jupyters/RUDOLPH-2_7B-fine-tune_rus.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "99fe2e79-81e4-428a-b977-2e830e780806",
   "metadata": {},
   "source": [
    "## Fine-tuning RUDOLPH on text QA"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5685796b-1a81-442c-8a8e-59b297791be2",
   "metadata": {},
   "source": [
    "RUDOLPH был обучен на 4-х задачах (text2image, image2text и text2text), причем предобучение на задачу генерации текста осуществлялось только в левых текстовых токенах. Использование данных задач для предобучения позволяет модели научиться сопоставлять тексты и картинки, а также кодировать лингвистическую информацию о естественном языке в генерируемые векторные представления.\n",
    "\n",
    "После пре-трейна, RUDOLPH может быть дообучен под более узкую задачу, для которой входом или выходом является текст, картинка либо их комбинация.\n",
    "\n",
    "В данном ноутбуке описан процесс дообучения RUDOLPH 2.7B под текстовую задачу ответа на вопросы по данному текстн (machine reading comprehension (MRC))."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d9bed231-463e-462e-956c-87efdd5c6e88",
   "metadata": {},
   "source": [
    "# Данные"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3844ce7b-e325-45b8-8f74-00fcfebcc614",
   "metadata": {},
   "source": [
    "Используем данные из набора [sberQuAD](https://huggingface.co/datasets/sberquad) - это датасет, содержащий текстовый абзац, вопрос по прочитанному тексту и фрагмент текста, являющийся ответом на заданный вопрос.\n",
    "\n",
    "Размер датасета:\n",
    "\n",
    "* train: 45328 \n",
    "* validation: 5036\n",
    "* test: 23936"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "544214ad-0cfe-43c9-98b4-9d2d7165e316",
   "metadata": {},
   "source": [
    "Cделаем импорт необходимых зависимостей"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "e70136e9-c67c-4474-8a8a-27ef812e72e3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: rudolph==0.0.1rc10 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (0.0.1rc10)\n",
      "Requirement already satisfied: torchvision in /home/user/conda/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (0.10.1+cu111)\n",
      "Requirement already satisfied: einops~=0.3.2 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (0.3.2)\n",
      "Requirement already satisfied: omegaconf>=2.0.0 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (2.1.1)\n",
      "Requirement already satisfied: youtokentome>=1.0.6 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (1.0.6)\n",
      "Requirement already satisfied: transformers~=4.10.2 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (4.10.3)\n",
      "Requirement already satisfied: matplotlib in /home/user/conda/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (3.5.1)\n",
      "Requirement already satisfied: more-itertools>=8.10.0 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (8.12.0)\n",
      "Requirement already satisfied: torch in /home/user/conda/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (1.9.1+cu111)\n",
      "Requirement already satisfied: rudalle==1.1.3 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (1.1.3)\n",
      "Requirement already satisfied: taming-transformers==0.0.1 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudolph==0.0.1rc10) (0.0.1)\n",
      "Requirement already satisfied: segmentation-models-pytorch==0.1.3 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudalle==1.1.3->rudolph==0.0.1rc10) (0.1.3)\n",
      "Requirement already satisfied: opencv-python==4.5.4.60 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudalle==1.1.3->rudolph==0.0.1rc10) (4.5.4.60)\n",
      "Requirement already satisfied: PyWavelets==1.1.1 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from rudalle==1.1.3->rudolph==0.0.1rc10) (1.1.1)\n",
      "Requirement already satisfied: pytorch-lightning>=1.0.8 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.5.10)\n",
      "Requirement already satisfied: tqdm in /home/user/conda/lib/python3.7/site-packages (from taming-transformers==0.0.1->rudolph==0.0.1rc10) (4.62.3)\n",
      "Requirement already satisfied: numpy in /home/user/conda/lib/python3.7/site-packages (from taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.21.5)\n",
      "Requirement already satisfied: efficientnet-pytorch==0.6.3 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from segmentation-models-pytorch==0.1.3->rudalle==1.1.3->rudolph==0.0.1rc10) (0.6.3)\n",
      "Requirement already satisfied: timm==0.3.2 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from segmentation-models-pytorch==0.1.3->rudalle==1.1.3->rudolph==0.0.1rc10) (0.3.2)\n",
      "Requirement already satisfied: pretrainedmodels==0.7.4 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from segmentation-models-pytorch==0.1.3->rudalle==1.1.3->rudolph==0.0.1rc10) (0.7.4)\n",
      "Requirement already satisfied: munch in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from pretrainedmodels==0.7.4->segmentation-models-pytorch==0.1.3->rudalle==1.1.3->rudolph==0.0.1rc10) (2.5.0)\n",
      "Requirement already satisfied: PyYAML>=5.1.0 in /home/user/conda/lib/python3.7/site-packages (from omegaconf>=2.0.0->rudolph==0.0.1rc10) (6.0)\n",
      "Requirement already satisfied: antlr4-python3-runtime==4.8 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from omegaconf>=2.0.0->rudolph==0.0.1rc10) (4.8)\n",
      "Requirement already satisfied: typing-extensions in /home/user/conda/lib/python3.7/site-packages (from torch->rudolph==0.0.1rc10) (4.0.1)\n",
      "Requirement already satisfied: pillow>=5.3.0 in /home/user/conda/lib/python3.7/site-packages (from torchvision->rudolph==0.0.1rc10) (9.0.0)\n",
      "Requirement already satisfied: filelock in /home/user/conda/lib/python3.7/site-packages (from transformers~=4.10.2->rudolph==0.0.1rc10) (3.4.2)\n",
      "Requirement already satisfied: tokenizers<0.11,>=0.10.1 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from transformers~=4.10.2->rudolph==0.0.1rc10) (0.10.3)\n",
      "Requirement already satisfied: packaging in /home/user/conda/lib/python3.7/site-packages (from transformers~=4.10.2->rudolph==0.0.1rc10) (21.3)\n",
      "Requirement already satisfied: importlib-metadata in /home/user/conda/lib/python3.7/site-packages (from transformers~=4.10.2->rudolph==0.0.1rc10) (4.10.1)\n",
      "Requirement already satisfied: sacremoses in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from transformers~=4.10.2->rudolph==0.0.1rc10) (0.0.47)\n",
      "Requirement already satisfied: regex!=2019.12.17 in /home/user/conda/lib/python3.7/site-packages (from transformers~=4.10.2->rudolph==0.0.1rc10) (2022.1.18)\n",
      "Requirement already satisfied: requests in /home/user/conda/lib/python3.7/site-packages (from transformers~=4.10.2->rudolph==0.0.1rc10) (2.27.1)\n",
      "Requirement already satisfied: huggingface-hub>=0.0.12 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from transformers~=4.10.2->rudolph==0.0.1rc10) (0.9.1)\n",
      "Requirement already satisfied: Click>=7.0 in /home/user/conda/lib/python3.7/site-packages (from youtokentome>=1.0.6->rudolph==0.0.1rc10) (8.0.3)\n",
      "Requirement already satisfied: cycler>=0.10 in /home/user/conda/lib/python3.7/site-packages (from matplotlib->rudolph==0.0.1rc10) (0.11.0)\n",
      "Requirement already satisfied: fonttools>=4.22.0 in /home/user/conda/lib/python3.7/site-packages (from matplotlib->rudolph==0.0.1rc10) (4.28.5)\n",
      "Requirement already satisfied: pyparsing>=2.2.1 in /home/user/conda/lib/python3.7/site-packages (from matplotlib->rudolph==0.0.1rc10) (3.0.7)\n",
      "Requirement already satisfied: python-dateutil>=2.7 in /home/user/conda/lib/python3.7/site-packages (from matplotlib->rudolph==0.0.1rc10) (2.8.2)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /home/user/conda/lib/python3.7/site-packages (from matplotlib->rudolph==0.0.1rc10) (1.3.2)\n",
      "Requirement already satisfied: six>=1.5 in /home/user/conda/lib/python3.7/site-packages (from python-dateutil>=2.7->matplotlib->rudolph==0.0.1rc10) (1.16.0)\n",
      "Requirement already satisfied: future>=0.17.1 in /home/user/conda/lib/python3.7/site-packages (from pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (0.18.2)\n",
      "Requirement already satisfied: pyDeprecate==0.3.1 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (0.3.1)\n",
      "Requirement already satisfied: fsspec[http]!=2021.06.0,>=2021.05.0 in /home/user/conda/lib/python3.7/site-packages (from pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (2022.1.0)\n",
      "Requirement already satisfied: torchmetrics>=0.4.1 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (0.7.2)\n",
      "Requirement already satisfied: setuptools==59.5.0 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (59.5.0)\n",
      "Requirement already satisfied: tensorboard>=2.2.0 in /home/user/conda/lib/python3.7/site-packages (from pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (2.3.0)\n",
      "Requirement already satisfied: zipp>=0.5 in /home/user/conda/lib/python3.7/site-packages (from importlib-metadata->transformers~=4.10.2->rudolph==0.0.1rc10) (3.7.0)\n",
      "Requirement already satisfied: urllib3<1.27,>=1.21.1 in /home/user/conda/lib/python3.7/site-packages (from requests->transformers~=4.10.2->rudolph==0.0.1rc10) (1.26.8)\n",
      "Requirement already satisfied: idna<4,>=2.5 in /home/user/conda/lib/python3.7/site-packages (from requests->transformers~=4.10.2->rudolph==0.0.1rc10) (3.3)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /home/user/conda/lib/python3.7/site-packages (from requests->transformers~=4.10.2->rudolph==0.0.1rc10) (2021.10.8)\n",
      "Requirement already satisfied: charset-normalizer~=2.0.0 in /home/user/conda/lib/python3.7/site-packages (from requests->transformers~=4.10.2->rudolph==0.0.1rc10) (2.0.10)\n",
      "Requirement already satisfied: joblib in /home/user/conda/lib/python3.7/site-packages (from sacremoses->transformers~=4.10.2->rudolph==0.0.1rc10) (1.1.0)\n",
      "Requirement already satisfied: aiohttp in /home/user/conda/lib/python3.7/site-packages (from fsspec[http]!=2021.06.0,>=2021.05.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (3.8.1)\n",
      "Requirement already satisfied: protobuf>=3.6.0 in /home/user/conda/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (3.19.3)\n",
      "Requirement already satisfied: tensorboard-plugin-wit>=1.6.0 in /home/user/conda/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.8.1)\n",
      "Requirement already satisfied: absl-py>=0.4 in /home/user/conda/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.0.0)\n",
      "Requirement already satisfied: wheel>=0.26 in /home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (0.37.1)\n",
      "Requirement already satisfied: google-auth-oauthlib<0.5,>=0.4.1 in /home/user/conda/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (0.4.6)\n",
      "Requirement already satisfied: markdown>=2.6.8 in /home/user/conda/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (3.3.6)\n",
      "Requirement already satisfied: grpcio>=1.24.3 in /home/user/conda/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.43.0)\n",
      "Requirement already satisfied: google-auth<2,>=1.6.3 in /home/user/conda/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.35.0)\n",
      "Requirement already satisfied: werkzeug>=0.11.15 in /home/user/conda/lib/python3.7/site-packages (from tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (2.0.2)\n",
      "Requirement already satisfied: cachetools<5.0,>=2.0.0 in /home/user/conda/lib/python3.7/site-packages (from google-auth<2,>=1.6.3->tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (4.2.4)\n",
      "Requirement already satisfied: rsa<5,>=3.1.4 in /home/user/conda/lib/python3.7/site-packages (from google-auth<2,>=1.6.3->tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (4.8)\n",
      "Requirement already satisfied: pyasn1-modules>=0.2.1 in /home/user/conda/lib/python3.7/site-packages (from google-auth<2,>=1.6.3->tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (0.2.8)\n",
      "Requirement already satisfied: requests-oauthlib>=0.7.0 in /home/user/conda/lib/python3.7/site-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.3.0)\n",
      "Requirement already satisfied: attrs>=17.3.0 in /home/user/conda/lib/python3.7/site-packages (from aiohttp->fsspec[http]!=2021.06.0,>=2021.05.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (21.4.0)\n",
      "Requirement already satisfied: aiosignal>=1.1.2 in /home/user/conda/lib/python3.7/site-packages (from aiohttp->fsspec[http]!=2021.06.0,>=2021.05.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.2.0)\n",
      "Requirement already satisfied: frozenlist>=1.1.1 in /home/user/conda/lib/python3.7/site-packages (from aiohttp->fsspec[http]!=2021.06.0,>=2021.05.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.3.0)\n",
      "Requirement already satisfied: multidict<7.0,>=4.5 in /home/user/conda/lib/python3.7/site-packages (from aiohttp->fsspec[http]!=2021.06.0,>=2021.05.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (5.2.0)\n",
      "Requirement already satisfied: async-timeout<5.0,>=4.0.0a3 in /home/user/conda/lib/python3.7/site-packages (from aiohttp->fsspec[http]!=2021.06.0,>=2021.05.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (4.0.2)\n",
      "Requirement already satisfied: asynctest==0.13.0 in /home/user/conda/lib/python3.7/site-packages (from aiohttp->fsspec[http]!=2021.06.0,>=2021.05.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (0.13.0)\n",
      "Requirement already satisfied: yarl<2.0,>=1.0 in /home/user/conda/lib/python3.7/site-packages (from aiohttp->fsspec[http]!=2021.06.0,>=2021.05.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (1.7.2)\n",
      "Requirement already satisfied: pyasn1<0.5.0,>=0.4.6 in /home/user/conda/lib/python3.7/site-packages (from pyasn1-modules>=0.2.1->google-auth<2,>=1.6.3->tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (0.4.8)\n",
      "Requirement already satisfied: oauthlib>=3.0.0 in /home/user/conda/lib/python3.7/site-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard>=2.2.0->pytorch-lightning>=1.0.8->taming-transformers==0.0.1->rudolph==0.0.1rc10) (3.1.1)\n"
     ]
    }
   ],
   "source": [
    "!pip install rudolph==0.0.1rc10 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "0063ccff-1602-4427-b7fc-e8d02a733b12",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from rudalle import get_vae\n",
    "from rudalle.utils import seed_everything\n",
    "\n",
    "import sys\n",
    "sys.path.insert(0, 'ru-dolph')\n",
    "\n",
    "import os\n",
    "from glob import glob\n",
    "from os.path import join\n",
    "from datetime import datetime\n",
    "\n",
    "import torch\n",
    "import bitsandbytes as bnb\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "import torchvision\n",
    "import transformers\n",
    "import more_itertools\n",
    "import numpy as np\n",
    "import torch.nn.functional as F\n",
    "import matplotlib.pyplot as plt\n",
    "import torchvision.transforms as T\n",
    "from tqdm.auto import tqdm\n",
    "from einops import rearrange\n",
    "import youtokentome as yttm\n",
    "\n",
    "import torchvision\n",
    "from rudalle.image_prompts import ImagePrompts\n",
    "from rudolph.model import get_rudolph_model, ruDolphModel, FP16Module\n",
    "from rudolph import utils\n",
    "from rudolph.model.utils import get_attention_mask\n",
    "from rudalle import get_tokenizer, get_vae\n",
    "from rudolph.api import ruDolphApi\n",
    "\n",
    "import datasets\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "69c45243-c0a7-47c1-8efa-9f89777062fa",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = 'cuda'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "afe1a829-6ace-49f5-8a42-177a18551c1e",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/jovyan/.imgenv-a100-0/lib/python3.7/site-packages/huggingface_hub/file_download.py:624: FutureWarning: `cached_download` is the legacy way to download files from the HF hub, please consider upgrading to `hf_hub_download`\n",
      "  FutureWarning,\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3c7d16967234498a88641d469f636041",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading:   0%|          | 0.00/5.33G [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model = get_rudolph_model('2.7B', pretrained=True, fp16=True, device='cuda')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "988a2dde-1237-430a-a1f6-fb16d5365fce",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tokenizer --> ready\n",
      "Working with z of shape (1, 256, 32, 32) = 262144 dimensions.\n",
      "vae --> ready\n"
     ]
    }
   ],
   "source": [
    "tokenizer = get_tokenizer()\n",
    "vae = get_vae(dwt=False).to(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f85fbcae-556d-4ed8-93bc-5c0d1d1771f9",
   "metadata": {},
   "source": [
    "## Данные для обучения"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7a9eec01-485f-49a8-be1d-be5e8f89f9b4",
   "metadata": {},
   "source": [
    "Выгрузим данные с помощью библиотеки datasets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "0d78815f-2550-416e-99c9-a6d7e37c5c12",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Reusing dataset sberquad (/home/jovyan/.cache/huggingface/datasets/sberquad/sberquad/1.0.0/62115d937acf2634cfacbfee10c13a7ee39df3ce345bb45af7088676f9811e77)\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "888288eb0b6f4c4bab64b661d660a0b7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/3 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from datasets import list_datasets\n",
    "dataset = datasets.load_dataset('sberquad')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2133c061-9b99-4c7d-b746-234fc146e149",
   "metadata": {},
   "source": [
    "Нам понадобятся три поля, содержащихся в наборе (`'context'`, `'question'`, `'answers'`)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "c8c9ffa4-89b2-4c8f-9b21-390ec34009ef",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DatasetDict({\n",
       "    train: Dataset({\n",
       "        features: ['id', 'title', 'context', 'question', 'answers'],\n",
       "        num_rows: 45328\n",
       "    })\n",
       "    validation: Dataset({\n",
       "        features: ['id', 'title', 'context', 'question', 'answers'],\n",
       "        num_rows: 5036\n",
       "    })\n",
       "    test: Dataset({\n",
       "        features: ['id', 'title', 'context', 'question', 'answers'],\n",
       "        num_rows: 23936\n",
       "    })\n",
       "})"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "cb80953b-17c5-44c1-9bd0-0d541feab581",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data = dataset['train']\n",
    "val_data = dataset['validation']\n",
    "test_data = dataset['test']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "8c632846-e4ae-4c18-a7bc-b5da9e3aa366",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Контекст + вопрос:  В протерозойских отложениях органические остатки встречаются намного чаще, чем в архейских. Они представлены известковыми выделениями сине-зелёных водорослей, ходами червей, остатками кишечнополостных. Кроме известковых водорослей, к числу древнейших растительных остатков относятся скопления графито-углистого вещества, образовавшегося в результате разложения Corycium enigmaticum. В кремнистых сланцах железорудной формации Канады найдены нитевидные водоросли, грибные нити и формы, близкие современным кокколитофоридам. В железистых кварцитах Северной Америки и Сибири обнаружены железистые продукты жизнедеятельности бактерий.чем представлены органические остатки?\n",
      "----------------------------------------------------------------------------------------------------\n",
      "Ответ:  ['известковыми выделениями сине-зелёных водорослей']\n"
     ]
    }
   ],
   "source": [
    "print('Контекст + вопрос: ', train_data['context'][0] + train_data['question'][0])\n",
    "print('-'*100)\n",
    "print('Ответ: ', train_data['answers'][0]['text'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "af540ea2-a16d-4735-a743-d4f1db72060e",
   "metadata": {},
   "source": [
    "## Подготовка загрузчика данных"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "af3ae029-aeb4-4973-bd0c-72ee7f5feb91",
   "metadata": {},
   "source": [
    "Fine-tuning RUDOLPH-а будум проводить следующим образом: контекст и вопрос будем подавать в левые текстовые токены, а ответ на вопрос будем ожидать в правых текстовых токенах. Вместо токенов изображения подаются нули."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1771292e-8f2b-408f-bb12-7986e511bfce",
   "metadata": {},
   "source": [
    "При предобучении RUDOLPH-а использовался набор спец-символов для левых и правых текстовых токенов, который служил для определения типа решаемой задачи (`<LT_T2I>` и `<LT_T2I>`).\n",
    "\n",
    "Добавим в словарь дополнительные спец-символы (`<LT_TQA>` и `<LT_TQA>`), которые добавляются в начало левых и правых текстовых токенов, соответственно. С помощью данных спец-токенов добавляем модели знания о решаемой задаче.\n",
    "\n",
    "Входная последовательность будет выглядеть следующим образом: \n",
    "```<BOS> <LT_TQA> <context> <question> <EOS>``` ```<zeros-image-tokens>```\n",
    "```<BOS> <RT_TQA> <answer> <EOS>```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "508a3702-7490-4186-ae2a-9e3bffeeceb3",
   "metadata": {},
   "outputs": [],
   "source": [
    "class Args():\n",
    "    def __init__(self, model, checkpoint_path):\n",
    "        self.device = model.get_param('device')\n",
    "        self.l_text_seq_length = model.get_param('l_text_seq_length')\n",
    "        self.r_text_seq_length = model.get_param('r_text_seq_length')\n",
    "        self.image_tokens_per_dim = model.get_param('image_tokens_per_dim')\n",
    "        self.image_seq_length = model.get_param('image_seq_length')\n",
    "        self.epochs = 1\n",
    "        self.save_path= checkpoint_path\n",
    "        self.model_name = 'rudolph_sberquad_'\n",
    "        self.save_every = 1000\n",
    "        self.bs = 4\n",
    "        self.clip = 1.0\n",
    "        self.lr = 2e-5\n",
    "        self.wandb = False\n",
    "        self.lt_loss_weight = 0.01\n",
    "        self.img_loss_weight = 0.01\n",
    "        self.rt_loss_weight = 7\n",
    "        self.image_size = self.image_tokens_per_dim * 8\n",
    "        \n",
    "checkpoint_path = './model/checkpoints/'\n",
    "args = Args(model, checkpoint_path)\n",
    "if not os.path.exists(args.save_path):\n",
    "    os.makedirs(args.save_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "98628c06-f2af-4a63-9b5f-a10e80075a95",
   "metadata": {},
   "outputs": [],
   "source": [
    "SPC_TOKENS = {\n",
    "    '<LT_UNK>': 16384,\n",
    "    '<RT_UNK>': 16385,\n",
    "    '<LT_T2I>': 16386,\n",
    "    '<LT_I2T>': 16387,\n",
    "    '<LT_T2T>': 16388,\n",
    "    '<RT_I2T>': 16389,\n",
    "    \n",
    "    '<LT_TQA>': 16390,\n",
    "    '<RT_TQA>': 16391,\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a390ee81-663d-4e49-a077-ebccbaf2de68",
   "metadata": {},
   "source": [
    "# Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "5cff6b5e-a00e-4a11-a5a5-50cf802f1b8f",
   "metadata": {},
   "outputs": [],
   "source": [
    "class SQuADDataset(Dataset):\n",
    "    spc_id = -1\n",
    "    \n",
    "    def __init__(self, dataset, tokenizer, args):\n",
    "        self.tokenizer = tokenizer\n",
    "        self.dataset = dataset\n",
    "        self.args = args\n",
    "        self.spc_tokens = SPC_TOKENS\n",
    "        \n",
    "    def __len__(self):\n",
    "        return len(self.dataset)\n",
    "    \n",
    "    def encode_text(self, text, text_seq_length, add_special = True):\n",
    "        tokens = self.tokenizer.tokenizer.encode([text], output_type=yttm.OutputType.ID)[0]\n",
    "        bos = [self.tokenizer.bos_id]\n",
    "        if add_special:\n",
    "            bos.append(self.spc_id)\n",
    "        tokens = bos + tokens + [self.tokenizer.eos_id]\n",
    "        return self.tokenizer.prepare_tokens(tokens, text_seq_length)\n",
    "   \n",
    "    def __getitem__(self, item):\n",
    "        \n",
    "        left_text = self.dataset['context'][item] + ' ' + self.dataset['question'][item]\n",
    "        right_text = self.dataset['answers'][item]['text'][0]\n",
    "        \n",
    "        left_special_token = '<LT_TQA>'\n",
    "        right_special_token = '<RT_TQA>'\n",
    "        \n",
    "        left_text = left_text.lower().strip()\n",
    "        left_encoded_text = self.encode_text(left_text, text_seq_length=self.args.l_text_seq_length)\n",
    "        left_encoded_text[torch.where(left_encoded_text == self.spc_id)] = self.spc_tokens[left_special_token]\n",
    "        \n",
    "        right_text = right_text.lower().strip()\n",
    "        right_encoded_text = self.encode_text(right_text, text_seq_length=self.args.r_text_seq_length)\n",
    "        right_encoded_text[torch.where(right_encoded_text == self.spc_id)] = self.spc_tokens[right_special_token]\n",
    "        \n",
    "        return left_encoded_text, right_encoded_text\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "579b2f82-fb4b-4e2e-9148-72b7ec199189",
   "metadata": {},
   "source": [
    "Функция для декодирования последовательности текстовых токенов."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "6baeb414-edc8-41ef-b22d-7fdb6e82834e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def decode_text(tokenizer, encoded, ignore_ids):\n",
    "    return tokenizer.tokenizer.decode(encoded.cpu().numpy().tolist(), ignore_ids=ignore_ids)[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79f361a4-9908-47a5-8dbd-541b59fad5d2",
   "metadata": {},
   "source": [
    "Посмотрим на датасет"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "3aa241bc-3aab-4b4c-9a55-8624fd2bf541",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_qa = SQuADDataset(train_data, tokenizer, args)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "0e7e76eb-2bbe-4cd6-aa6b-7848ac1511b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "idx = random.randint(0, len(dataset_qa)-1)\n",
    "left_encoded_text, right_encoded_text = dataset_qa[idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "05655094-bae9-4794-b626-cc88e0467064",
   "metadata": {},
   "outputs": [],
   "source": [
    "spc_id = -1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "f25254e1-cebb-4f7b-885e-76d1a80b7fb1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Токены, которые будут игнорироваться токенизатором при декодировании.\n",
    "ignore_ids = [\n",
    "    tokenizer.eos_id, tokenizer.bos_id, tokenizer.unk_id, tokenizer.pad_id,\n",
    "    spc_id, *list(SPC_TOKENS.values())\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "b1420126-33a6-45ad-8b03-9450f5550df5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Контекст + Вопрос:  половая зрелость может наступать в разные сроки и при разной длине. как правило, половое созревание наступает у самцов окуня в возрасте 2—3 лет[27][109], у самок — несколько позднее, в 4—5 лет[155]. для некоторых водоёмов, например, для озёр убинское (новосибирская область), тразимено (италия) и пауни (новая зеландия), отмечены случаи, когда окунь становился половозрелым в возрасте 1 года[7][156]. у окуней из низовий енисея половое созревание может наступить только в 6 лет[156], а в озере хубсугул — в 7 лет[45]. отмечалось увеличение возраста полового созревания самок в реках после создания водохранилищ[157]. когда наступает половое созревания у окуней в озере хубсугул?\n",
      "Ответ:  в 7 лет\n"
     ]
    }
   ],
   "source": [
    "print('Контекст + Вопрос: ', decode_text(tokenizer, left_encoded_text, ignore_ids))\n",
    "print('Ответ: ', decode_text(tokenizer, right_encoded_text, ignore_ids))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9a4b8a1b-3043-4a24-bb9b-08f59f31bc19",
   "metadata": {},
   "source": [
    "## RUDOLPH fine-tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "7ed8f73c-0026-4f19-8c55-aedde2415f4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "try:\n",
    "    if args.wandb:\n",
    "        wandb.init(project = args.model_name)\n",
    "except:\n",
    "    args.wandb = False\n",
    "    print('If you want to use wandb logs pls login via wandb -login')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c58de043-8ca6-4a40-b082-38bf7bfff6ae",
   "metadata": {},
   "source": [
    "Можно заморозить некоторые веса модели для ускорения процесса обучения."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "99f955a7-e91c-4833-a123-f21f81bcb557",
   "metadata": {},
   "outputs": [],
   "source": [
    "def freeze(\n",
    "    model,\n",
    "    freeze_emb=False,\n",
    "    freeze_ln=False,\n",
    "    freeze_attn=True,\n",
    "    freeze_ff=True,\n",
    "    freeze_other=False,\n",
    "):\n",
    "    for name, p in model.module.named_parameters():\n",
    "        name = name.lower()\n",
    "        if 'ln' in name or 'norm' in name:\n",
    "            p.requires_grad = not freeze_ln\n",
    "        elif 'embeddings' in name:\n",
    "            p.requires_grad = not freeze_emb\n",
    "        elif 'mlp' in name:\n",
    "            p.requires_grad = not freeze_ff\n",
    "        elif 'attn' in name:\n",
    "            p.requires_grad = not freeze_attn\n",
    "        else:\n",
    "            p.requires_grad = not freeze_other\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "fc83cc32-5058-4cfd-8de8-20e7f75bb740",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataloader = DataLoader(dataset_qa, batch_size=args.bs, shuffle=True, drop_last=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "c90d755a-7b38-4ef0-85c1-34fed218b8d1",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.train()\n",
    "optimizer = bnb.optim.Adam8bit(model.parameters(), lr=args.lr)\n",
    "\n",
    "scheduler = torch.optim.lr_scheduler.OneCycleLR(\n",
    "    optimizer, max_lr=args.lr, final_div_factor=500, \n",
    "    steps_per_epoch=len(train_dataloader), epochs=args.epochs\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "e99e7000-4685-4533-be3c-9c4adaec9a5b",
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_checkpoint(model, optimizer, losslogger, filename='checkpoint.pt'):\n",
    "    # Note: Input model & optimizer should be pre-defined.  This routine only updates their states.\n",
    "    start_epoch = 0\n",
    "    if os.path.isfile(filename):\n",
    "        print(\"=> loading checkpoint '{}'\".format(filename))\n",
    "        checkpoint = torch.load(filename)\n",
    "        start_epoch = checkpoint['epoch']\n",
    "        model.load_state_dict(checkpoint['state_dict'])\n",
    "        optimizer.load_state_dict(checkpoint['optimizer'])\n",
    "        losslogger = checkpoint['losslogger']\n",
    "        print(\"=> loaded checkpoint '{}' (epoch {})\"\n",
    "                  .format(filename, checkpoint['epoch']))\n",
    "    else:\n",
    "        print(\"=> no checkpoint found at '{}'\".format(filename))\n",
    "\n",
    "    return model, optimizer, start_epoch, losslogger"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "2e6dc457-b807-414e-83f4-6a3585686eae",
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(model, optimizer, scheduler, args: Args, train_dataloader, from_checkpoints = False):\n",
    "    \"\"\"\n",
    "      args - arguments for training\n",
    "\n",
    "      train_dataloader - SQuADDataset class with text_q - text_a pairs in batch\n",
    "      \"\"\"\n",
    "    loss_logs = []\n",
    "    try:\n",
    "        t_steps = len(train_dataloader)*args.epochs\n",
    "        progress = tqdm(total = t_steps, desc='🦌🦌🦌finetuning process🦌🦌🦌')\n",
    "    \n",
    "        save_counter = 0\n",
    "        start_epoch = 0\n",
    "        \n",
    "        if from_checkpoints:\n",
    "            model, optimizer, start_epoch, loss_logs = load_checkpoint(model, optimizer, loss_logs, filename=os.path.join(args.save_path,f\"rudolph_sberquad_state6.pt\"))\n",
    "\n",
    "        for epoch in range(start_epoch, args.epochs):\n",
    "      \n",
    "          for encoded_left_text, encoded_right_text in train_dataloader: #, prompt, word\n",
    "        \n",
    "            bs_text = encoded_left_text.shape[0]\n",
    "        \n",
    "            save_counter+=1\n",
    "\n",
    "            model.zero_grad()\n",
    "\n",
    "            image_seq_length = args.image_tokens_per_dim ** 2\n",
    "            total_seq_length = args.l_text_seq_length + image_seq_length + args.r_text_seq_length\n",
    "        \n",
    "            attention_mask_text = get_attention_mask(bs_text, args.l_text_seq_length,\n",
    "                                                 args.image_tokens_per_dim, \n",
    "                                                 args.r_text_seq_length, args.device)\n",
    "        \n",
    "            image_input_ids_text = torch.zeros((bs_text, image_seq_length), dtype=torch.int32).to(args.device)\n",
    "            input_ids_text = torch.cat((encoded_left_text.to(args.device), image_input_ids_text, encoded_right_text.to(args.device)), dim=1)\n",
    "        \n",
    "            loss, loss_values = model.forward(input_ids_text, attention_mask_text, \n",
    "                                                             lt_loss_weight=args.lt_loss_weight,\n",
    "                                                             img_loss_weight=args.img_loss_weight, \n",
    "                                                             rt_loss_weight=args.rt_loss_weight, return_loss=True)\n",
    "            \n",
    "\n",
    "            loss.backward()\n",
    "            \n",
    "            torch.nn.utils.clip_grad_norm_(model.parameters(),args.clip)\n",
    "            optimizer.step()\n",
    "            scheduler.step()\n",
    "            optimizer.zero_grad()\n",
    "            if save_counter % args.save_every == 0:\n",
    "                print(f'Saving checkpoint here {args.model_name}_rudolph_{save_counter}.pt')\n",
    "                plt.plot(loss_logs)\n",
    "                plt.show()\n",
    "                state = {'epoch': epoch + 1, 'state_dict': model.state_dict(),\n",
    "                         'optimizer': optimizer.state_dict(), 'losslogger': loss_logs}\n",
    "                torch.save(\n",
    "                    model.state_dict(),\n",
    "                    os.path.join(args.save_path,f\"{args.model_name}rudolph_{save_counter}.pt\")\n",
    "                )\n",
    "                torch.save(\n",
    "                    state,\n",
    "                    os.path.join(args.save_path,f\"{args.model_name}state_{save_counter}.pt\")\n",
    "                )\n",
    "\n",
    "            if args.wandb:\n",
    "                wandb.log({\"loss\":  loss.item()})\n",
    "            loss_logs+=[loss.item()]\n",
    "            progress.update()\n",
    "            progress.set_postfix({\"loss\": loss.item()})\n",
    "    \n",
    "        print(f'Complitly tuned and saved here  {args.model_name}__textqa_last.pt')\n",
    "        plt.plot(loss_logs)\n",
    "        plt.show()\n",
    "        torch.save(\n",
    "            model.state_dict(),\n",
    "            os.path.join(args.save_path,f\"{args.model_name}textqa_last.pt\")\n",
    "        )\n",
    "    except KeyboardInterrupt:\n",
    "    \n",
    "    \n",
    "        print(f'What for did you stopped? Please change model_path to /{args.save_path}/{args.model_name}_rudolf_Failed_train')\n",
    "        plt.plot(loss_logs)\n",
    "        plt.show()\n",
    "    \n",
    "        torch.save(\n",
    "                model.state_dict(),\n",
    "                os.path.join(args.save_path,f\"{args.model_name}_rudolf_Failed_train.pt\")\n",
    "                )\n",
    "    except Exception as err:\n",
    "        print(f'Failed with {err}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "4d4f0ea0-139f-432c-842b-1c7a1f066e4a",
   "metadata": {},
   "outputs": [],
   "source": [
    "#!rm -r ./model/checkpoints"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "e7caf877-683d-467f-bb7b-a2902c978f24",
   "metadata": {},
   "outputs": [],
   "source": [
    "#!mkdir ./model/checkpoints"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "ea1eef0a-4aba-4e95-a427-ac370118e62a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1f53c4b0c9884b54871a3772ec43c931",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "🦌🦌🦌finetuning process🦌🦌🦌:   0%|          | 0/11332 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_1000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_2000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_3000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_4000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_5000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_6000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_7000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_8000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_9000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_10000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saving checkpoint here rudolph_sberquad__rudolph_11000.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Complitly tuned and saved here  rudolph_sberquad___textqa_last.pt\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "model = freeze(\n",
    "    model=model,\n",
    "    freeze_emb=False,\n",
    "    freeze_ln=False,\n",
    "    freeze_attn=True,\n",
    "    freeze_ff=True,\n",
    "    freeze_other=False,\n",
    ") \n",
    "\n",
    "train(model, optimizer, scheduler, args, train_dataloader)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d06f09bd-e6d6-444f-9223-9a642b2060fc",
   "metadata": {},
   "source": [
    "## Inference "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fdd58c01-f390-4f21-8f15-166845c8c0e3",
   "metadata": {},
   "source": [
    "Итак, RUDOLPH научился отвечать на вопросы по прочитанному тексту, время посмотреть на ответы."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ad9f67e7-586b-47e3-a7d3-78ceabf94136",
   "metadata": {},
   "source": [
    "В классе RuDolphApi присутствует большое количество методов для запуска RUDOLPH-а на разнообразных задачах (generate_text - для генерации текста в левых токенах, image_captioning - генерация описания изображения, dream - генерация изображения по запросу, ...)\n",
    "\n",
    "Для получения ответов в необходимом нам формате создадим собственный класс ruDolphTextQAApi, который наследует от ruDolphApi, и определим в нем функцию generate_text_answers, которая принимает на вход левые текстовые токены и токены изображения и генерирует ответ в правых текстовых токенах."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "56b727ed-aad6-4e7b-b242-1e201907e33a",
   "metadata": {},
   "outputs": [],
   "source": [
    "class ruDolphTextQAApi(ruDolphApi):\n",
    "    def __init__(self, model, tokenizer, vae, spc_tokens=None, quite=False, *, bs=24, q=0.5, txt_top_k=64,\n",
    "                 img_top_k=768, txt_top_p=0.8, img_top_p=0.99, txt_temperature=0.9, img_temperature=1.0):\n",
    "        \n",
    "        super().__init__(model = model, tokenizer = tokenizer, vae = vae, spc_tokens = spc_tokens, quite = quite, bs = bs, q = q, txt_top_k = txt_top_k,\n",
    "                 img_top_k = img_top_k, txt_top_p = txt_top_p, img_top_p = img_top_p, txt_temperature = txt_temperature, img_temperature = img_temperature)\n",
    "    \n",
    "    def generate_text_answers(self, image_tokens, left_text, vocab_size, top_k, top_p, temperature=1.0, use_cache=True, template='', allowed_token_ids=None, special_token='<RT_TQA>'):\n",
    "        '''\n",
    "            Generate right text tokens.\n",
    "        '''\n",
    "        self.model.eval()\n",
    "        torch.cuda.empty_cache()\n",
    "        generated_tokens = []\n",
    "        chunk_bs = left_text.shape[0]\n",
    "\n",
    "        # В случае если мы передаем какое либо унифицированное начало для всех ответов в правые текстовые токены\n",
    "        template = template.lower().strip()\n",
    "        template_encoded = self.encode_text(template, text_seq_length=self.r_text_seq_length)\n",
    "    \n",
    "        template_size = (template_encoded != 0).sum() - 1  # eos\n",
    "        \n",
    "        template_encoded = template_encoded[:template_size]\n",
    "        template_encoded[torch.where(template_encoded == self.spc_id)] = self.spc_tokens[special_token]\n",
    "        \n",
    "        #print('Decoded sequence', self.decode_text(template_encoded))\n",
    "\n",
    "        with torch.no_grad():\n",
    "            attention_mask = self.get_attention_mask(chunk_bs)\n",
    "\n",
    "            out = torch.cat((\n",
    "                left_text.to(self.device),\n",
    "                image_tokens,\n",
    "                template_encoded.repeat(chunk_bs, 1).to(self.device),\n",
    "            ), dim=1)\n",
    "\n",
    "            cache = None\n",
    "            iter_range = range(\n",
    "                self.l_text_seq_length + self.image_seq_length + template_size, \n",
    "                self.l_text_seq_length + self.image_seq_length + self.r_text_seq_length\n",
    "            )\n",
    "            \n",
    "            if not self.quite:\n",
    "                iter_range = tqdm(iter_range)\n",
    "                \n",
    "            for _ in iter_range:  \n",
    "                logits, cache = self.model(out, attention_mask, cache=cache, use_cache=use_cache, return_loss=False)\n",
    "                \n",
    "                logits = logits[:, -1, :self.vocab_size]\n",
    "            \n",
    "                if allowed_token_ids:\n",
    "                    logits = logits[:, allowed_token_ids]\n",
    "            \n",
    "                logits /= temperature\n",
    "                filtered_logits = transformers.top_k_top_p_filtering(logits, top_k=top_k, top_p=top_p)\n",
    "                probs = torch.nn.functional.softmax(filtered_logits, dim=-1)\n",
    "                sample = torch.multinomial(probs, 1)\n",
    "            \n",
    "                if allowed_token_ids:\n",
    "                        sample = torch.tensor(allowed_token_ids).to(self.device)[sample]\n",
    "            \n",
    "                indexes = torch.where(sample >= self.vocab_size - self.l_text_seq_length)\n",
    "                sample[indexes] = 3\n",
    "                out = torch.cat((out, sample), dim=-1)\n",
    "\n",
    "            generated_tokens.append(out[:, -self.r_text_seq_length:])\n",
    "\n",
    "        tokens = torch.cat(generated_tokens)[:,1:]\n",
    "\n",
    "        texts = []\n",
    "\n",
    "        for i in range(tokens.shape[0]):\n",
    "            end = torch.where(tokens[i] == 3)[0]\n",
    "            if (len(end) > 0):\n",
    "                end = end[0]\n",
    "            else:\n",
    "                end = tokens[i].shape[0]\n",
    "            \n",
    "            text = self.decode_text(tokens[i][:end+1]).strip()#end+1\n",
    "            if text:\n",
    "                texts.append(text)\n",
    "            else:\n",
    "                texts.append('пустая строка')\n",
    "        return texts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "e1bdb7f4-5bfc-40e0-bf3e-d263af86e1b5",
   "metadata": {},
   "outputs": [],
   "source": [
    "def inference_textqa(model, tokenizer, vae, dataloader, show_answers = False):\n",
    "    api = ruDolphTextQAApi(model, tokenizer, vae)\n",
    "    \n",
    "    vocab_size = model.get_param('vocab_size')\n",
    "    image_tokens_per_dim = model.get_param('image_tokens_per_dim')\n",
    "    l_text_seq_length = model.get_param('l_text_seq_length')\n",
    "    r_text_seq_length = model.get_param('r_text_seq_length')\n",
    "    image_seq_length = model.get_param('image_seq_length')\n",
    "    device = model.get_param('device')\n",
    "    \n",
    "    spc_id = -1\n",
    "    \n",
    "    vocab = tokenizer.tokenizer.vocab()\n",
    "    \n",
    "    ignore_ids = [\n",
    "    tokenizer.eos_id, tokenizer.bos_id, tokenizer.unk_id, tokenizer.pad_id,\n",
    "    spc_id, *list(SPC_TOKENS.values())\n",
    "    ]\n",
    "    \n",
    "    # allowed_token_ids - список токенов, которые может сгенерировать модель (в случае, если мы хотим ограничить выходной словарь)\n",
    "    allowed_token_ids = []\n",
    "    for i, token in enumerate(vocab):\n",
    "        allowed_token_ids.append(i)\n",
    "        \n",
    "    generated_answers = []\n",
    "    for batch in tqdm(dataloader):\n",
    "        left_text, right_text = batch\n",
    "        left_text = left_text.to(device)\n",
    "        bs = left_text.shape[0]\n",
    "        image_input_ids_text = torch.zeros((bs, image_seq_length), dtype=torch.int32).to(args.device)\n",
    "        \n",
    "        ## Генерируем правые текстовые токены\n",
    "        texts = api.generate_text_answers(image_input_ids_text, left_text, vocab_size, \n",
    "                                    top_k=32, top_p=0.8, temperature=1.0, template = '', \n",
    "                                    allowed_token_ids = allowed_token_ids, special_token='<RT_UNK>')\n",
    "            \n",
    "        generated_answers.extend(texts)\n",
    "        ## Посмотрим на ответы, сгенерированные моделью \n",
    "        if (show_answers):\n",
    "            i = random.choice(range(len(texts)))\n",
    "            print('Контекст + вопрос :', decode_text(tokenizer, left_text[i], ignore_ids))\n",
    "            print('*'*80)\n",
    "            print('Ответ :', texts[i])\n",
    "    return generated_answers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "c58d0e83-ac64-4297-a986-5874fe6e791e",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "767ad0eb-6687-414c-99c8-5edb19e3126d",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_qa_test = SQuADDataset(val_data, tokenizer, args)\n",
    "test_dataloader = DataLoader(dataset_qa_test, batch_size=26, shuffle=False, drop_last=False)\n",
    "\n",
    "#model.load_state_dict(torch.load(checkpoint_path + 'rudolph_sberquad__rudolph_5000.pt'))\n",
    "model.eval();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "76e5342d-e700-414f-9b53-f361aabf7279",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f48a00195c974a1381581f6e0f0398a8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/194 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ed0bbc97b55343a29cf65f2316ff3a9c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "847ca5b6a62642a2b1710590e7b98d45",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "74492ba148254e1390d76567979851dd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "71c41c625af545c8b44e84e3015cc137",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "61016f126131448a96cf462a08f907c2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7168d0f5948f4517a16909e259e9fc4e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a2fab58f637c4881a28053e3a7e4a394",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "22749e7db49046468a4a9c38858e43e3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5e3bf7e5253541539f3379ca22662ad8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f2e564435112478ca27c1cc234823bbe",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3e7bb3b1a65148788bd31fe3ff395537",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "acdbcdb4902048f2a7f3b1978e77e9f8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1e04ee7276dc4770a3ec939554de6ad3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e44c519888594a028e864d3461233dc0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "28da58674a1346ee804a5e162e703640",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a5efc0201d394cba88d2ad8f8f439ca8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "514c74d0edd74acfa5938892e4456239",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "959f418fbad54946933343da1f56351e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "580bcffa7d5e4cd1a44d587937e268da",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "56031bf2cf854791b1d257407b67f980",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "78d82ae7f0ae44c4a9bda7526feb970a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2eae21b3f6bd42ad80b6ea5c25f73bba",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f729adf708dd435e8d59a1ddac3893ea",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d76a334431414bc7b5cb58fcdf6742bc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3e2b23e8b23a49689d2100b6b2d0678d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ce1515c1cd894dad905b63b58a354729",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "135ecc0bd8484eaf94ba0fcc14323c9f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a8890615466f4188826224b6651692f2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "22ff9291c3784e8c9f670468ea7e304c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e5b2d8bad3a54cbe9f7de684512716fa",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d5e8f950b2a8460488aca553e8ae57ac",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3a07093355fb42eebed3e700bae38551",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "794aa897a6e54e10916921c7656a0d55",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2917fcd530874edf92f4c906fa5d7592",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "be98f2d060ee4bda954d04b8a417f97e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f05cddddfe144e83afa61bf02c4d8c7f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "64ef37984c914f59893224992e7d9cb3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a17a5944f3034d5da1e4f40a03aca64c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "496272d167114be4ba9ff5b7c8f53bf9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bc9dd1f13ba84480b776aca32e41c24a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0b3b060d522241c9bd44731c2c1d66de",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bd1379da82d84c6487c4a3d4dc691071",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a727cccfe62e456bb6429d79175eb9e5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1b95bbf98b1247eca0210c832b1e9e50",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f82547c09cd547dd9693796b14b9b4ed",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5ce22c90fdc94e2e8a22664cf648815d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9742be6136c043daa12949dfce9991d2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e3af66bf7ed2492c9d40369c403278e1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f0c24f79ae40406da98f29bd15b009ed",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "00e8d2bf602345d89d06dfbe8ea354ab",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b7f270874c4f4aae80824d5eca85c095",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "647a84f813d2453784f19044a9473f2b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "79657758c6a746b0ab6a631f7f07ed2d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "03d7620671d544df900173151b08ad69",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f075996a34e7486b870bb6358bd41f19",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "944b5756cc6241899fd1ffcd33b28cc8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "IOPub message rate exceeded.\n",
      "The notebook server will temporarily stop sending output\n",
      "to the client in order to avoid crashing it.\n",
      "To change this limit, set the config variable\n",
      "`--NotebookApp.iopub_msg_rate_limit`.\n",
      "\n",
      "Current values:\n",
      "NotebookApp.iopub_msg_rate_limit=1000.0 (msgs/sec)\n",
      "NotebookApp.rate_limit_window=3.0 (secs)\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f1dd770a742c4ac5a40c21f18da1bd93",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d49712e01902494aa19c6e96650f34c8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8678ffc9ba2d4e0684c1e432d739c08d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "689df303914743609245278a7e767f10",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ffedf65c9021400baeb4f706be919e1a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "99b6d9da7a3c4ee5804fdc11d9802ad4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d7728bc289b24d578e142a2f964c2bac",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "96956cf02963406d98a96f1ea1b9ec10",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e95ccd0c2c9f4a12a75274a8e084395c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "01440678d0ef4f6281d2239c8d45f4e5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ed6d0e004a5448c1b8bb24951d0e65f1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7d4430b4c3444814870de02f0bb2811a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b132c487f215465b99b221bebe81c20f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "475d56b610854c42aa7544f46a409d00",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f718f7ce3fc5448b8d58af972f70b8dd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "57ea5862148b4c59b955029c9dca6436",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2cdfa628487b489399ce5384bd664707",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ac95046654b84ac4bf99765ab41d27b2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "14031750d00c42eeb6045e48bd4311c6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dc82b9df3bce4027ac8d70a5bbd21c10",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ec87f228f67d4c0e809f08a6ab450869",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5851fcbe33a2488e83ba1e719155cc57",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e21ce19664a14fc0be912952962c9dc0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2205f74929474d0d921c0324fb0eb157",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "44ee183f52364fb98e1d9be4b3755edb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fb17161ebad84f8fa5cb57cdce82242d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a950b1e5b1c14a71bc11355b1b7a8893",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cc303fee2279431fb2335260ed924747",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "47472e1af5ff4aafb9c89c70393e2402",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c7b2e2db9b0b4720bc50dbda8b869cd0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "942a80755b434e74bf63960a154eba0a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5d9c6a9eb27c450ba73e24041a51c677",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "78bc8e427f084d3bb8d05aaddfbd26c8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e897c72025a54f0ca4daa50a42cb22d0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dd78abb86e724d6dbbd2e8223cd7f954",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7b32c69a86414cd7b4608216bbd84af8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b57bfe7ce6244869bf207c30f7cc96a4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0135f971d02d403cbe3604b32c6f39cf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3250a2c0d33145be97d92540bab65ee0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cb95ac299d94484d9ad6f13bed235a94",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f47fe21319aa4d89be5db1718b0ca55c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e2c4366970ea4ff985da603a8fd019a7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f5ae6630467f4307baae9537022a6006",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5f017724792c4df68dfe46455c14f6be",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "515bf1adbba6409187f220239ae73a33",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7a60b48684c54d4fbb3a8a5bdfe52ef2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "03a2c6c16c5e4a87a5f1826e386c8d67",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e5c8405e0524462a82d22880ba8f4002",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "30b9668bc24f4bbc9094924e8931f04f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7545d468e0a74c0389b27b34848cb6d1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9daa806c57a8413aa80098fe7f811e34",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cdb54af86f0a405bb8641242f9d898b9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f873625ca90640b5b51bfa835b899854",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "66ada3e94df4424aaba0a6328bf368f3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "286405ecd7254994ab401304e68ccd07",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bc8309039b4e4e68af69e5891b944d15",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "eb6498cc0dac40489d5bb63e886e941e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ec08fe0298514e389975f4f18f66d247",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0abe81fc019f45fa98fe60962ccf7567",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4c39706c85014bc3a20c602a975dd962",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0dc689b0654a4a55b0dddbd039516481",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "26cd990ecdc642acab83b6c211cb5a60",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0b6b62dc380f408fb3a3498e49198026",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1715153285284f8288e1eefd2e4b5272",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a88556101a7e4d1b86e35b98d2b305fe",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cf2678ce774148d2b0938246246d64aa",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "693e700ed9fa4c8f83ca0357179112eb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0cd3b533548540b7b870f5dd44d86d22",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "72c88a29f6634d30a9c716dc0ae0b744",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2d3b4bb2ae604963abf47546662b370d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b9f7fad1d69a485988d3dc59443eea8b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8bd86837edc74e0db85439b137aab0ea",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "11f7b7f09cd74f4ba7a1991f8cad554f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "859154b2823d4f7694c66aaf2283b2b5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "903ecde016dc4a68b9cc84c2d5523786",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "dc61c63617094f7980453b7ef86044c1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "105964d9747a4ae5b14a64059dfddf99",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "aa3abfdf2370470faea480de8338bf19",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "818b23c26f42423f8e7ae199d550d052",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5f9119bce840424f991e0a7bea94dce5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2441c6a85c5a4ed4a90fb97da79d7b87",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c173431b7dd24f4986681a8097cf07d7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "31727213fcf4496293cd1c28d29aa7e0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "00ee8baf9449436a814d60a99b8526fc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2c4d75fa29244b038b6026d77b1eaef9",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "713f01ca858a4db08007157be82154e0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "379fc430a736465e864f059f3a9ed1a5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "207bd5f11a064e71bbdf976bf26cffa0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5886117a506e49309d93037092740b3b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3e610d1adbd644eda39c65f2a77f890e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fe55eec6739f489e9c08e1634a5a1dff",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f51e6dc842a8415e93ab0d7c07fb0013",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9c9ca297af394cb1ba30e68605e0ce6a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a0cac422f22e4ab089487a326df796e2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7732f8d734e04ad885dc90fcf50b797c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bdbca1c55f9b4f678e048153b4f987b2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6a3f48a4a668473dba50b25d3874d9fa",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "68456e76e2084456a393b6eeb9fb8b5d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "32d87891068c427086b57a73d1f476e1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "267576543aba49a7ad111ce602df2132",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "da2702dd54a94e95b40163a7f192bcea",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3f71c2219f204be28c9bc80a35b0346e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6a103d2e6d154a20961907a7bcf3a71d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "254e15919a4842449a03666fa2d5cd9e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cc15019183ed4e49b35c0834275c4fe1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cda5ba7976fb4d0bb615e5bad2a447e5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6b4674def08f40e7ab42323e2dee78fb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8e5e1af9b0614e1abd21b3d7d91e1242",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "616eab9310f74e4b849369e47d1e2d40",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "530f2e044599413f95da8db8760ad11b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3ed6a2deaa6f42008f0ff21fd7228511",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1735770f65bd429e917e419ca39896d0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8e85ccc40266425baf45bf2cdacff3ea",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2f07c85e28f643aa8b98d7e36fa5b2e7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d87919155ec04f19b15af272643aa3bb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ea699457157f46c0ae71c91467fba20e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0a17bc96b1a44fc794afce65feaffb31",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7e0eb956e6ee4371a07c6abff0484904",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "17cfbff43f724cc4bbadabb927c08726",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fee9c73c38a64ca8b3b6e9c8f54b25c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8738417c00f94a6088a8baf11232e615",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2dbd6e84bf46456a87f37d0d193a1db1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5618ec85899842c38dbaa0586bf0e732",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "599276e607a2418b9bf0627eb7009581",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ac71bd1c98c54ed0b8bb3764f5031f1e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "30fc0a64700742f98bc6671c4d9ec749",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5b19fa77136347d1a88e358af9c13259",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3285bb4d25344e8eba383dbecdfa43af",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c9287a457f9b40d488ab3eef548f339a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "463ffa3028c042cea4fdeec7d7ccb447",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ef10e075be2741028273f5e9ce29e441",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2be8998586114bae996911de4e88a7fc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "085270485eaa433aa173e1d23ca54fca",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2276a20232694f30bd1f1bb04a458939",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictions = inference_textqa(model, tokenizer, vae, test_dataloader)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f9295e07-f5e6-4436-a20d-104ea33c81de",
   "metadata": {},
   "source": [
    "Запишем список истинных ответов для валидационной выборки и посмотрим на истинный и сгенерированный ответы."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "54a0d18c-2707-447b-9586-bf7937dd3af8",
   "metadata": {},
   "outputs": [],
   "source": [
    "gt = list(map(lambda x: x['text'][0], val_data['answers']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "93774f32-fb14-429f-805f-6bdd9349540e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Контекст: Персонажи манги Наруто получали как положительные, так и негативные отзывы в публикациях, посвящённых непосредственно манге, а также её аниме-экранизации и других медиа-изданиям. Так, сайт Active Anime хвалил серию за то, что персонажи не вырезаны из одного картона и дал высокую оценку очень трогательным эмоциональным испытаниям , которые они проходят в течение сюжета[35]. Anime Boredom согласился с такой оценкой, отмечая уровень хорошо проработанных и интересных персонажей , которые правдоподобно и удачно вписываются в комедийную либо драматичную сюжетную линию[36].\n",
      "****************************************************************************************************\n",
      "Вопрос: За что сайт Active Anime хвалил серию ?\n"
     ]
    }
   ],
   "source": [
    "i = random.choice(range(len(gt)))\n",
    "print('Контекст:', val_data['context'][i])\n",
    "print('*'*100)\n",
    "print('Вопрос:', val_data['question'][i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "1bbca4f2-2ff2-4dd4-a908-0fb4a6dfd633",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Истинный ответ:  персонажи не вырезаны из одного картона \n",
      "Предсказанный ответ: за то, что персонажи не вырезаны из одного картона\n"
     ]
    }
   ],
   "source": [
    "print('Истинный ответ:', gt[i])\n",
    "print('Предсказанный ответ:', predictions[i])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a829a791-d78d-4966-aec7-f3796f97eee3",
   "metadata": {},
   "source": [
    "Рассчитаем метрику f1_score на сгенерированных текстовых ответах и истинных ответах для val части датасета sberQuAD."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "44713732-0ba5-4267-8466-5435bede6f22",
   "metadata": {},
   "outputs": [],
   "source": [
    "from f1_score import compute_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "29e261b3-9c83-46a0-86d7-6e139a957371",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.43181463990937896"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "compute_f1(gt, predictions)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f867175a-fa89-4647-b67c-5cdf3efeaaf0",
   "metadata": {},
   "source": [
    "## Генерация ответов для текстов из другого домена"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "61bfefe0-6229-4c94-900e-06f97b9cf2b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def encode_text(text, tokenizer, text_seq_length, spc_id, add_special = True):\n",
    "        tokens = tokenizer.tokenizer.encode([text], output_type=yttm.OutputType.ID)[0]\n",
    "        bos = [tokenizer.bos_id]\n",
    "        if add_special:\n",
    "            bos.append(spc_id)\n",
    "        tokens = bos + tokens + [tokenizer.eos_id]\n",
    "        return tokenizer.prepare_tokens(tokens, text_seq_length)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "ea9a0f55-c4f6-4898-921c-16c095e954e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_textqa_answer(model, tokenizer, vae, question):\n",
    "    \n",
    "    api = ruDolphTextQAApi(model, tokenizer, vae)\n",
    "    \n",
    "    vocab_size = model.get_param('vocab_size')\n",
    "    image_tokens_per_dim = model.get_param('image_tokens_per_dim')\n",
    "    l_text_seq_length = model.get_param('l_text_seq_length')\n",
    "    r_text_seq_length = model.get_param('r_text_seq_length')\n",
    "    image_seq_length = model.get_param('image_seq_length')\n",
    "    device = model.get_param('device')\n",
    "    \n",
    "    spc_id = -1\n",
    "    \n",
    "    vocab = tokenizer.tokenizer.vocab()\n",
    "    \n",
    "    ignore_ids = [\n",
    "    tokenizer.eos_id, tokenizer.bos_id, tokenizer.unk_id, tokenizer.pad_id,\n",
    "    spc_id, *list(SPC_TOKENS.values())\n",
    "    ]\n",
    "    \n",
    "    # allowed_token_ids - список токенов, которые может сгенерировать модель (в случае, если мы хотим ограничить выходной словарь)\n",
    "    allowed_token_ids = []\n",
    "    for i, token in enumerate(vocab):\n",
    "        allowed_token_ids.append(i)\n",
    "        \n",
    "    left_text_encoded = encode_text(question.lower().strip(), tokenizer, args.l_text_seq_length, spc_id).to(device)\n",
    "    left_text_encoded = torch.unsqueeze(left_text_encoded, 0)\n",
    "    left_special_token = '<LT_TQA>'\n",
    "    \n",
    "    left_text_encoded[torch.where(left_text_encoded == spc_id)] = SPC_TOKENS[left_special_token]\n",
    "    \n",
    "    image_input_ids_text = torch.zeros((1, image_seq_length), dtype=torch.int32).to(args.device)\n",
    "\n",
    "    \n",
    "    answer = api.generate_text_answers(image_input_ids_text, left_text_encoded, vocab_size, \n",
    "                                    top_k=32, top_p=0.8, temperature=1.0, template = '', \n",
    "                                    allowed_token_ids = allowed_token_ids, special_token='<RT_UNK>')\n",
    "            \n",
    "    print('Контекст + вопрос :', question)\n",
    "    print('*'*80)\n",
    "    print('Ответ :', answer[0])\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "dc6ca213-92f7-4af5-90bc-e9daecaa1883",
   "metadata": {},
   "outputs": [],
   "source": [
    "text = 'Уже в детстве Уолт Дисней отличался актёрским талантом. \\\n",
    "Он очень любил фильмы с Чарли Чаплином. Несмотря на большую нагрузку и усталость \\\n",
    "Уолту нравилось ходить в школу. Там он развлекал друзей, пародируя героев Чарли Чаплина. \\\n",
    "Таланту Уолта как художника также было где разгуляться. Иногда учителя вызывали Уолта и \\\n",
    "предлагали ему проиллюстрировать тему урока на школьной доске.'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "33870d79-7bcc-46c3-be0d-b576380524aa",
   "metadata": {},
   "outputs": [],
   "source": [
    "question = 'Кого Уолт Дисней пародировал в школе?'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "6034e86e-76e0-401c-b72b-2ad41c275fcc",
   "metadata": {},
   "outputs": [],
   "source": [
    "left_template = text + ' ' + question"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "87b3de33-8842-4ac0-8e3a-f4596b8c2c8f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cf16c239176c43b7bf15f6e9030b7915",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/126 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Контекст + вопрос : Уже в детстве Уолт Дисней отличался актёрским талантом. Он очень любил фильмы с Чарли Чаплином. Несмотря на большую нагрузку и усталость Уолту нравилось ходить в школу. Там он развлекал друзей, пародируя героев Чарли Чаплина. Таланту Уолта как художника также было где разгуляться. Иногда учителя вызывали Уолта и предлагали ему проиллюстрировать тему урока на школьной доске. Кого Уолт Дисней пародировал в школе?\n",
      "********************************************************************************\n",
      "Ответ : героев чарли чаплина\n"
     ]
    }
   ],
   "source": [
    "get_textqa_answer(model, tokenizer, vae, left_template)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "823aa2ab-2b72-433d-b115-872937aa172d",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}