456 - val_top5: 0.4083\n",
+            "Epoch 78/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 0.8670 - top1: 0.7444 - top3: 0.9056 - top5: 0.9639 - val_loss: 10.4329 - val_top1: 0.0767 - val_top3: 0.2611 - val_top5: 0.4106\n",
+            "Epoch 79/100\n",
+            "12/12 [==============================] - 3s 271ms/step - loss: 0.9323 - top1: 0.7194 - top3: 0.9139 - top5: 0.9639 - val_loss: 7.3033 - val_top1: 0.0806 - val_top3: 0.2478 - val_top5: 0.3978\n",
+            "Epoch 80/100\n",
+            "12/12 [==============================] - 3s 262ms/step - loss: 0.8617 - top1: 0.7250 - top3: 0.9083 - top5: 0.9528 - val_loss: 8.0876 - val_top1: 0.0922 - val_top3: 0.2622 - val_top5: 0.4061\n",
+            "Epoch 81/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 0.8069 - top1: 0.7500 - top3: 0.9306 - top5: 0.9694 - val_loss: 7.8995 - val_top1: 0.0956 - val_top3: 0.2633 - val_top5: 0.4167\n",
+            "Epoch 82/100\n",
+            "12/12 [==============================] - 3s 264ms/step - loss: 0.7686 - top1: 0.7444 - top3: 0.9278 - top5: 0.9833 - val_loss: 7.6313 - val_top1: 0.0889 - val_top3: 0.2378 - val_top5: 0.3744\n",
+            "Epoch 83/100\n",
+            "12/12 [==============================] - 3s 269ms/step - loss: 0.6171 - top1: 0.8056 - top3: 0.9639 - top5: 0.9889 - val_loss: 9.3126 - val_top1: 0.0806 - val_top3: 0.2578 - val_top5: 0.3972\n",
+            "Epoch 84/100\n",
+            "12/12 [==============================] - 5s 428ms/step - loss: 0.6450 - top1: 0.7972 - top3: 0.9556 - top5: 0.9833 - val_loss: 7.2606 - val_top1: 0.0939 - val_top3: 0.2622 - val_top5: 0.4033\n",
+            "Epoch 85/100\n",
+            "12/12 [==============================] - 3s 277ms/step - loss: 0.7591 - top1: 0.7583 - top3: 0.9306 - top5: 0.9778 - val_loss: 7.9695 - val_top1: 0.0817 - val_top3: 0.2511 - val_top5: 0.4044\n",
+            "Epoch 86/100\n",
+            "12/12 [==============================] - 3s 280ms/step - loss: 0.6382 - top1: 0.8083 - top3: 0.9611 - top5: 0.9833 - val_loss: 9.0153 - val_top1: 0.0928 - val_top3: 0.2994 - val_top5: 0.4361\n",
+            "Epoch 87/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 0.7005 - top1: 0.7806 - top3: 0.9417 - top5: 0.9806 - val_loss: 7.1955 - val_top1: 0.1172 - val_top3: 0.3050 - val_top5: 0.4556\n",
+            "Epoch 88/100\n",
+            "12/12 [==============================] - 3s 257ms/step - loss: 0.6239 - top1: 0.7889 - top3: 0.9583 - top5: 0.9833 - val_loss: 7.9878 - val_top1: 0.1167 - val_top3: 0.3033 - val_top5: 0.4650\n",
+            "Epoch 89/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 0.7165 - top1: 0.7528 - top3: 0.9500 - top5: 0.9889 - val_loss: 7.6096 - val_top1: 0.1233 - val_top3: 0.3378 - val_top5: 0.4722\n",
+            "Epoch 90/100\n",
+            "12/12 [==============================] - 3s 279ms/step - loss: 0.5109 - top1: 0.8444 - top3: 0.9611 - top5: 0.9889 - val_loss: 7.5592 - val_top1: 0.1189 - val_top3: 0.3117 - val_top5: 0.4750\n",
+            "Epoch 91/100\n",
+            "12/12 [==============================] - 3s 266ms/step - loss: 0.6646 - top1: 0.7806 - top3: 0.9500 - top5: 0.9861 - val_loss: 7.7273 - val_top1: 0.1139 - val_top3: 0.3011 - val_top5: 0.4678\n",
+            "Epoch 92/100\n",
+            "12/12 [==============================] - 3s 290ms/step - loss: 0.5290 - top1: 0.8417 - top3: 0.9639 - top5: 0.9972 - val_loss: 6.4798 - val_top1: 0.1144 - val_top3: 0.3350 - val_top5: 0.4811\n",
+            "Epoch 93/100\n",
+            "12/12 [==============================] - 3s 288ms/step - loss: 0.4698 - top1: 0.8444 - top3: 0.9694 - top5: 0.9833 - val_loss: 9.0025 - val_top1: 0.0967 - val_top3: 0.3039 - val_top5: 0.4639\n",
+            "Epoch 94/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 0.4942 - top1: 0.8556 - top3: 0.9694 - top5: 0.9944 - val_loss: 5.8329 - val_top1: 0.1367 - val_top3: 0.3278 - val_top5: 0.4911\n",
+            "Epoch 95/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 0.5575 - top1: 0.8306 - top3: 0.9639 - top5: 0.9889 - val_loss: 8.4929 - val_top1: 0.0911 - val_top3: 0.2606 - val_top5: 0.4161\n",
+            "Epoch 96/100\n",
+            "12/12 [==============================] - 3s 276ms/step - loss: 0.4945 - top1: 0.8306 - top3: 0.9694 - top5: 0.9889 - val_loss: 6.0370 - val_top1: 0.1211 - val_top3: 0.3067 - val_top5: 0.4567\n",
+            "Epoch 97/100\n",
+            "12/12 [==============================] - 3s 266ms/step - loss: 0.4584 - top1: 0.8472 - top3: 0.9778 - top5: 0.9861 - val_loss: 8.9154 - val_top1: 0.0961 - val_top3: 0.2961 - val_top5: 0.4417\n",
+            "Epoch 98/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 0.4219 - top1: 0.8778 - top3: 0.9639 - top5: 0.9806 - val_loss: 8.9500 - val_top1: 0.1122 - val_top3: 0.3139 - val_top5: 0.4828\n",
+            "Epoch 99/100\n",
+            "12/12 [==============================] - 3s 262ms/step - loss: 0.4460 - top1: 0.8389 - top3: 0.9722 - top5: 0.9944 - val_loss: 6.8160 - val_top1: 0.1228 - val_top3: 0.3239 - val_top5: 0.4844\n",
+            "Epoch 100/100\n",
+            "12/12 [==============================] - 3s 257ms/step - loss: 0.3929 - top1: 0.8861 - top3: 0.9722 - top5: 0.9944 - val_loss: 9.7742 - val_top1: 0.0961 - val_top3: 0.2744 - val_top5: 0.4300\n"
+          ]
+        }
+      ],
+      "source": [
+        "baseline_history = baseline_model.fit(\n",
+        "    labeled_train_dataset.unbatch().batch(30), epochs=100, validation_data=test_dataset\n",
+        ")\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 46,
+      "metadata": {
+        "executionInfo": {
+          "elapsed": 2,
+          "status": "ok",
+          "timestamp": 1672852750128,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "XzfBlyktdVLe"
+      },
+      "outputs": [],
+      "source": [
+        "# Define the contrastive model with model-subclassing\n",
+        "class ContrastiveModel(keras.Model):\n",
+        "    def __init__(self):\n",
+        "        super().__init__()\n",
+        "\n",
+        "        self.temperature = temperature\n",
+        "        self.contrastive_augmenter = get_augmenter(**contrastive_augmentation)\n",
+        "        self.classification_augmenter = get_augmenter(**classification_augmentation)\n",
+        "        self.encoder = get_encoder()\n",
+        "        \n",
+        "        # Non-linear MLP as projection head\n",
+        "        self.projection_head = keras.Sequential(\n",
+        "            [\n",
+        "                keras.Input(shape=(width,)),\n",
+        "                layers.Dense(width, activation=\"relu\"),\n",
+        "                layers.Dense(width),\n",
+        "            ],\n",
+        "            name=\"projection_head\",\n",
+        "        )\n",
+        "        \n",
+        "        # Single dense layer for linear probing\n",
+        "        self.linear_probe = keras.Sequential(\n",
+        "            [layers.Input(shape=(width,)), layers.Dense(len(CLASSES))], name=\"linear_probe\"\n",
+        "        )\n",
+        "\n",
+        "        self.encoder.summary()\n",
+        "        self.projection_head.summary()\n",
+        "        self.linear_probe.summary()\n",
+        "\n",
+        "    def compile(self, contrastive_optimizer, probe_optimizer, **kwargs):\n",
+        "        super().compile(**kwargs)\n",
+        "\n",
+        "        self.contrastive_optimizer = contrastive_optimizer\n",
+        "        self.probe_optimizer = probe_optimizer\n",
+        "\n",
+        "        # self.contrastive_loss will be defined as a method\n",
+        "        self.probe_loss = keras.losses.SparseCategoricalCrossentropy(from_logits=True)\n",
+        "\n",
+        "        self.contrastive_loss_tracker = keras.metrics.Mean(name=\"c_loss\")\n",
+        "        self.contrastive_accuracy = keras.metrics.SparseCategoricalAccuracy(\n",
+        "            name=\"c_acc\"\n",
+        "        )\n",
+        "        self.probe_loss_tracker = keras.metrics.Mean(name=\"p_loss\")\n",
+        "\n",
+        "        self.probe_accuracy = keras.metrics.SparseCategoricalAccuracy(name=\"top1\")\n",
+        "        self.top3_accuracy = tf.keras.metrics.SparseTopKCategoricalAccuracy(k=3, name=\"top3\")\n",
+        "        self.top5_accuracy = tf.keras.metrics.SparseTopKCategoricalAccuracy(k=5, name=\"top5\")\n",
+        "\n",
+        "\n",
+        "    @property\n",
+        "    def metrics(self):\n",
+        "        return [\n",
+        "            self.contrastive_loss_tracker,\n",
+        "            self.contrastive_accuracy,\n",
+        "            self.probe_loss_tracker,\n",
+        "            self.probe_accuracy,\n",
+        "            self.top3_accuracy,\n",
+        "            self.top5_accuracy,\n",
+        "        ]\n",
+        "\n",
+        "    def contrastive_loss(self, projections_1, projections_2):\n",
+        "        # InfoNCE loss (information noise-contrastive estimation)\n",
+        "        # NT-Xent loss (normalized temperature-scaled cross entropy)\n",
+        "\n",
+        "        # Cosine similarity: the dot product of the l2-normalized feature vectors\n",
+        "        projections_1 = tf.math.l2_normalize(projections_1, axis=1)\n",
+        "        projections_2 = tf.math.l2_normalize(projections_2, axis=1)\n",
+        "        similarities = (\n",
+        "            tf.matmul(projections_1, projections_2, transpose_b=True) / self.temperature\n",
+        "        )\n",
+        "\n",
+        "        # The similarity between the representations of two augmented views of the\n",
+        "        # same image should be higher than their similarity with other views\n",
+        "        batch_size = tf.shape(projections_1)[0]\n",
+        "        contrastive_labels = tf.range(batch_size)\n",
+        "        self.contrastive_accuracy.update_state(contrastive_labels, similarities)\n",
+        "        self.contrastive_accuracy.update_state(\n",
+        "            contrastive_labels, tf.transpose(similarities)\n",
+        "        )\n",
+        "\n",
+        "        # The temperature-scaled similarities are used as logits for cross-entropy\n",
+        "        # a symmetrized version of the loss is used here\n",
+        "        loss_1_2 = keras.losses.sparse_categorical_crossentropy(\n",
+        "            contrastive_labels, similarities, from_logits=True\n",
+        "        )\n",
+        "        loss_2_1 = keras.losses.sparse_categorical_crossentropy(\n",
+        "            contrastive_labels, tf.transpose(similarities), from_logits=True\n",
+        "        )\n",
+        "        return (loss_1_2 + loss_2_1) / 2\n",
+        "\n",
+        "    def train_step(self, data):\n",
+        "        (unlabeled_images, _), (labeled_images, labels) = data\n",
+        "\n",
+        "        # Both labeled and unlabeled images are used, without labels\n",
+        "        images = tf.concat((unlabeled_images, labeled_images), axis=0)\n",
+        "        # Each image is augmented twice, differently\n",
+        "        augmented_images_1 = self.contrastive_augmenter(images, training=True)\n",
+        "        augmented_images_2 = self.contrastive_augmenter(images, training=True)\n",
+        "        with tf.GradientTape() as tape:\n",
+        "            features_1 = self.encoder(augmented_images_1, training=True)\n",
+        "            features_2 = self.encoder(augmented_images_2, training=True)\n",
+        "            # The representations are passed through a projection mlp\n",
+        "            projections_1 = self.projection_head(features_1, training=True)\n",
+        "            projections_2 = self.projection_head(features_2, training=True)\n",
+        "            contrastive_loss = self.contrastive_loss(projections_1, projections_2)\n",
+        "        gradients = tape.gradient(\n",
+        "            contrastive_loss,\n",
+        "            self.encoder.trainable_weights + self.projection_head.trainable_weights,\n",
+        "        )\n",
+        "        self.contrastive_optimizer.apply_gradients(\n",
+        "            zip(\n",
+        "                gradients,\n",
+        "                self.encoder.trainable_weights + self.projection_head.trainable_weights,\n",
+        "            )\n",
+        "        )\n",
+        "        self.contrastive_loss_tracker.update_state(contrastive_loss)\n",
+        "\n",
+        "        # Labels are only used in evalutation for an on-the-fly logistic regression\n",
+        "        preprocessed_images = self.classification_augmenter(\n",
+        "            labeled_images, training=True\n",
+        "        )\n",
+        "        with tf.GradientTape() as tape:\n",
+        "            # the encoder is used in inference mode here to avoid regularization\n",
+        "            # and updating the batch normalization paramers if they are used\n",
+        "            features = self.encoder(preprocessed_images, training=False)\n",
+        "            class_logits = self.linear_probe(features, training=True)\n",
+        "            probe_loss = self.probe_loss(labels, class_logits)\n",
+        "            # tf.print(\"labels: \", labels, summarize=-1)\n",
+        "            # tf.print(\"class_logits: \", class_logits, summarize=-1)\n",
+        "            # tf.print(\"probe_loss: \", probe_loss, summarize=-1)\n",
+        "        gradients = tape.gradient(probe_loss, self.linear_probe.trainable_weights)\n",
+        "        self.probe_optimizer.apply_gradients(\n",
+        "            zip(gradients, self.linear_probe.trainable_weights)\n",
+        "        )\n",
+        "        self.probe_loss_tracker.update_state(probe_loss)\n",
+        "        self.probe_accuracy.update_state(labels, class_logits)\n",
+        "        self.top3_accuracy.update_state(labels, class_logits)\n",
+        "        self.top5_accuracy.update_state(labels, class_logits)\n",
+        "\n",
+        "        return {m.name: m.result() for m in self.metrics}\n",
+        "\n",
+        "    def test_step(self, data):\n",
+        "        labeled_images, labels = data\n",
+        "\n",
+        "        # For testing the components are used with a training=False flag\n",
+        "        preprocessed_images = self.classification_augmenter(\n",
+        "            labeled_images, training=False\n",
+        "        )\n",
+        "        features = self.encoder(preprocessed_images, training=False)\n",
+        "        class_logits = self.linear_probe(features, training=False)\n",
+        "        probe_loss = self.probe_loss(labels, class_logits)\n",
+        "        self.probe_loss_tracker.update_state(probe_loss)\n",
+        "        self.probe_accuracy.update_state(labels, class_logits)\n",
+        "        self.top3_accuracy.update_state(labels, class_logits)\n",
+        "        self.top5_accuracy.update_state(labels, class_logits)\n",
+        "\n",
+        "        # Only the probe metrics are logged at test time\n",
+        "        return {m.name: m.result() for m in self.metrics[2:]}\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 47,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "executionInfo": {
+          "elapsed": 1163,
+          "status": "ok",
+          "timestamp": 1672852753912,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "Dy_1kCq3EBeC",
+        "outputId": "4bdbbd5f-6523-459e-bbda-21faef22a5a5"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Model: \"encoder\"\n",
+            "_________________________________________________________________\n",
+            " Layer (type)                Output Shape              Param #   \n",
+            "=================================================================\n",
+            " mobilenet_1.00_128 (Functio  (None, 4, 4, 1024)       3228864   \n",
+            " nal)                                                            \n",
+            "                                                                 \n",
+            " flatten_16 (Flatten)        (None, 16384)             0         \n",
+            "                                                                 \n",
+            " dense_16 (Dense)            (None, 128)               2097280   \n",
+            "                                                                 \n",
+            "=================================================================\n",
+            "Total params: 5,326,144\n",
+            "Trainable params: 5,304,256\n",
+            "Non-trainable params: 21,888\n",
+            "_________________________________________________________________\n",
+            "Model: \"projection_head\"\n",
+            "_________________________________________________________________\n",
+            " Layer (type)                Output Shape              Param #   \n",
+            "=================================================================\n",
+            " dense_17 (Dense)            (None, 128)               16512     \n",
+            "                                                                 \n",
+            " dense_18 (Dense)            (None, 128)               16512     \n",
+            "                                                                 \n",
+            "=================================================================\n",
+            "Total params: 33,024\n",
+            "Trainable params: 33,024\n",
+            "Non-trainable params: 0\n",
+            "_________________________________________________________________\n",
+            "Model: \"linear_probe\"\n",
+            "_________________________________________________________________\n",
+            " Layer (type)                Output Shape              Param #   \n",
+            "=================================================================\n",
+            " dense_19 (Dense)            (None, 18)                2322      \n",
+            "                                                                 \n",
+            "=================================================================\n",
+            "Total params: 2,322\n",
+            "Trainable params: 2,322\n",
+            "Non-trainable params: 0\n",
+            "_________________________________________________________________\n"
+          ]
+        }
+      ],
+      "source": [
+        "# Contrastive pretraining\n",
+        "pretraining_model = ContrastiveModel()\n",
+        "\n",
+        "pretraining_model.compile(\n",
+        "    contrastive_optimizer=keras.optimizers.Adam(),\n",
+        "    probe_optimizer=keras.optimizers.Adam(),\n",
+        ")\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 48,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "executionInfo": {
+          "elapsed": 7760227,
+          "status": "ok",
+          "timestamp": 1672860585998,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "54hLdY3qEyhv",
+        "outputId": "54a5cf60-1e09-47e7-a4a9-8724b641dfc7"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Epoch 1/100\n",
+            "36/36 [==============================] - 123s 2s/step - c_loss: 4.5858 - c_acc: 0.0221 - p_loss: 3.3480 - top1: 0.0583 - top3: 0.1444 - top5: 0.2250 - val_p_loss: 3.2398 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 2/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 4.3233 - c_acc: 0.0345 - p_loss: 3.1677 - top1: 0.0444 - top3: 0.1500 - top5: 0.2639 - val_p_loss: 3.0327 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 3/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 4.1441 - c_acc: 0.0593 - p_loss: 3.3851 - top1: 0.0472 - top3: 0.1639 - top5: 0.2639 - val_p_loss: 3.3707 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 4/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 4.0228 - c_acc: 0.0749 - p_loss: 3.7335 - top1: 0.0389 - top3: 0.1361 - top5: 0.2361 - val_p_loss: 3.9840 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 5/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 3.9168 - c_acc: 0.0817 - p_loss: 3.7751 - top1: 0.0417 - top3: 0.1611 - top5: 0.2972 - val_p_loss: 3.9558 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 6/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 3.7263 - c_acc: 0.1096 - p_loss: 3.8780 - top1: 0.0500 - top3: 0.1694 - top5: 0.2667 - val_p_loss: 4.0290 - val_top1: 0.0556 - val_top3: 0.1667 - val_top5: 0.2778\n",
+            "Epoch 7/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 3.6288 - c_acc: 0.1346 - p_loss: 3.9558 - top1: 0.0444 - top3: 0.1500 - top5: 0.2528 - val_p_loss: 3.5124 - val_top1: 0.0506 - val_top3: 0.1583 - val_top5: 0.2567\n",
+            "Epoch 8/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 3.4966 - c_acc: 0.1503 - p_loss: 4.8822 - top1: 0.0556 - top3: 0.1528 - top5: 0.2694 - val_p_loss: 6.9783 - val_top1: 0.0467 - val_top3: 0.1422 - val_top5: 0.2444\n",
+            "Epoch 9/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 3.4071 - c_acc: 0.1678 - p_loss: 8.0423 - top1: 0.0667 - top3: 0.1694 - top5: 0.2361 - val_p_loss: 14.2141 - val_top1: 0.0517 - val_top3: 0.1411 - val_top5: 0.2578\n",
+            "Epoch 10/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 3.3193 - c_acc: 0.1788 - p_loss: 15.4594 - top1: 0.0667 - top3: 0.1583 - top5: 0.2722 - val_p_loss: 20.4377 - val_top1: 0.0667 - val_top3: 0.1733 - val_top5: 0.2894\n",
+            "Epoch 11/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 3.2232 - c_acc: 0.2154 - p_loss: 15.9685 - top1: 0.0639 - top3: 0.1722 - top5: 0.3250 - val_p_loss: 13.7904 - val_top1: 0.0833 - val_top3: 0.2356 - val_top5: 0.3744\n",
+            "Epoch 12/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 3.1142 - c_acc: 0.2345 - p_loss: 15.0086 - top1: 0.0667 - top3: 0.2444 - top5: 0.3694 - val_p_loss: 16.3864 - val_top1: 0.0861 - val_top3: 0.2378 - val_top5: 0.3794\n",
+            "Epoch 13/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 3.0456 - c_acc: 0.2443 - p_loss: 12.2634 - top1: 0.0778 - top3: 0.2611 - top5: 0.4111 - val_p_loss: 11.8923 - val_top1: 0.0872 - val_top3: 0.2433 - val_top5: 0.3789\n",
+            "Epoch 14/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 2.9622 - c_acc: 0.2731 - p_loss: 13.0498 - top1: 0.1167 - top3: 0.2556 - top5: 0.4194 - val_p_loss: 14.6520 - val_top1: 0.0939 - val_top3: 0.2606 - val_top5: 0.3961\n",
+            "Epoch 15/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.8636 - c_acc: 0.2771 - p_loss: 12.1137 - top1: 0.1111 - top3: 0.2750 - top5: 0.4167 - val_p_loss: 12.7450 - val_top1: 0.1072 - val_top3: 0.2728 - val_top5: 0.4033\n",
+            "Epoch 16/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.7781 - c_acc: 0.3039 - p_loss: 12.0209 - top1: 0.1361 - top3: 0.3167 - top5: 0.4806 - val_p_loss: 14.7042 - val_top1: 0.0956 - val_top3: 0.2672 - val_top5: 0.4144\n",
+            "Epoch 17/100\n",
+            "36/36 [==============================] - 74s 2s/step - c_loss: 2.7030 - c_acc: 0.3223 - p_loss: 11.7686 - top1: 0.1194 - top3: 0.2917 - top5: 0.4528 - val_p_loss: 13.2911 - val_top1: 0.0950 - val_top3: 0.2611 - val_top5: 0.4111\n",
+            "Epoch 18/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.6417 - c_acc: 0.3504 - p_loss: 13.2002 - top1: 0.1333 - top3: 0.3167 - top5: 0.4778 - val_p_loss: 18.4293 - val_top1: 0.0872 - val_top3: 0.2328 - val_top5: 0.3683\n",
+            "Epoch 19/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 2.5437 - c_acc: 0.3527 - p_loss: 12.8320 - top1: 0.1250 - top3: 0.3333 - top5: 0.4833 - val_p_loss: 14.6945 - val_top1: 0.0861 - val_top3: 0.2589 - val_top5: 0.4083\n",
+            "Epoch 20/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.5015 - c_acc: 0.3711 - p_loss: 11.3523 - top1: 0.1167 - top3: 0.2889 - top5: 0.4333 - val_p_loss: 10.8001 - val_top1: 0.0878 - val_top3: 0.2778 - val_top5: 0.4394\n",
+            "Epoch 21/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 2.4045 - c_acc: 0.3948 - p_loss: 12.3947 - top1: 0.1333 - top3: 0.2806 - top5: 0.4611 - val_p_loss: 13.1532 - val_top1: 0.1183 - val_top3: 0.2933 - val_top5: 0.4294\n",
+            "Epoch 22/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 2.3753 - c_acc: 0.4043 - p_loss: 11.7714 - top1: 0.1528 - top3: 0.3417 - top5: 0.5111 - val_p_loss: 13.3982 - val_top1: 0.1172 - val_top3: 0.3033 - val_top5: 0.4522\n",
+            "Epoch 23/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 2.3454 - c_acc: 0.4131 - p_loss: 11.6782 - top1: 0.1500 - top3: 0.3722 - top5: 0.5000 - val_p_loss: 15.9087 - val_top1: 0.1050 - val_top3: 0.2589 - val_top5: 0.3933\n",
+            "Epoch 24/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.2702 - c_acc: 0.4275 - p_loss: 13.5934 - top1: 0.1250 - top3: 0.3222 - top5: 0.4972 - val_p_loss: 15.7255 - val_top1: 0.1067 - val_top3: 0.2850 - val_top5: 0.4317\n",
+            "Epoch 25/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 2.2127 - c_acc: 0.4388 - p_loss: 12.0966 - top1: 0.1611 - top3: 0.3944 - top5: 0.5389 - val_p_loss: 14.6215 - val_top1: 0.1178 - val_top3: 0.2922 - val_top5: 0.4367\n",
+            "Epoch 26/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.0818 - c_acc: 0.4717 - p_loss: 11.5599 - top1: 0.1306 - top3: 0.3972 - top5: 0.5500 - val_p_loss: 14.2833 - val_top1: 0.1144 - val_top3: 0.3056 - val_top5: 0.4450\n",
+            "Epoch 27/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.0961 - c_acc: 0.4699 - p_loss: 12.1741 - top1: 0.1556 - top3: 0.3472 - top5: 0.4917 - val_p_loss: 14.5990 - val_top1: 0.1122 - val_top3: 0.2900 - val_top5: 0.4328\n",
+            "Epoch 28/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.0545 - c_acc: 0.4784 - p_loss: 11.3273 - top1: 0.1500 - top3: 0.3639 - top5: 0.5194 - val_p_loss: 13.4422 - val_top1: 0.1256 - val_top3: 0.3117 - val_top5: 0.4589\n",
+            "Epoch 29/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 2.0048 - c_acc: 0.4900 - p_loss: 10.6976 - top1: 0.1750 - top3: 0.3750 - top5: 0.5556 - val_p_loss: 13.0510 - val_top1: 0.1222 - val_top3: 0.3039 - val_top5: 0.4456\n",
+            "Epoch 30/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 1.9558 - c_acc: 0.5009 - p_loss: 10.5623 - top1: 0.1806 - top3: 0.3722 - top5: 0.5111 - val_p_loss: 14.6967 - val_top1: 0.1406 - val_top3: 0.3233 - val_top5: 0.4789\n",
+            "Epoch 31/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.8821 - c_acc: 0.5165 - p_loss: 11.8279 - top1: 0.1556 - top3: 0.4306 - top5: 0.5833 - val_p_loss: 15.6465 - val_top1: 0.1294 - val_top3: 0.2994 - val_top5: 0.4567\n",
+            "Epoch 32/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 1.8157 - c_acc: 0.5367 - p_loss: 10.8144 - top1: 0.1556 - top3: 0.4250 - top5: 0.5694 - val_p_loss: 13.1304 - val_top1: 0.1328 - val_top3: 0.3183 - val_top5: 0.4694\n",
+            "Epoch 33/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.7672 - c_acc: 0.5437 - p_loss: 10.5795 - top1: 0.1639 - top3: 0.4000 - top5: 0.5500 - val_p_loss: 14.8528 - val_top1: 0.1172 - val_top3: 0.3033 - val_top5: 0.4617\n",
+            "Epoch 34/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.7541 - c_acc: 0.5567 - p_loss: 11.7489 - top1: 0.2056 - top3: 0.4500 - top5: 0.5806 - val_p_loss: 14.1244 - val_top1: 0.1578 - val_top3: 0.3661 - val_top5: 0.5194\n",
+            "Epoch 35/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.7076 - c_acc: 0.5707 - p_loss: 11.5770 - top1: 0.1806 - top3: 0.4333 - top5: 0.5944 - val_p_loss: 16.3164 - val_top1: 0.1378 - val_top3: 0.3267 - val_top5: 0.4839\n",
+            "Epoch 36/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.6321 - c_acc: 0.5764 - p_loss: 11.7358 - top1: 0.1750 - top3: 0.3889 - top5: 0.5583 - val_p_loss: 15.7149 - val_top1: 0.1206 - val_top3: 0.3028 - val_top5: 0.4633\n",
+            "Epoch 37/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.6170 - c_acc: 0.5939 - p_loss: 11.5618 - top1: 0.1639 - top3: 0.3639 - top5: 0.5472 - val_p_loss: 14.4601 - val_top1: 0.1378 - val_top3: 0.3267 - val_top5: 0.4922\n",
+            "Epoch 38/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.5744 - c_acc: 0.5957 - p_loss: 10.3889 - top1: 0.1917 - top3: 0.4306 - top5: 0.5778 - val_p_loss: 15.2097 - val_top1: 0.1300 - val_top3: 0.3317 - val_top5: 0.4678\n",
+            "Epoch 39/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.5715 - c_acc: 0.5942 - p_loss: 11.6093 - top1: 0.1500 - top3: 0.3750 - top5: 0.5667 - val_p_loss: 14.9725 - val_top1: 0.1339 - val_top3: 0.3200 - val_top5: 0.4828\n",
+            "Epoch 40/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.4884 - c_acc: 0.6280 - p_loss: 11.1567 - top1: 0.2083 - top3: 0.4167 - top5: 0.5944 - val_p_loss: 13.5139 - val_top1: 0.1472 - val_top3: 0.3444 - val_top5: 0.4906\n",
+            "Epoch 41/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.4570 - c_acc: 0.6274 - p_loss: 9.7995 - top1: 0.2056 - top3: 0.4556 - top5: 0.5861 - val_p_loss: 14.9412 - val_top1: 0.1406 - val_top3: 0.3478 - val_top5: 0.5083\n",
+            "Epoch 42/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.4383 - c_acc: 0.6384 - p_loss: 10.9668 - top1: 0.1972 - top3: 0.4444 - top5: 0.6333 - val_p_loss: 14.3386 - val_top1: 0.1444 - val_top3: 0.3439 - val_top5: 0.4989\n",
+            "Epoch 43/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.4022 - c_acc: 0.6427 - p_loss: 10.5637 - top1: 0.1944 - top3: 0.4306 - top5: 0.5917 - val_p_loss: 11.8865 - val_top1: 0.1494 - val_top3: 0.3656 - val_top5: 0.5122\n",
+            "Epoch 44/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.4039 - c_acc: 0.6402 - p_loss: 9.6471 - top1: 0.2389 - top3: 0.4861 - top5: 0.6306 - val_p_loss: 12.6986 - val_top1: 0.1600 - val_top3: 0.3711 - val_top5: 0.5433\n",
+            "Epoch 45/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.3578 - c_acc: 0.6566 - p_loss: 9.8112 - top1: 0.2111 - top3: 0.4778 - top5: 0.6556 - val_p_loss: 14.4274 - val_top1: 0.1394 - val_top3: 0.3433 - val_top5: 0.5100\n",
+            "Epoch 46/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.3658 - c_acc: 0.6559 - p_loss: 9.4212 - top1: 0.2333 - top3: 0.4944 - top5: 0.6667 - val_p_loss: 13.3540 - val_top1: 0.1467 - val_top3: 0.3594 - val_top5: 0.5178\n",
+            "Epoch 47/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.3597 - c_acc: 0.6552 - p_loss: 9.6175 - top1: 0.2139 - top3: 0.4833 - top5: 0.6639 - val_p_loss: 12.4351 - val_top1: 0.1561 - val_top3: 0.3739 - val_top5: 0.5317\n",
+            "Epoch 48/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.3374 - c_acc: 0.6688 - p_loss: 10.1261 - top1: 0.2306 - top3: 0.4500 - top5: 0.6139 - val_p_loss: 13.9852 - val_top1: 0.1406 - val_top3: 0.3550 - val_top5: 0.5261\n",
+            "Epoch 49/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.2790 - c_acc: 0.6794 - p_loss: 9.9443 - top1: 0.2417 - top3: 0.5000 - top5: 0.6528 - val_p_loss: 14.9369 - val_top1: 0.1600 - val_top3: 0.3617 - val_top5: 0.5183\n",
+            "Epoch 50/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.3141 - c_acc: 0.6706 - p_loss: 9.6863 - top1: 0.2556 - top3: 0.4889 - top5: 0.6306 - val_p_loss: 14.2001 - val_top1: 0.1528 - val_top3: 0.3522 - val_top5: 0.5222\n",
+            "Epoch 51/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.2776 - c_acc: 0.6753 - p_loss: 10.1876 - top1: 0.2250 - top3: 0.4694 - top5: 0.6278 - val_p_loss: 15.1207 - val_top1: 0.1433 - val_top3: 0.3583 - val_top5: 0.5144\n",
+            "Epoch 52/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.2598 - c_acc: 0.6881 - p_loss: 9.7904 - top1: 0.2444 - top3: 0.5056 - top5: 0.6639 - val_p_loss: 13.4377 - val_top1: 0.1589 - val_top3: 0.3722 - val_top5: 0.5461\n",
+            "Epoch 53/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.2444 - c_acc: 0.6850 - p_loss: 9.2116 - top1: 0.2056 - top3: 0.5667 - top5: 0.6972 - val_p_loss: 13.8246 - val_top1: 0.1656 - val_top3: 0.3944 - val_top5: 0.5550\n",
+            "Epoch 54/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.2020 - c_acc: 0.6977 - p_loss: 9.9388 - top1: 0.2333 - top3: 0.4833 - top5: 0.6528 - val_p_loss: 14.0555 - val_top1: 0.1633 - val_top3: 0.3883 - val_top5: 0.5372\n",
+            "Epoch 55/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.2038 - c_acc: 0.7052 - p_loss: 9.7780 - top1: 0.2611 - top3: 0.5194 - top5: 0.7028 - val_p_loss: 14.8887 - val_top1: 0.1639 - val_top3: 0.3667 - val_top5: 0.5317\n",
+            "Epoch 56/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.1454 - c_acc: 0.7235 - p_loss: 9.1994 - top1: 0.2556 - top3: 0.5167 - top5: 0.6889 - val_p_loss: 13.5513 - val_top1: 0.1806 - val_top3: 0.4122 - val_top5: 0.5683\n",
+            "Epoch 57/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.1542 - c_acc: 0.7082 - p_loss: 9.1845 - top1: 0.2472 - top3: 0.5389 - top5: 0.6889 - val_p_loss: 12.4345 - val_top1: 0.1667 - val_top3: 0.3861 - val_top5: 0.5356\n",
+            "Epoch 58/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.1234 - c_acc: 0.7193 - p_loss: 8.2431 - top1: 0.2583 - top3: 0.5611 - top5: 0.7472 - val_p_loss: 13.5931 - val_top1: 0.1806 - val_top3: 0.3828 - val_top5: 0.5550\n",
+            "Epoch 59/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.1042 - c_acc: 0.7294 - p_loss: 8.6656 - top1: 0.2694 - top3: 0.5611 - top5: 0.7167 - val_p_loss: 12.3844 - val_top1: 0.1772 - val_top3: 0.4056 - val_top5: 0.5800\n",
+            "Epoch 60/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.1115 - c_acc: 0.7183 - p_loss: 8.8357 - top1: 0.2667 - top3: 0.5583 - top5: 0.7250 - val_p_loss: 12.8478 - val_top1: 0.1706 - val_top3: 0.3783 - val_top5: 0.5556\n",
+            "Epoch 61/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.0781 - c_acc: 0.7362 - p_loss: 9.4532 - top1: 0.2250 - top3: 0.5361 - top5: 0.7472 - val_p_loss: 17.2815 - val_top1: 0.1567 - val_top3: 0.3761 - val_top5: 0.5456\n",
+            "Epoch 62/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.1008 - c_acc: 0.7309 - p_loss: 10.2386 - top1: 0.2806 - top3: 0.5611 - top5: 0.7194 - val_p_loss: 15.6925 - val_top1: 0.1617 - val_top3: 0.3811 - val_top5: 0.5306\n",
+            "Epoch 63/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.0996 - c_acc: 0.7226 - p_loss: 9.4412 - top1: 0.2611 - top3: 0.5222 - top5: 0.7167 - val_p_loss: 15.7595 - val_top1: 0.1517 - val_top3: 0.3578 - val_top5: 0.5222\n",
+            "Epoch 64/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.0744 - c_acc: 0.7355 - p_loss: 9.0500 - top1: 0.3028 - top3: 0.5583 - top5: 0.7361 - val_p_loss: 14.0074 - val_top1: 0.1733 - val_top3: 0.4017 - val_top5: 0.5628\n",
+            "Epoch 65/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.0754 - c_acc: 0.7338 - p_loss: 9.6131 - top1: 0.2556 - top3: 0.5111 - top5: 0.6917 - val_p_loss: 13.8470 - val_top1: 0.1783 - val_top3: 0.3933 - val_top5: 0.5367\n",
+            "Epoch 66/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.0128 - c_acc: 0.7580 - p_loss: 9.7066 - top1: 0.2667 - top3: 0.5361 - top5: 0.7083 - val_p_loss: 14.1280 - val_top1: 0.1750 - val_top3: 0.4033 - val_top5: 0.5694\n",
+            "Epoch 67/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 1.0364 - c_acc: 0.7472 - p_loss: 8.0576 - top1: 0.3222 - top3: 0.5611 - top5: 0.7028 - val_p_loss: 13.4836 - val_top1: 0.1850 - val_top3: 0.4056 - val_top5: 0.5722\n",
+            "Epoch 68/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9931 - c_acc: 0.7585 - p_loss: 8.1475 - top1: 0.2750 - top3: 0.5944 - top5: 0.7583 - val_p_loss: 13.4056 - val_top1: 0.1722 - val_top3: 0.4133 - val_top5: 0.5761\n",
+            "Epoch 69/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9510 - c_acc: 0.7684 - p_loss: 7.4657 - top1: 0.3111 - top3: 0.6139 - top5: 0.7750 - val_p_loss: 11.9338 - val_top1: 0.1722 - val_top3: 0.4072 - val_top5: 0.5706\n",
+            "Epoch 70/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9853 - c_acc: 0.7615 - p_loss: 9.1287 - top1: 0.2500 - top3: 0.5556 - top5: 0.6944 - val_p_loss: 13.7218 - val_top1: 0.1778 - val_top3: 0.4044 - val_top5: 0.5722\n",
+            "Epoch 71/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9714 - c_acc: 0.7633 - p_loss: 7.2602 - top1: 0.3167 - top3: 0.6028 - top5: 0.7694 - val_p_loss: 14.3024 - val_top1: 0.1606 - val_top3: 0.3989 - val_top5: 0.5406\n",
+            "Epoch 72/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9744 - c_acc: 0.7622 - p_loss: 8.1309 - top1: 0.2861 - top3: 0.5556 - top5: 0.7778 - val_p_loss: 14.0316 - val_top1: 0.1672 - val_top3: 0.3983 - val_top5: 0.5450\n",
+            "Epoch 73/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9731 - c_acc: 0.7683 - p_loss: 7.9508 - top1: 0.3306 - top3: 0.6222 - top5: 0.7639 - val_p_loss: 14.2259 - val_top1: 0.1900 - val_top3: 0.4178 - val_top5: 0.5728\n",
+            "Epoch 74/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9397 - c_acc: 0.7739 - p_loss: 8.4960 - top1: 0.3333 - top3: 0.5917 - top5: 0.7472 - val_p_loss: 14.2826 - val_top1: 0.1789 - val_top3: 0.4128 - val_top5: 0.5683\n",
+            "Epoch 75/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9270 - c_acc: 0.7807 - p_loss: 7.8952 - top1: 0.3278 - top3: 0.6194 - top5: 0.7722 - val_p_loss: 16.1326 - val_top1: 0.1722 - val_top3: 0.3911 - val_top5: 0.5617\n",
+            "Epoch 76/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8777 - c_acc: 0.7956 - p_loss: 8.2881 - top1: 0.2972 - top3: 0.5917 - top5: 0.7667 - val_p_loss: 16.7344 - val_top1: 0.1739 - val_top3: 0.3917 - val_top5: 0.5378\n",
+            "Epoch 77/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8832 - c_acc: 0.7904 - p_loss: 7.9199 - top1: 0.2750 - top3: 0.5750 - top5: 0.7722 - val_p_loss: 13.3840 - val_top1: 0.1767 - val_top3: 0.3833 - val_top5: 0.5467\n",
+            "Epoch 78/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8948 - c_acc: 0.7913 - p_loss: 8.0667 - top1: 0.3250 - top3: 0.6250 - top5: 0.7556 - val_p_loss: 16.1052 - val_top1: 0.1811 - val_top3: 0.3856 - val_top5: 0.5283\n",
+            "Epoch 79/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9031 - c_acc: 0.7859 - p_loss: 8.1451 - top1: 0.3139 - top3: 0.5750 - top5: 0.7444 - val_p_loss: 13.8611 - val_top1: 0.1794 - val_top3: 0.3939 - val_top5: 0.5522\n",
+            "Epoch 80/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8900 - c_acc: 0.7917 - p_loss: 8.5767 - top1: 0.2889 - top3: 0.5611 - top5: 0.7472 - val_p_loss: 15.4069 - val_top1: 0.1744 - val_top3: 0.3856 - val_top5: 0.5589\n",
+            "Epoch 81/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.9119 - c_acc: 0.7836 - p_loss: 8.8942 - top1: 0.3389 - top3: 0.6139 - top5: 0.7528 - val_p_loss: 18.1046 - val_top1: 0.1811 - val_top3: 0.3883 - val_top5: 0.5450\n",
+            "Epoch 82/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8728 - c_acc: 0.8066 - p_loss: 8.0897 - top1: 0.3222 - top3: 0.6028 - top5: 0.7583 - val_p_loss: 14.7780 - val_top1: 0.2000 - val_top3: 0.4072 - val_top5: 0.5767\n",
+            "Epoch 83/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8771 - c_acc: 0.7904 - p_loss: 7.0899 - top1: 0.3722 - top3: 0.6444 - top5: 0.8056 - val_p_loss: 13.9046 - val_top1: 0.1883 - val_top3: 0.4044 - val_top5: 0.5583\n",
+            "Epoch 84/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 0.8539 - c_acc: 0.7995 - p_loss: 7.1536 - top1: 0.3528 - top3: 0.6139 - top5: 0.7667 - val_p_loss: 13.9198 - val_top1: 0.1900 - val_top3: 0.4206 - val_top5: 0.5694\n",
+            "Epoch 85/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8026 - c_acc: 0.8149 - p_loss: 7.3116 - top1: 0.3194 - top3: 0.6111 - top5: 0.7917 - val_p_loss: 13.8208 - val_top1: 0.1722 - val_top3: 0.4056 - val_top5: 0.5578\n",
+            "Epoch 86/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 0.8241 - c_acc: 0.8045 - p_loss: 7.9628 - top1: 0.3250 - top3: 0.6611 - top5: 0.8083 - val_p_loss: 13.7853 - val_top1: 0.1989 - val_top3: 0.4428 - val_top5: 0.5967\n",
+            "Epoch 87/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8377 - c_acc: 0.8020 - p_loss: 7.1914 - top1: 0.3611 - top3: 0.6778 - top5: 0.7833 - val_p_loss: 13.0113 - val_top1: 0.1967 - val_top3: 0.4161 - val_top5: 0.5761\n",
+            "Epoch 88/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 0.8241 - c_acc: 0.8096 - p_loss: 7.6905 - top1: 0.3028 - top3: 0.6389 - top5: 0.7833 - val_p_loss: 14.5146 - val_top1: 0.1750 - val_top3: 0.4250 - val_top5: 0.5844\n",
+            "Epoch 89/100\n",
+            "36/36 [==============================] - 71s 2s/step - c_loss: 0.7880 - c_acc: 0.8149 - p_loss: 7.3010 - top1: 0.3472 - top3: 0.6528 - top5: 0.7972 - val_p_loss: 13.7444 - val_top1: 0.1772 - val_top3: 0.4017 - val_top5: 0.5489\n",
+            "Epoch 90/100\n",
+            "36/36 [==============================] - 71s 2s/step - c_loss: 0.8125 - c_acc: 0.8174 - p_loss: 6.4776 - top1: 0.3861 - top3: 0.6611 - top5: 0.7944 - val_p_loss: 15.6358 - val_top1: 0.1939 - val_top3: 0.4200 - val_top5: 0.5689\n",
+            "Epoch 91/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 0.8296 - c_acc: 0.8148 - p_loss: 7.4747 - top1: 0.3833 - top3: 0.6417 - top5: 0.8028 - val_p_loss: 17.4104 - val_top1: 0.1894 - val_top3: 0.4356 - val_top5: 0.5833\n",
+            "Epoch 92/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8112 - c_acc: 0.8141 - p_loss: 7.1051 - top1: 0.3444 - top3: 0.6361 - top5: 0.7750 - val_p_loss: 13.0306 - val_top1: 0.2122 - val_top3: 0.4350 - val_top5: 0.5800\n",
+            "Epoch 93/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 0.7822 - c_acc: 0.8203 - p_loss: 7.0252 - top1: 0.3667 - top3: 0.6389 - top5: 0.7944 - val_p_loss: 13.1084 - val_top1: 0.1972 - val_top3: 0.4222 - val_top5: 0.5722\n",
+            "Epoch 94/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.7701 - c_acc: 0.8247 - p_loss: 7.4745 - top1: 0.3444 - top3: 0.6278 - top5: 0.7806 - val_p_loss: 14.6724 - val_top1: 0.1956 - val_top3: 0.4322 - val_top5: 0.5772\n",
+            "Epoch 95/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 0.7577 - c_acc: 0.8263 - p_loss: 7.1471 - top1: 0.3639 - top3: 0.6861 - top5: 0.8083 - val_p_loss: 17.3091 - val_top1: 0.1844 - val_top3: 0.4156 - val_top5: 0.5700\n",
+            "Epoch 96/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.8027 - c_acc: 0.8143 - p_loss: 8.1323 - top1: 0.3333 - top3: 0.6250 - top5: 0.7806 - val_p_loss: 14.7396 - val_top1: 0.1756 - val_top3: 0.3994 - val_top5: 0.5689\n",
+            "Epoch 97/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.7414 - c_acc: 0.8308 - p_loss: 7.8203 - top1: 0.3583 - top3: 0.6500 - top5: 0.8028 - val_p_loss: 18.2735 - val_top1: 0.1711 - val_top3: 0.3839 - val_top5: 0.5378\n",
+            "Epoch 98/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 0.7463 - c_acc: 0.8347 - p_loss: 7.8030 - top1: 0.3417 - top3: 0.6417 - top5: 0.8167 - val_p_loss: 14.3665 - val_top1: 0.2022 - val_top3: 0.4178 - val_top5: 0.5839\n",
+            "Epoch 99/100\n",
+            "36/36 [==============================] - 72s 2s/step - c_loss: 0.7436 - c_acc: 0.8356 - p_loss: 7.8708 - top1: 0.3583 - top3: 0.6389 - top5: 0.7889 - val_p_loss: 13.9370 - val_top1: 0.1989 - val_top3: 0.4144 - val_top5: 0.5406\n",
+            "Epoch 100/100\n",
+            "36/36 [==============================] - 73s 2s/step - c_loss: 0.7453 - c_acc: 0.8360 - p_loss: 7.7088 - top1: 0.3889 - top3: 0.6722 - top5: 0.7917 - val_p_loss: 16.2370 - val_top1: 0.1833 - val_top3: 0.4211 - val_top5: 0.5722\n"
+          ]
+        }
+      ],
+      "source": [
+        "pretraining_history = pretraining_model.fit(\n",
+        "    train_dataset, epochs=100, validation_data=test_dataset\n",
+        ")\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 53,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "executionInfo": {
+          "elapsed": 621,
+          "status": "ok",
+          "timestamp": 1672861180504,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "5xr-xUKdQxeJ",
+        "outputId": "c6f8ff85-2ada-4fe6-b3d7-7bec59b75d5f"
+      },
+      "outputs": [
+        {
+          "data": {
+            "text/plain": [
+              "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x7effafd6ed30>"
+            ]
+          },
+          "execution_count": 53,
+          "metadata": {},
+          "output_type": "execute_result"
+        }
+      ],
+      "source": [
+        "# pretraining_model.save_weights('./pretraining_model_weights')\n",
+        "pretraining_model.load_weights('./pretraining_model_weights')\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": 54,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "executionInfo": {
+          "elapsed": 427696,
+          "status": "ok",
+          "timestamp": 1672861609362,
+          "user": {
+            "displayName": "Laureηt",
+            "userId": "01668146649491175447"
+          },
+          "user_tz": -60
+        },
+        "id": "7QIoLloWtwhE",
+        "outputId": "3e76a184-0573-40da-e37b-c31d58ec1430"
+      },
+      "outputs": [
+        {
+          "name": "stdout",
+          "output_type": "stream",
+          "text": [
+            "Epoch 1/100\n",
+            "12/12 [==============================] - 8s 318ms/step - loss: 73.9322 - top1: 0.0944 - top3: 0.2444 - top5: 0.3528 - val_loss: 99.8719 - val_top1: 0.0611 - val_top3: 0.2300 - val_top5: 0.3511\n",
+            "Epoch 2/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 25.3839 - top1: 0.1861 - top3: 0.4361 - top5: 0.5806 - val_loss: 95.9250 - val_top1: 0.1106 - val_top3: 0.2767 - val_top5: 0.4217\n",
+            "Epoch 3/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 12.0852 - top1: 0.3222 - top3: 0.6028 - top5: 0.7111 - val_loss: 71.7181 - val_top1: 0.1150 - val_top3: 0.2944 - val_top5: 0.4400\n",
+            "Epoch 4/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 8.0282 - top1: 0.4167 - top3: 0.6833 - top5: 0.7944 - val_loss: 70.8434 - val_top1: 0.1139 - val_top3: 0.2989 - val_top5: 0.4267\n",
+            "Epoch 5/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 6.3533 - top1: 0.4083 - top3: 0.7111 - top5: 0.8111 - val_loss: 61.7378 - val_top1: 0.1428 - val_top3: 0.3306 - val_top5: 0.4567\n",
+            "Epoch 6/100\n",
+            "12/12 [==============================] - 3s 276ms/step - loss: 4.8435 - top1: 0.5111 - top3: 0.7528 - top5: 0.8528 - val_loss: 55.3876 - val_top1: 0.1544 - val_top3: 0.3294 - val_top5: 0.4722\n",
+            "Epoch 7/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 3.9249 - top1: 0.5000 - top3: 0.8000 - top5: 0.9083 - val_loss: 47.2079 - val_top1: 0.1461 - val_top3: 0.3333 - val_top5: 0.4594\n",
+            "Epoch 8/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 3.7509 - top1: 0.6056 - top3: 0.8750 - top5: 0.9361 - val_loss: 51.1446 - val_top1: 0.1183 - val_top3: 0.3044 - val_top5: 0.4478\n",
+            "Epoch 9/100\n",
+            "12/12 [==============================] - 3s 280ms/step - loss: 2.7583 - top1: 0.6444 - top3: 0.8528 - top5: 0.9472 - val_loss: 49.8988 - val_top1: 0.1317 - val_top3: 0.3328 - val_top5: 0.4761\n",
+            "Epoch 10/100\n",
+            "12/12 [==============================] - 3s 284ms/step - loss: 2.9460 - top1: 0.6472 - top3: 0.9000 - top5: 0.9500 - val_loss: 32.7369 - val_top1: 0.1650 - val_top3: 0.3578 - val_top5: 0.4989\n",
+            "Epoch 11/100\n",
+            "12/12 [==============================] - 3s 267ms/step - loss: 2.4896 - top1: 0.6778 - top3: 0.9028 - top5: 0.9611 - val_loss: 36.0452 - val_top1: 0.1517 - val_top3: 0.3672 - val_top5: 0.5006\n",
+            "Epoch 12/100\n",
+            "12/12 [==============================] - 3s 262ms/step - loss: 2.2882 - top1: 0.6611 - top3: 0.8917 - top5: 0.9500 - val_loss: 28.7688 - val_top1: 0.1450 - val_top3: 0.3406 - val_top5: 0.4967\n",
+            "Epoch 13/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 1.9085 - top1: 0.7222 - top3: 0.9083 - top5: 0.9500 - val_loss: 27.5473 - val_top1: 0.1472 - val_top3: 0.3439 - val_top5: 0.4867\n",
+            "Epoch 14/100\n",
+            "12/12 [==============================] - 4s 348ms/step - loss: 1.7324 - top1: 0.7444 - top3: 0.9194 - top5: 0.9639 - val_loss: 30.1766 - val_top1: 0.1450 - val_top3: 0.3517 - val_top5: 0.4894\n",
+            "Epoch 15/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 2.3391 - top1: 0.7250 - top3: 0.9139 - top5: 0.9750 - val_loss: 17.1605 - val_top1: 0.1956 - val_top3: 0.3794 - val_top5: 0.5300\n",
+            "Epoch 16/100\n",
+            "12/12 [==============================] - 3s 271ms/step - loss: 1.8701 - top1: 0.7417 - top3: 0.9222 - top5: 0.9722 - val_loss: 22.7560 - val_top1: 0.1856 - val_top3: 0.3750 - val_top5: 0.5272\n",
+            "Epoch 17/100\n",
+            "12/12 [==============================] - 3s 284ms/step - loss: 1.9086 - top1: 0.7361 - top3: 0.9250 - top5: 0.9722 - val_loss: 25.2979 - val_top1: 0.1839 - val_top3: 0.3817 - val_top5: 0.5194\n",
+            "Epoch 18/100\n",
+            "12/12 [==============================] - 3s 266ms/step - loss: 1.3392 - top1: 0.7889 - top3: 0.9639 - top5: 0.9972 - val_loss: 27.3306 - val_top1: 0.1611 - val_top3: 0.3589 - val_top5: 0.5011\n",
+            "Epoch 19/100\n",
+            "12/12 [==============================] - 3s 269ms/step - loss: 1.7008 - top1: 0.7667 - top3: 0.9472 - top5: 0.9917 - val_loss: 20.3858 - val_top1: 0.1794 - val_top3: 0.3939 - val_top5: 0.5583\n",
+            "Epoch 20/100\n",
+            "12/12 [==============================] - 3s 284ms/step - loss: 1.5783 - top1: 0.7750 - top3: 0.9444 - top5: 0.9722 - val_loss: 20.2930 - val_top1: 0.2317 - val_top3: 0.4322 - val_top5: 0.5889\n",
+            "Epoch 21/100\n",
+            "12/12 [==============================] - 3s 264ms/step - loss: 1.1688 - top1: 0.8306 - top3: 0.9667 - top5: 0.9889 - val_loss: 32.4590 - val_top1: 0.1756 - val_top3: 0.3856 - val_top5: 0.5406\n",
+            "Epoch 22/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 1.3821 - top1: 0.8111 - top3: 0.9611 - top5: 0.9861 - val_loss: 29.8482 - val_top1: 0.1944 - val_top3: 0.3939 - val_top5: 0.5439\n",
+            "Epoch 23/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 1.5115 - top1: 0.8222 - top3: 0.9750 - top5: 0.9861 - val_loss: 25.6451 - val_top1: 0.1783 - val_top3: 0.3856 - val_top5: 0.5250\n",
+            "Epoch 24/100\n",
+            "12/12 [==============================] - 3s 287ms/step - loss: 1.3507 - top1: 0.7944 - top3: 0.9528 - top5: 0.9806 - val_loss: 25.1630 - val_top1: 0.2011 - val_top3: 0.4211 - val_top5: 0.5733\n",
+            "Epoch 25/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 1.0129 - top1: 0.8500 - top3: 0.9750 - top5: 0.9917 - val_loss: 28.6518 - val_top1: 0.1906 - val_top3: 0.3906 - val_top5: 0.5328\n",
+            "Epoch 26/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 1.2421 - top1: 0.8556 - top3: 0.9778 - top5: 0.9972 - val_loss: 26.6470 - val_top1: 0.1761 - val_top3: 0.4022 - val_top5: 0.5306\n",
+            "Epoch 27/100\n",
+            "12/12 [==============================] - 3s 269ms/step - loss: 0.7551 - top1: 0.8806 - top3: 0.9806 - top5: 0.9917 - val_loss: 28.5631 - val_top1: 0.1639 - val_top3: 0.3789 - val_top5: 0.5200\n",
+            "Epoch 28/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 0.9173 - top1: 0.8694 - top3: 0.9639 - top5: 0.9833 - val_loss: 27.9307 - val_top1: 0.1728 - val_top3: 0.3828 - val_top5: 0.5139\n",
+            "Epoch 29/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 1.0824 - top1: 0.8472 - top3: 0.9778 - top5: 0.9944 - val_loss: 38.9105 - val_top1: 0.1406 - val_top3: 0.3322 - val_top5: 0.4844\n",
+            "Epoch 30/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 1.1366 - top1: 0.8417 - top3: 0.9722 - top5: 0.9917 - val_loss: 24.5217 - val_top1: 0.2011 - val_top3: 0.4050 - val_top5: 0.5400\n",
+            "Epoch 31/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 1.2931 - top1: 0.8417 - top3: 0.9556 - top5: 0.9806 - val_loss: 26.1612 - val_top1: 0.2150 - val_top3: 0.4339 - val_top5: 0.5628\n",
+            "Epoch 32/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 1.0229 - top1: 0.8528 - top3: 0.9806 - top5: 0.9917 - val_loss: 30.4698 - val_top1: 0.1706 - val_top3: 0.3689 - val_top5: 0.5189\n",
+            "Epoch 33/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 1.3514 - top1: 0.8500 - top3: 0.9694 - top5: 0.9944 - val_loss: 33.3621 - val_top1: 0.1611 - val_top3: 0.3494 - val_top5: 0.5044\n",
+            "Epoch 34/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 0.9171 - top1: 0.8806 - top3: 0.9750 - top5: 0.9917 - val_loss: 44.8007 - val_top1: 0.1478 - val_top3: 0.3650 - val_top5: 0.5200\n",
+            "Epoch 35/100\n",
+            "12/12 [==============================] - 3s 268ms/step - loss: 0.9032 - top1: 0.8750 - top3: 0.9917 - top5: 0.9944 - val_loss: 24.2278 - val_top1: 0.1767 - val_top3: 0.4144 - val_top5: 0.5567\n",
+            "Epoch 36/100\n",
+            "12/12 [==============================] - 3s 284ms/step - loss: 0.4459 - top1: 0.9111 - top3: 0.9944 - top5: 1.0000 - val_loss: 17.2528 - val_top1: 0.2106 - val_top3: 0.4400 - val_top5: 0.5744\n",
+            "Epoch 37/100\n",
+            "12/12 [==============================] - 3s 280ms/step - loss: 0.6986 - top1: 0.8917 - top3: 0.9694 - top5: 0.9917 - val_loss: 19.7720 - val_top1: 0.1856 - val_top3: 0.3856 - val_top5: 0.5361\n",
+            "Epoch 38/100\n",
+            "12/12 [==============================] - 4s 303ms/step - loss: 0.7881 - top1: 0.8806 - top3: 0.9667 - top5: 0.9944 - val_loss: 19.8641 - val_top1: 0.1828 - val_top3: 0.3900 - val_top5: 0.5411\n",
+            "Epoch 39/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 0.7422 - top1: 0.9028 - top3: 0.9917 - top5: 1.0000 - val_loss: 21.8557 - val_top1: 0.2033 - val_top3: 0.4039 - val_top5: 0.5478\n",
+            "Epoch 40/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 0.4641 - top1: 0.9167 - top3: 0.9861 - top5: 0.9944 - val_loss: 35.8059 - val_top1: 0.1756 - val_top3: 0.3861 - val_top5: 0.5344\n",
+            "Epoch 41/100\n",
+            "12/12 [==============================] - 3s 259ms/step - loss: 0.6262 - top1: 0.9139 - top3: 0.9917 - top5: 0.9972 - val_loss: 33.7794 - val_top1: 0.1694 - val_top3: 0.3789 - val_top5: 0.5272\n",
+            "Epoch 42/100\n",
+            "12/12 [==============================] - 3s 285ms/step - loss: 0.9856 - top1: 0.8722 - top3: 0.9806 - top5: 0.9972 - val_loss: 27.1626 - val_top1: 0.1917 - val_top3: 0.4022 - val_top5: 0.5489\n",
+            "Epoch 43/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 0.8370 - top1: 0.8806 - top3: 0.9861 - top5: 0.9972 - val_loss: 25.7161 - val_top1: 0.1989 - val_top3: 0.3950 - val_top5: 0.5389\n",
+            "Epoch 44/100\n",
+            "12/12 [==============================] - 3s 275ms/step - loss: 0.5446 - top1: 0.9222 - top3: 0.9889 - top5: 0.9944 - val_loss: 22.7228 - val_top1: 0.1939 - val_top3: 0.4056 - val_top5: 0.5428\n",
+            "Epoch 45/100\n",
+            "12/12 [==============================] - 3s 276ms/step - loss: 0.7556 - top1: 0.9083 - top3: 0.9750 - top5: 0.9917 - val_loss: 21.4619 - val_top1: 0.2200 - val_top3: 0.4344 - val_top5: 0.5733\n",
+            "Epoch 46/100\n",
+            "12/12 [==============================] - 3s 278ms/step - loss: 0.5049 - top1: 0.9194 - top3: 0.9861 - top5: 0.9944 - val_loss: 24.7316 - val_top1: 0.1967 - val_top3: 0.4139 - val_top5: 0.5594\n",
+            "Epoch 47/100\n",
+            "12/12 [==============================] - 3s 259ms/step - loss: 0.6334 - top1: 0.9056 - top3: 0.9861 - top5: 0.9889 - val_loss: 24.1323 - val_top1: 0.1994 - val_top3: 0.4156 - val_top5: 0.5489\n",
+            "Epoch 48/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 0.7063 - top1: 0.9000 - top3: 0.9889 - top5: 0.9972 - val_loss: 22.2957 - val_top1: 0.1883 - val_top3: 0.3944 - val_top5: 0.5339\n",
+            "Epoch 49/100\n",
+            "12/12 [==============================] - 3s 271ms/step - loss: 0.9505 - top1: 0.8806 - top3: 0.9722 - top5: 0.9833 - val_loss: 18.4470 - val_top1: 0.1911 - val_top3: 0.3928 - val_top5: 0.5433\n",
+            "Epoch 50/100\n",
+            "12/12 [==============================] - 3s 267ms/step - loss: 0.9428 - top1: 0.8750 - top3: 0.9889 - top5: 0.9944 - val_loss: 25.4363 - val_top1: 0.2006 - val_top3: 0.4250 - val_top5: 0.5544\n",
+            "Epoch 51/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 0.9024 - top1: 0.8944 - top3: 0.9833 - top5: 0.9917 - val_loss: 27.8479 - val_top1: 0.2094 - val_top3: 0.4161 - val_top5: 0.5583\n",
+            "Epoch 52/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 0.8420 - top1: 0.8944 - top3: 0.9806 - top5: 0.9944 - val_loss: 25.7100 - val_top1: 0.2111 - val_top3: 0.4206 - val_top5: 0.5606\n",
+            "Epoch 53/100\n",
+            "12/12 [==============================] - 3s 264ms/step - loss: 0.9583 - top1: 0.8917 - top3: 0.9806 - top5: 0.9889 - val_loss: 23.2428 - val_top1: 0.2000 - val_top3: 0.4178 - val_top5: 0.5589\n",
+            "Epoch 54/100\n",
+            "12/12 [==============================] - 4s 313ms/step - loss: 0.9405 - top1: 0.8889 - top3: 0.9778 - top5: 0.9889 - val_loss: 22.2511 - val_top1: 0.2128 - val_top3: 0.4189 - val_top5: 0.5744\n",
+            "Epoch 55/100\n",
+            "12/12 [==============================] - 4s 348ms/step - loss: 1.0060 - top1: 0.8889 - top3: 0.9778 - top5: 0.9917 - val_loss: 21.1396 - val_top1: 0.2167 - val_top3: 0.4267 - val_top5: 0.5694\n",
+            "Epoch 56/100\n",
+            "12/12 [==============================] - 3s 271ms/step - loss: 0.8457 - top1: 0.8917 - top3: 0.9833 - top5: 0.9944 - val_loss: 27.9648 - val_top1: 0.1894 - val_top3: 0.3956 - val_top5: 0.5356\n",
+            "Epoch 57/100\n",
+            "12/12 [==============================] - 4s 370ms/step - loss: 0.7562 - top1: 0.9000 - top3: 0.9778 - top5: 0.9917 - val_loss: 29.6968 - val_top1: 0.1728 - val_top3: 0.3667 - val_top5: 0.5089\n",
+            "Epoch 58/100\n",
+            "12/12 [==============================] - 4s 344ms/step - loss: 0.6295 - top1: 0.9472 - top3: 0.9861 - top5: 0.9972 - val_loss: 30.9653 - val_top1: 0.1767 - val_top3: 0.3800 - val_top5: 0.5150\n",
+            "Epoch 59/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 0.8466 - top1: 0.9167 - top3: 0.9806 - top5: 0.9944 - val_loss: 33.5612 - val_top1: 0.1844 - val_top3: 0.3822 - val_top5: 0.5072\n",
+            "Epoch 60/100\n",
+            "12/12 [==============================] - 3s 288ms/step - loss: 0.8133 - top1: 0.8972 - top3: 0.9806 - top5: 1.0000 - val_loss: 30.8591 - val_top1: 0.1594 - val_top3: 0.3578 - val_top5: 0.5067\n",
+            "Epoch 61/100\n",
+            "12/12 [==============================] - 3s 266ms/step - loss: 0.7173 - top1: 0.9139 - top3: 0.9917 - top5: 0.9972 - val_loss: 23.8336 - val_top1: 0.1778 - val_top3: 0.3967 - val_top5: 0.5422\n",
+            "Epoch 62/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 1.1505 - top1: 0.8694 - top3: 0.9778 - top5: 0.9917 - val_loss: 26.8529 - val_top1: 0.1533 - val_top3: 0.3483 - val_top5: 0.5061\n",
+            "Epoch 63/100\n",
+            "12/12 [==============================] - 4s 350ms/step - loss: 1.0769 - top1: 0.9000 - top3: 0.9750 - top5: 0.9889 - val_loss: 27.4314 - val_top1: 0.2011 - val_top3: 0.3956 - val_top5: 0.5339\n",
+            "Epoch 64/100\n",
+            "12/12 [==============================] - 3s 257ms/step - loss: 0.5940 - top1: 0.9167 - top3: 0.9889 - top5: 1.0000 - val_loss: 30.1204 - val_top1: 0.2022 - val_top3: 0.3894 - val_top5: 0.5317\n",
+            "Epoch 65/100\n",
+            "12/12 [==============================] - 3s 279ms/step - loss: 0.6377 - top1: 0.9028 - top3: 0.9889 - top5: 0.9972 - val_loss: 30.7100 - val_top1: 0.1883 - val_top3: 0.3789 - val_top5: 0.5122\n",
+            "Epoch 66/100\n",
+            "12/12 [==============================] - 3s 259ms/step - loss: 0.7521 - top1: 0.9000 - top3: 0.9861 - top5: 0.9917 - val_loss: 29.6171 - val_top1: 0.1928 - val_top3: 0.4200 - val_top5: 0.5456\n",
+            "Epoch 67/100\n",
+            "12/12 [==============================] - 3s 261ms/step - loss: 0.3921 - top1: 0.9417 - top3: 0.9972 - top5: 1.0000 - val_loss: 33.0342 - val_top1: 0.1950 - val_top3: 0.4033 - val_top5: 0.5444\n",
+            "Epoch 68/100\n",
+            "12/12 [==============================] - 4s 370ms/step - loss: 0.3799 - top1: 0.9333 - top3: 0.9972 - top5: 1.0000 - val_loss: 21.0015 - val_top1: 0.1956 - val_top3: 0.3972 - val_top5: 0.5572\n",
+            "Epoch 69/100\n",
+            "12/12 [==============================] - 5s 388ms/step - loss: 0.3579 - top1: 0.9528 - top3: 0.9972 - top5: 1.0000 - val_loss: 24.9785 - val_top1: 0.1833 - val_top3: 0.3983 - val_top5: 0.5411\n",
+            "Epoch 70/100\n",
+            "12/12 [==============================] - 4s 350ms/step - loss: 0.5754 - top1: 0.9417 - top3: 0.9889 - top5: 0.9944 - val_loss: 32.8437 - val_top1: 0.1883 - val_top3: 0.4056 - val_top5: 0.5383\n",
+            "Epoch 71/100\n",
+            "12/12 [==============================] - 5s 419ms/step - loss: 0.6394 - top1: 0.9083 - top3: 0.9889 - top5: 1.0000 - val_loss: 28.1090 - val_top1: 0.1972 - val_top3: 0.4033 - val_top5: 0.5317\n",
+            "Epoch 72/100\n",
+            "12/12 [==============================] - 4s 338ms/step - loss: 0.8495 - top1: 0.9083 - top3: 0.9833 - top5: 0.9944 - val_loss: 33.4919 - val_top1: 0.1828 - val_top3: 0.3983 - val_top5: 0.5378\n",
+            "Epoch 73/100\n",
+            "12/12 [==============================] - 3s 263ms/step - loss: 0.5934 - top1: 0.9222 - top3: 0.9889 - top5: 0.9972 - val_loss: 36.6565 - val_top1: 0.2128 - val_top3: 0.3878 - val_top5: 0.5233\n",
+            "Epoch 74/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 0.6069 - top1: 0.9306 - top3: 0.9833 - top5: 0.9972 - val_loss: 46.0970 - val_top1: 0.1750 - val_top3: 0.3539 - val_top5: 0.4933\n",
+            "Epoch 75/100\n",
+            "12/12 [==============================] - 3s 264ms/step - loss: 0.6728 - top1: 0.9306 - top3: 0.9889 - top5: 1.0000 - val_loss: 36.7551 - val_top1: 0.1861 - val_top3: 0.4011 - val_top5: 0.5278\n",
+            "Epoch 76/100\n",
+            "12/12 [==============================] - 3s 277ms/step - loss: 0.4724 - top1: 0.9278 - top3: 0.9972 - top5: 1.0000 - val_loss: 31.9093 - val_top1: 0.1672 - val_top3: 0.3889 - val_top5: 0.5311\n",
+            "Epoch 77/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 0.7074 - top1: 0.9278 - top3: 0.9889 - top5: 0.9972 - val_loss: 29.2835 - val_top1: 0.2039 - val_top3: 0.3994 - val_top5: 0.5450\n",
+            "Epoch 78/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 0.6212 - top1: 0.9139 - top3: 0.9833 - top5: 0.9917 - val_loss: 29.7789 - val_top1: 0.2117 - val_top3: 0.4039 - val_top5: 0.5439\n",
+            "Epoch 79/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 0.7132 - top1: 0.9111 - top3: 0.9833 - top5: 0.9944 - val_loss: 34.1109 - val_top1: 0.2189 - val_top3: 0.4022 - val_top5: 0.5278\n",
+            "Epoch 80/100\n",
+            "12/12 [==============================] - 3s 276ms/step - loss: 0.3696 - top1: 0.9389 - top3: 0.9972 - top5: 1.0000 - val_loss: 35.3221 - val_top1: 0.2128 - val_top3: 0.4028 - val_top5: 0.5400\n",
+            "Epoch 81/100\n",
+            "12/12 [==============================] - 3s 279ms/step - loss: 0.4482 - top1: 0.9389 - top3: 0.9889 - top5: 0.9944 - val_loss: 30.8614 - val_top1: 0.2011 - val_top3: 0.4094 - val_top5: 0.5517\n",
+            "Epoch 82/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 0.4886 - top1: 0.9556 - top3: 1.0000 - top5: 1.0000 - val_loss: 29.4703 - val_top1: 0.2189 - val_top3: 0.4139 - val_top5: 0.5556\n",
+            "Epoch 83/100\n",
+            "12/12 [==============================] - 3s 280ms/step - loss: 0.8582 - top1: 0.9306 - top3: 0.9833 - top5: 0.9917 - val_loss: 28.8590 - val_top1: 0.1744 - val_top3: 0.3800 - val_top5: 0.5406\n",
+            "Epoch 84/100\n",
+            "12/12 [==============================] - 3s 275ms/step - loss: 0.3760 - top1: 0.9389 - top3: 0.9944 - top5: 1.0000 - val_loss: 27.9458 - val_top1: 0.2061 - val_top3: 0.4228 - val_top5: 0.5767\n",
+            "Epoch 85/100\n",
+            "12/12 [==============================] - 3s 266ms/step - loss: 0.6781 - top1: 0.9389 - top3: 0.9944 - top5: 0.9972 - val_loss: 40.9894 - val_top1: 0.1772 - val_top3: 0.3894 - val_top5: 0.5450\n",
+            "Epoch 86/100\n",
+            "12/12 [==============================] - 3s 286ms/step - loss: 0.4348 - top1: 0.9306 - top3: 0.9944 - top5: 1.0000 - val_loss: 29.0168 - val_top1: 0.2006 - val_top3: 0.4117 - val_top5: 0.5456\n",
+            "Epoch 87/100\n",
+            "12/12 [==============================] - 3s 262ms/step - loss: 0.5136 - top1: 0.9444 - top3: 0.9917 - top5: 0.9972 - val_loss: 26.6945 - val_top1: 0.2139 - val_top3: 0.4161 - val_top5: 0.5622\n",
+            "Epoch 88/100\n",
+            "12/12 [==============================] - 3s 278ms/step - loss: 0.6469 - top1: 0.9167 - top3: 0.9917 - top5: 0.9944 - val_loss: 26.5667 - val_top1: 0.1989 - val_top3: 0.4039 - val_top5: 0.5633\n",
+            "Epoch 89/100\n",
+            "12/12 [==============================] - 3s 264ms/step - loss: 0.4760 - top1: 0.9278 - top3: 1.0000 - top5: 1.0000 - val_loss: 28.4160 - val_top1: 0.1961 - val_top3: 0.3922 - val_top5: 0.5433\n",
+            "Epoch 90/100\n",
+            "12/12 [==============================] - 3s 282ms/step - loss: 0.3644 - top1: 0.9472 - top3: 0.9944 - top5: 0.9972 - val_loss: 26.3260 - val_top1: 0.2022 - val_top3: 0.4083 - val_top5: 0.5594\n",
+            "Epoch 91/100\n",
+            "12/12 [==============================] - 4s 321ms/step - loss: 0.4228 - top1: 0.9583 - top3: 0.9944 - top5: 0.9972 - val_loss: 30.1662 - val_top1: 0.1794 - val_top3: 0.3956 - val_top5: 0.5444\n",
+            "Epoch 92/100\n",
+            "12/12 [==============================] - 3s 265ms/step - loss: 0.4653 - top1: 0.9361 - top3: 0.9917 - top5: 0.9944 - val_loss: 31.0508 - val_top1: 0.1883 - val_top3: 0.4067 - val_top5: 0.5378\n",
+            "Epoch 93/100\n",
+            "12/12 [==============================] - 4s 329ms/step - loss: 0.4649 - top1: 0.9333 - top3: 0.9889 - top5: 0.9944 - val_loss: 27.2362 - val_top1: 0.2050 - val_top3: 0.4439 - val_top5: 0.5739\n",
+            "Epoch 94/100\n",
+            "12/12 [==============================] - 3s 271ms/step - loss: 0.8348 - top1: 0.9111 - top3: 0.9806 - top5: 0.9889 - val_loss: 21.8538 - val_top1: 0.2117 - val_top3: 0.4550 - val_top5: 0.5978\n",
+            "Epoch 95/100\n",
+            "12/12 [==============================] - 3s 281ms/step - loss: 0.4281 - top1: 0.9500 - top3: 0.9944 - top5: 1.0000 - val_loss: 23.9799 - val_top1: 0.2172 - val_top3: 0.4439 - val_top5: 0.5839\n",
+            "Epoch 96/100\n",
+            "12/12 [==============================] - 3s 258ms/step - loss: 0.2019 - top1: 0.9556 - top3: 0.9972 - top5: 1.0000 - val_loss: 25.4205 - val_top1: 0.2144 - val_top3: 0.4411 - val_top5: 0.5872\n",
+            "Epoch 97/100\n",
+            "12/12 [==============================] - 3s 279ms/step - loss: 0.3994 - top1: 0.9500 - top3: 0.9889 - top5: 0.9972 - val_loss: 26.6242 - val_top1: 0.2211 - val_top3: 0.4456 - val_top5: 0.5767\n",
+            "Epoch 98/100\n",
+            "12/12 [==============================] - 4s 354ms/step - loss: 0.4678 - top1: 0.9333 - top3: 0.9972 - top5: 1.0000 - val_loss: 32.3052 - val_top1: 0.2011 - val_top3: 0.4217 - val_top5: 0.5661\n",
+            "Epoch 99/100\n",
+            "12/12 [==============================] - 3s 260ms/step - loss: 0.4220 - top1: 0.9472 - top3: 0.9972 - top5: 1.0000 - val_loss: 32.3346 - val_top1: 0.1806 - val_top3: 0.4178 - val_top5: 0.5650\n",
+            "Epoch 100/100\n",
+            "12/12 [==============================] - 3s 278ms/step - loss: 0.4012 - top1: 0.9556 - top3: 0.9917 - top5: 0.9944 - val_loss: 22.5723 - val_top1: 0.2083 - val_top3: 0.4283 - val_top5: 0.5794\n"
+          ]
+        }
+      ],
+      "source": [
+        "# Supervised finetuning of the pretrained encoder\n",
+        "finetuning_model = keras.Sequential(\n",
+        "    [\n",
+        "        layers.Input(shape=IMG_SHAPE),\n",
+        "        get_augmenter(**classification_augmentation),\n",
+        "        pretraining_model.encoder,\n",
+        "        layers.Dense(len(CLASSES)),\n",
+        "    ],\n",
+        "    name=\"finetuning_model\",\n",
+        ")\n",
+        "finetuning_model.compile(\n",
+        "    optimizer=keras.optimizers.Adam(),\n",
+        "    loss=keras.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
+        "    metrics=[\n",
+        "        keras.metrics.SparseCategoricalAccuracy(name=\"top1\"),\n",
+        "        tf.keras.metrics.SparseTopKCategoricalAccuracy(k=3, name=\"top3\"),\n",
+        "        tf.keras.metrics.SparseTopKCategoricalAccuracy(k=5, name=\"top5\")\n",
+        "    ],\n",
+        ")\n",
+        "\n",
+        "finetuning_history = finetuning_model.fit(\n",
+        "    labeled_train_dataset.unbatch().batch(30), epochs=100, validation_data=test_dataset\n",
+        ")\n"
+      ]
+    }
+  ],
+  "metadata": {
+    "accelerator": "GPU",
+    "colab": {
+      "authorship_tag": "ABX9TyMry1qjHt3ff4mTAl8MEvzL",
+      "provenance": []
+    },
+    "gpuClass": "standard",
+    "kernelspec": {
+      "display_name": "Python 3",
+      "language": "python",
+      "name": "python3"
+    },
+    "language_info": {
+      "name": "python",
+      "version": "3.10.9 (main, Dec  6 2022, 18:44:57) [GCC 11.3.0]"
+    },
+    "vscode": {
+      "interpreter": {
+        "hash": "7815da974dc6b70eef1620f3b951c28a97a88e2199d769663112ac895c9b37ee"
+      }
+    },
+    "widgets": {
+      "application/vnd.jupyter.widget-state+json": {
+        "022331d5cd3e43c58fac533918517d47": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": "hidden",
+            "width": null
+          }
+        },
+        "0b9a6a3607be486cb44deab621476db3": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "0bbc472ae27441dfb3ff3e9b58d56e71": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_d83d0f7418d04166975f056bb4c71f43",
+            "placeholder": "​",
+            "style": "IPY_MODEL_23ea228dcb214d6185701e9c7a33f9d7",
+            "value": "Shuffling ~/tensorflow_datasets/projsemisup/1.1.0.incompleteUMKSYI/projsemisup-unlabelled.tfrecord*...:  97%"
+          }
+        },
+        "0bf4eaad630549a6b489172a93b8f2ee": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_db025fec91104f209b62774850ad33bb",
+            "placeholder": "​",
+            "style": "IPY_MODEL_eb50e6dc7ec945a9940e6f69e608a4fd",
+            "value": " 1137987320/1137987320 [00:00&lt;00:00, 15048536683.58 MiB/s]"
+          }
+        },
+        "0ccd60e561e94ceebd7fd9fb209f669b": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "16a63953514a422f9a7f8d385f115308": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "182e7f0633264ae0aa5c97ea14f6b7df": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "1bf1a0aa3dde41ceb1267c0fc9754986": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "1c39c0bac3664b8c9eb97ba37d29bbf2": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "1f58412868b448808175d63bcfb447da": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_0b9a6a3607be486cb44deab621476db3",
+            "placeholder": "​",
+            "style": "IPY_MODEL_625edd164dcf4932a36dec0851a79c89",
+            "value": " 3/3 [02:52&lt;00:00, 74.60s/ splits]"
+          }
+        },
+        "208d7f6179e347fe943a2ef9fb56d9a9": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_c80e7fc56adb400e907cd62b150b978d",
+            "placeholder": "​",
+            "style": "IPY_MODEL_ad2cd0ee331141fd8252cd600ccc5fff",
+            "value": "Generating splits...: 100%"
+          }
+        },
+        "22f5687d1c514a26b1cff927b3243955": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": "20px"
+          }
+        },
+        "23ea228dcb214d6185701e9c7a33f9d7": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "266c3c3c39d147b2b4c95af49fb36f42": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "282fb71f47174efc997f8d259b785d61": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "2c99fb33ca8a43f5a7efc29a975c2bd6": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": "hidden",
+            "width": null
+          }
+        },
+        "2e58e17884b14d4e935bcf11fb01acf6": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_a1b346004e3c48a890b99bbfff63b83f",
+            "placeholder": "​",
+            "style": "IPY_MODEL_5f1c14872085484fa1e40e9f75bb1f40",
+            "value": "Generating unlabelled examples...: "
+          }
+        },
+        "30093792a4584cea92bd8fe8e2c41687": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "3273f7f1d44b4741af5873ec8e411381": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_973841b307dd46b59175ca8545169581",
+              "IPY_MODEL_51b36f07ba624efa93a48ee34cf906c5",
+              "IPY_MODEL_d014273e04c7430d99b746883d71a85b"
+            ],
+            "layout": "IPY_MODEL_ae66b99545a84856af3793aa59273def"
+          }
+        },
+        "34415bcdaaa54e089630854f1df6114c": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "344ab71ee68b4af28cb2948cc3e2ef15": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "37beb756d55345e39d41c42bd5f4dc16": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_f3cbe23c356d43679cc4ff486fca3545",
+            "placeholder": "​",
+            "style": "IPY_MODEL_bf7519c8ab95488fa5f79cb35c8db1e1",
+            "value": "Shuffling ~/tensorflow_datasets/projsemisup/1.1.0.incompleteUMKSYI/projsemisup-train.tfrecord*...:   0%"
+          }
+        },
+        "3b3435d568f344e5a284f663743f0690": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "3c1c6edfd6994df38afb1ab8d21a757a": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "3fdb7263acbe430ba27d42f46fbffdc0": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "success",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_22f5687d1c514a26b1cff927b3243955",
+            "max": 1,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_3c1c6edfd6994df38afb1ab8d21a757a",
+            "value": 0
+          }
+        },
+        "40b74195618544d5a2c18e9307d8997f": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": "hidden",
+            "width": null
+          }
+        },
+        "42f427596cf948268ec0ae6a784229d1": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "45bccbfc63c14b308bbb6ad4c384ce74": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": "hidden",
+            "width": null
+          }
+        },
+        "47ae3e436a3548618f07a1a7a07e168e": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "49f5c7f0b5cc4e9999c6d905fdd74c1e": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": "hidden",
+            "width": null
+          }
+        },
+        "4f13be67e1c343c6adec61bc58ff90f1": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "4fb533e568294997a4a4f359416c1b5b": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_a77dac4501f147f2bad2378c4300dafd",
+            "placeholder": "​",
+            "style": "IPY_MODEL_78f1b7c6540d4687ade023641b47c5a4",
+            "value": "Extraction completed...: "
+          }
+        },
+        "51b36f07ba624efa93a48ee34cf906c5": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "info",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_94ff6626e3b84758a2ffb12a9973ded8",
+            "max": 1,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_edde9ce2a45344d6babb797dce13d486",
+            "value": 1
+          }
+        },
+        "59f1ab66e02c4ae5a6a840a1da0d1e8e": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "5d4921be4b2244db90796202da5faf04": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "5f1c14872085484fa1e40e9f75bb1f40": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "5fc847f952624fba849142a2d01b4500": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "625edd164dcf4932a36dec0851a79c89": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "63af397c06c242b2a6d9d4acd0232cb3": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_e86e376d77c24506a3541e0ab9da1c32",
+              "IPY_MODEL_f8774da5294944598574dfd2a8a35610",
+              "IPY_MODEL_a6b8c7ab593148768205ad47b7fea634"
+            ],
+            "layout": "IPY_MODEL_b00733d89e474697873f02f7e60fa7d4"
+          }
+        },
+        "65053c140cfd402eb64fdbc1cc32cfc3": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "info",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_69426376e67948f492cd8deded7efe2e",
+            "max": 1,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_f71989288503489b847e8f629658798b",
+            "value": 1
+          }
+        },
+        "674c7bed061442718b1520380e88d522": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_266c3c3c39d147b2b4c95af49fb36f42",
+            "placeholder": "​",
+            "style": "IPY_MODEL_344ab71ee68b4af28cb2948cc3e2ef15",
+            "value": " 0/360 [00:00&lt;?, ? examples/s]"
+          }
+        },
+        "69426376e67948f492cd8deded7efe2e": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": "20px"
+          }
+        },
+        "69c4ff515c4e4b1f9f7c313374b3aa34": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "6b1184b6610a4660a8d6e3d8efa59bc8": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "6bdb69e0423b4df68ae9408a01ab02cc": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_42f427596cf948268ec0ae6a784229d1",
+            "placeholder": "​",
+            "style": "IPY_MODEL_59f1ab66e02c4ae5a6a840a1da0d1e8e",
+            "value": " 20694/? [02:23&lt;00:00, 148.25 examples/s]"
+          }
+        },
+        "6d19543181544ddbac41e1357d448d34": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "6fc14bc422744eb29e13a1a0f31952e2": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": "20px"
+          }
+        },
+        "71d8a3f083ef44bfbfa64f2bbb10c1f0": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "72dfaac67f24431689781951c59d7feb": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "73d89e4492644d1a9e412b72d72a14de": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_a299ad4d23e74945824a9deecedfc76f",
+            "max": 1800,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_eababa6c94554cd484df6dd55dc994db",
+            "value": 1800
+          }
+        },
+        "74bfd61016d24ea7b677354077f04379": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "success",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_6fc14bc422744eb29e13a1a0f31952e2",
+            "max": 1,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_6d19543181544ddbac41e1357d448d34",
+            "value": 1
+          }
+        },
+        "78f1b7c6540d4687ade023641b47c5a4": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "7ed215974c6b4311bb34d4ba721f790b": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_a274c06ad80347ec9a5077cb6ce5eba1",
+              "IPY_MODEL_73d89e4492644d1a9e412b72d72a14de",
+              "IPY_MODEL_f499165e66704145bfb38bf241613b87"
+            ],
+            "layout": "IPY_MODEL_40b74195618544d5a2c18e9307d8997f"
+          }
+        },
+        "7f1567f89ac1487eb70960c65b6816ec": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_5fc847f952624fba849142a2d01b4500",
+            "max": 20697,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_1c39c0bac3664b8c9eb97ba37d29bbf2",
+            "value": 20697
+          }
+        },
+        "80bf30467e6942bbb164bf90d6aca854": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_282fb71f47174efc997f8d259b785d61",
+            "max": 360,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_a8796bad61a34df89c1154a40bd814e9",
+            "value": 360
+          }
+        },
+        "811af7ec09af455785951ea0137f2e3a": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "88fdfea6ec91465ba7cda5e9547599d6": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_b17b9124cc884e2ea22011cd5e5b3636",
+            "placeholder": "​",
+            "style": "IPY_MODEL_bb1ca1dec70249f182852c32a06bb4c4",
+            "value": " 0/0 [00:00&lt;?, ? file/s]"
+          }
+        },
+        "8bd20837baf947099b4fc9d6cb3e871e": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "92d5ca69eb10464cadd2ad8c6b92b5f1": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_bf108424901946eb8b189fc726300472",
+            "placeholder": "​",
+            "style": "IPY_MODEL_0ccd60e561e94ceebd7fd9fb209f669b",
+            "value": "Dl Completed...: 100%"
+          }
+        },
+        "94ff6626e3b84758a2ffb12a9973ded8": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": "20px"
+          }
+        },
+        "96b5945b44ff461387eed16bd3306a20": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "973841b307dd46b59175ca8545169581": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_3b3435d568f344e5a284f663743f0690",
+            "placeholder": "​",
+            "style": "IPY_MODEL_5d4921be4b2244db90796202da5faf04",
+            "value": "Generating train examples...: "
+          }
+        },
+        "9901ea583eb54f5494dabb99cfc99483": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_71d8a3f083ef44bfbfa64f2bbb10c1f0",
+            "max": 3,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_6b1184b6610a4660a8d6e3d8efa59bc8",
+            "value": 3
+          }
+        },
+        "9f4b3d82a9bf41d49154d125ea0d548a": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "a1b346004e3c48a890b99bbfff63b83f": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "a274c06ad80347ec9a5077cb6ce5eba1": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_16a63953514a422f9a7f8d385f115308",
+            "placeholder": "​",
+            "style": "IPY_MODEL_96b5945b44ff461387eed16bd3306a20",
+            "value": "Shuffling ~/tensorflow_datasets/projsemisup/1.1.0.incompleteUMKSYI/projsemisup-test.tfrecord*...:   0%"
+          }
+        },
+        "a299ad4d23e74945824a9deecedfc76f": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "a3a0c812bbc34c3c942d5372cc078b18": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": "20px"
+          }
+        },
+        "a6b8c7ab593148768205ad47b7fea634": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_e0c480f5b7254651a5af7e04410d9d43",
+            "placeholder": "​",
+            "style": "IPY_MODEL_69c4ff515c4e4b1f9f7c313374b3aa34",
+            "value": " 1796/? [00:21&lt;00:00, 100.58 examples/s]"
+          }
+        },
+        "a77dac4501f147f2bad2378c4300dafd": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "a8796bad61a34df89c1154a40bd814e9": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "ad2cd0ee331141fd8252cd600ccc5fff": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "ae66b99545a84856af3793aa59273def": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": "hidden",
+            "width": null
+          }
+        },
+        "b00733d89e474697873f02f7e60fa7d4": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": "hidden",
+            "width": null
+          }
+        },
+        "b02f0e154835453fa47b161ec4cefcb1": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": "20px"
+          }
+        },
+        "b073c9c0a0b54a12995ecefe0136e950": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "b17b9124cc884e2ea22011cd5e5b3636": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "b5c568b161f24e418d38ecc1821267ac": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_2e58e17884b14d4e935bcf11fb01acf6",
+              "IPY_MODEL_65053c140cfd402eb64fdbc1cc32cfc3",
+              "IPY_MODEL_6bdb69e0423b4df68ae9408a01ab02cc"
+            ],
+            "layout": "IPY_MODEL_2c99fb33ca8a43f5a7efc29a975c2bd6"
+          }
+        },
+        "b7e79680e03b46e1b33510af665107fe": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_208d7f6179e347fe943a2ef9fb56d9a9",
+              "IPY_MODEL_9901ea583eb54f5494dabb99cfc99483",
+              "IPY_MODEL_1f58412868b448808175d63bcfb447da"
+            ],
+            "layout": "IPY_MODEL_49f5c7f0b5cc4e9999c6d905fdd74c1e"
+          }
+        },
+        "b7f82bacd58940439544261b5e942048": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "b982bb2c7d784572aee1029aa217b969": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "bb1ca1dec70249f182852c32a06bb4c4": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "bf108424901946eb8b189fc726300472": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "bf7519c8ab95488fa5f79cb35c8db1e1": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "c1957bebe6f64eb0915d6e7a26157775": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_4fb533e568294997a4a4f359416c1b5b",
+              "IPY_MODEL_3fdb7263acbe430ba27d42f46fbffdc0",
+              "IPY_MODEL_88fdfea6ec91465ba7cda5e9547599d6"
+            ],
+            "layout": "IPY_MODEL_1bf1a0aa3dde41ceb1267c0fc9754986"
+          }
+        },
+        "c3b6e4177f2e4b029b1666cf4c1a62a6": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "c80e7fc56adb400e907cd62b150b978d": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "cf39226fb80c417c990d43244daff8b4": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "success",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_a3a0c812bbc34c3c942d5372cc078b18",
+            "max": 1,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_ed3859fe5b6f4c93855f0ef500d8bc91",
+            "value": 1
+          }
+        },
+        "d014273e04c7430d99b746883d71a85b": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_47ae3e436a3548618f07a1a7a07e168e",
+            "placeholder": "​",
+            "style": "IPY_MODEL_182e7f0633264ae0aa5c97ea14f6b7df",
+            "value": " 355/? [00:04&lt;00:00, 87.06 examples/s]"
+          }
+        },
+        "d18b6dd9bcd145ab9b4dcbd1bf217a39": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "d83d0f7418d04166975f056bb4c71f43": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "db025fec91104f209b62774850ad33bb": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "dd5eb0967fb6431697bd5901e1bb4fba": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_9f4b3d82a9bf41d49154d125ea0d548a",
+            "placeholder": "​",
+            "style": "IPY_MODEL_30093792a4584cea92bd8fe8e2c41687",
+            "value": " 19982/20697 [00:02&lt;00:00, 6871.05 examples/s]"
+          }
+        },
+        "de6a8641f28d48d0b773ac5ad800eec6": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_c3b6e4177f2e4b029b1666cf4c1a62a6",
+            "placeholder": "​",
+            "style": "IPY_MODEL_d18b6dd9bcd145ab9b4dcbd1bf217a39",
+            "value": "Dl Size...: 100%"
+          }
+        },
+        "e0c480f5b7254651a5af7e04410d9d43": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "e86e376d77c24506a3541e0ab9da1c32": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_8bd20837baf947099b4fc9d6cb3e871e",
+            "placeholder": "​",
+            "style": "IPY_MODEL_34415bcdaaa54e089630854f1df6114c",
+            "value": "Generating test examples...: "
+          }
+        },
+        "e9eb28d864f34cf8bba737909172eb2e": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_37beb756d55345e39d41c42bd5f4dc16",
+              "IPY_MODEL_80bf30467e6942bbb164bf90d6aca854",
+              "IPY_MODEL_674c7bed061442718b1520380e88d522"
+            ],
+            "layout": "IPY_MODEL_022331d5cd3e43c58fac533918517d47"
+          }
+        },
+        "eababa6c94554cd484df6dd55dc994db": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "eb50e6dc7ec945a9940e6f69e608a4fd": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "DescriptionStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "DescriptionStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "description_width": ""
+          }
+        },
+        "ed3859fe5b6f4c93855f0ef500d8bc91": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "edde9ce2a45344d6babb797dce13d486": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "f203872de5f94508a338390508651d0c": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_de6a8641f28d48d0b773ac5ad800eec6",
+              "IPY_MODEL_74bfd61016d24ea7b677354077f04379",
+              "IPY_MODEL_0bf4eaad630549a6b489172a93b8f2ee"
+            ],
+            "layout": "IPY_MODEL_4f13be67e1c343c6adec61bc58ff90f1"
+          }
+        },
+        "f2596bfef90f4a4095713b88b1ac16c1": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_0bbc472ae27441dfb3ff3e9b58d56e71",
+              "IPY_MODEL_7f1567f89ac1487eb70960c65b6816ec",
+              "IPY_MODEL_dd5eb0967fb6431697bd5901e1bb4fba"
+            ],
+            "layout": "IPY_MODEL_45bccbfc63c14b308bbb6ad4c384ce74"
+          }
+        },
+        "f3cbe23c356d43679cc4ff486fca3545": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        },
+        "f499165e66704145bfb38bf241613b87": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_ffd0376578b64811b35afde6c0cb63ab",
+            "placeholder": "​",
+            "style": "IPY_MODEL_72dfaac67f24431689781951c59d7feb",
+            "value": " 0/1800 [00:00&lt;?, ? examples/s]"
+          }
+        },
+        "f4d4b13251004dbfb7ba81f36289456b": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HBoxModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HBoxModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HBoxView",
+            "box_style": "",
+            "children": [
+              "IPY_MODEL_92d5ca69eb10464cadd2ad8c6b92b5f1",
+              "IPY_MODEL_cf39226fb80c417c990d43244daff8b4",
+              "IPY_MODEL_fc17720d2e434619bf5396de7467590c"
+            ],
+            "layout": "IPY_MODEL_b073c9c0a0b54a12995ecefe0136e950"
+          }
+        },
+        "f71989288503489b847e8f629658798b": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "ProgressStyleModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "ProgressStyleModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "StyleView",
+            "bar_color": null,
+            "description_width": ""
+          }
+        },
+        "f8774da5294944598574dfd2a8a35610": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "FloatProgressModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "FloatProgressModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "ProgressView",
+            "bar_style": "info",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_b02f0e154835453fa47b161ec4cefcb1",
+            "max": 1,
+            "min": 0,
+            "orientation": "horizontal",
+            "style": "IPY_MODEL_811af7ec09af455785951ea0137f2e3a",
+            "value": 1
+          }
+        },
+        "fc17720d2e434619bf5396de7467590c": {
+          "model_module": "@jupyter-widgets/controls",
+          "model_module_version": "1.5.0",
+          "model_name": "HTMLModel",
+          "state": {
+            "_dom_classes": [],
+            "_model_module": "@jupyter-widgets/controls",
+            "_model_module_version": "1.5.0",
+            "_model_name": "HTMLModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/controls",
+            "_view_module_version": "1.5.0",
+            "_view_name": "HTMLView",
+            "description": "",
+            "description_tooltip": null,
+            "layout": "IPY_MODEL_b7f82bacd58940439544261b5e942048",
+            "placeholder": "​",
+            "style": "IPY_MODEL_b982bb2c7d784572aee1029aa217b969",
+            "value": " 1/1 [00:00&lt;00:00,  9.53 url/s]"
+          }
+        },
+        "ffd0376578b64811b35afde6c0cb63ab": {
+          "model_module": "@jupyter-widgets/base",
+          "model_module_version": "1.2.0",
+          "model_name": "LayoutModel",
+          "state": {
+            "_model_module": "@jupyter-widgets/base",
+            "_model_module_version": "1.2.0",
+            "_model_name": "LayoutModel",
+            "_view_count": null,
+            "_view_module": "@jupyter-widgets/base",
+            "_view_module_version": "1.2.0",
+            "_view_name": "LayoutView",
+            "align_content": null,
+            "align_items": null,
+            "align_self": null,
+            "border": null,
+            "bottom": null,
+            "display": null,
+            "flex": null,
+            "flex_flow": null,
+            "grid_area": null,
+            "grid_auto_columns": null,
+            "grid_auto_flow": null,
+            "grid_auto_rows": null,
+            "grid_column": null,
+            "grid_gap": null,
+            "grid_row": null,
+            "grid_template_areas": null,
+            "grid_template_columns": null,
+            "grid_template_rows": null,
+            "height": null,
+            "justify_content": null,
+            "justify_items": null,
+            "left": null,
+            "margin": null,
+            "max_height": null,
+            "max_width": null,
+            "min_height": null,
+            "min_width": null,
+            "object_fit": null,
+            "object_position": null,
+            "order": null,
+            "overflow": null,
+            "overflow_x": null,
+            "overflow_y": null,
+            "padding": null,
+            "right": null,
+            "top": null,
+            "visibility": null,
+            "width": null
+          }
+        }
+      }
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}