{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.7.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"! pip install -q datasets tqdm","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2022-11-19T03:14:25.271684Z","iopub.execute_input":"2022-11-19T03:14:25.272085Z","iopub.status.idle":"2022-11-19T03:14:36.330283Z","shell.execute_reply.started":"2022-11-19T03:14:25.272003Z","shell.execute_reply":"2022-11-19T03:14:36.329169Z"},"trusted":true},"execution_count":1,"outputs":[{"name":"stdout","text":"\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n\u001b[0m","output_type":"stream"}]},{"cell_type":"code","source":"from huggingface_hub import notebook_login\nnotebook_login()","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:14:36.332732Z","iopub.execute_input":"2022-11-19T03:14:36.333127Z","iopub.status.idle":"2022-11-19T03:14:36.516419Z","shell.execute_reply.started":"2022-11-19T03:14:36.333085Z","shell.execute_reply":"2022-11-19T03:14:36.515499Z"},"trusted":true},"execution_count":2,"outputs":[{"output_type":"display_data","data":{"text/plain":"VBox(children=(HTML(value='<center> <img\\nsrc=https://huggingface.co/front/assets/huggingface_logo-noborder.sv…","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"c19b93a972a94db084d1d3bdec370216"}},"metadata":{}}]},{"cell_type":"code","source":"import collections\nimport random\nimport os\nimport time\nimport json\nfrom PIL import Image\nimport io\nimport urllib\nimport uuid\nfrom concurrent.futures import ThreadPoolExecutor\nfrom functools import partial\n\nimport numpy as np\nfrom tqdm import tqdm\nfrom datasets import load_dataset\nfrom datasets.utils.file_utils import get_datasets_user_agent\nimport matplotlib.pyplot as plt\nimport tensorflow as tf","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:14:42.774416Z","iopub.execute_input":"2022-11-19T03:14:42.774914Z","iopub.status.idle":"2022-11-19T03:14:47.504264Z","shell.execute_reply.started":"2022-11-19T03:14:42.774876Z","shell.execute_reply":"2022-11-19T03:14:47.503291Z"},"trusted":true},"execution_count":3,"outputs":[]},{"cell_type":"code","source":"iso639_3_letter_code = \"hau\"\n# iso639_3_letter_code = \"tha\"\n# iso639_3_letter_code = \"kir\"\n\n# Download the language specific dataset from HF.\ndataset = load_dataset(\"sil-ai/bloom-captioning\", iso639_3_letter_code, \n                       use_auth_token=True, download_mode='force_redownload')","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:14:47.506122Z","iopub.execute_input":"2022-11-19T03:14:47.507062Z","iopub.status.idle":"2022-11-19T03:15:01.570634Z","shell.execute_reply.started":"2022-11-19T03:14:47.507023Z","shell.execute_reply":"2022-11-19T03:15:01.569653Z"},"trusted":true},"execution_count":4,"outputs":[{"name":"stderr","text":"/opt/conda/lib/python3.7/site-packages/huggingface_hub/utils/_deprecation.py:97: FutureWarning: Deprecated argument(s) used in 'dataset_info': token. Will not be supported from version '0.12'.\n  warnings.warn(message, FutureWarning)\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"Downloading builder script:   0%|          | 0.00/41.6k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"501ec6c17b8f4814a81cf29a8b365606"}},"metadata":{}},{"name":"stdout","text":"Downloading and preparing dataset bloom_captioning/hau to /root/.cache/huggingface/datasets/sil-ai___bloom_captioning/hau/0.0.0/8efe15718b4a50170c9add75b453aec13ec1c5216111d21815428536fe5913ca...\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"Downloading data:   0%|          | 0.00/176M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"ef9c3e49b5674cb48a69456e719b3ac0"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Generating test split: 0 examples [00:00, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":""}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Generating validation split: 0 examples [00:00, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":""}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Generating train split: 0 examples [00:00, ? examples/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":""}},"metadata":{}},{"name":"stdout","text":"Dataset bloom_captioning downloaded and prepared to /root/.cache/huggingface/datasets/sil-ai___bloom_captioning/hau/0.0.0/8efe15718b4a50170c9add75b453aec13ec1c5216111d21815428536fe5913ca. Subsequent calls will reuse this data.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"  0%|          | 0/3 [00:00<?, ?it/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"7229bbedbccb42fd944ed9961ea77a73"}},"metadata":{}}]},{"cell_type":"code","source":"! rm -rf images\n! mkdir images\n\nUSER_AGENT = get_datasets_user_agent()\n\ndef fetch_single_image(image_url, timeout=None, retries=0):\n    request = urllib.request.Request(\n        image_url,\n        data=None,\n        headers={\"user-agent\": USER_AGENT},\n    )\n    with urllib.request.urlopen(request, timeout=timeout) as req:\n        if 'png' in image_url:\n          png = Image.open(io.BytesIO(req.read())).convert('RGBA')\n          png.load() # required for png.split()\n          background = Image.new(\"RGB\", png.size, (255, 255, 255))\n          background.paste(png, mask=png.split()[3]) # 3 is the alpha channel\n          image_id = str(uuid.uuid4())\n          image_path = \"images/\" + image_id + \".jpg\"\n          background.save(image_path, 'JPEG', quality=80)\n        else:\n          image = Image.open(io.BytesIO(req.read()))\n          image_id = str(uuid.uuid4())\n          image_path = \"images/\" + image_id + \".jpg\"\n          image.save(image_path)\n    return image_path\n\ndef fetch_images(batch, num_threads, timeout=None, retries=3):\n    fetch_single_image_with_args = partial(fetch_single_image, timeout=timeout, retries=retries)\n    with ThreadPoolExecutor(max_workers=num_threads) as executor:\n        batch[\"image_path\"] = list(executor.map(fetch_single_image_with_args, batch[\"image_url\"]))\n    return batch\n\nnum_threads = 10\ndataset = dataset.map(fetch_images, batched=True, batch_size=100, fn_kwargs={\"num_threads\": num_threads})","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:15:01.572569Z","iopub.execute_input":"2022-11-19T03:15:01.573410Z","iopub.status.idle":"2022-11-19T03:17:19.020675Z","shell.execute_reply.started":"2022-11-19T03:15:01.573368Z","shell.execute_reply":"2022-11-19T03:17:19.019715Z"},"trusted":true},"execution_count":5,"outputs":[{"output_type":"display_data","data":{"text/plain":"  0%|          | 0/1 [00:00<?, ?ba/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"485144dda5f24af19e8ef5b58edfa0f9"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"  0%|          | 0/1 [00:00<?, ?ba/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"7842927c597443e7a1e4734ba15e8b1e"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"  0%|          | 0/18 [00:00<?, ?ba/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"60d0138d8f974b78b06e5e3897e1c6b3"}},"metadata":{}}]},{"cell_type":"code","source":"!pip install -q googletrans==3.1.0a0\nimport googletrans\nprint(googletrans.LANGUAGES['ha'])\n\nfrom googletrans import Translator\ntranslator = Translator()","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:17:19.022555Z","iopub.execute_input":"2022-11-19T03:17:19.022941Z","iopub.status.idle":"2022-11-19T03:17:32.774253Z","shell.execute_reply.started":"2022-11-19T03:17:19.022900Z","shell.execute_reply":"2022-11-19T03:17:32.773167Z"},"trusted":true},"execution_count":6,"outputs":[{"name":"stdout","text":"\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\nbeatrix-jupyterlab 3.1.7 requires google-cloud-bigquery-storage, which is not installed.\npandas-profiling 3.1.0 requires markupsafe~=2.0.1, but you have markupsafe 2.1.1 which is incompatible.\ngoogle-api-core 1.33.2 requires protobuf!=3.20.0,!=3.20.1,!=4.21.0,!=4.21.1,!=4.21.2,!=4.21.3,!=4.21.4,!=4.21.5,<4.0.0dev,>=3.19.5, but you have protobuf 3.19.4 which is incompatible.\ngcsfs 2022.5.0 requires fsspec==2022.5.0, but you have fsspec 2022.8.2 which is incompatible.\napache-beam 2.40.0 requires dill<0.3.2,>=0.3.1.1, but you have dill 0.3.5.1 which is incompatible.\naiobotocore 2.4.0 requires botocore<1.27.60,>=1.27.59, but you have botocore 1.27.93 which is incompatible.\u001b[0m\u001b[31m\n\u001b[0m\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n\u001b[0mhausa\n","output_type":"stream"}]},{"cell_type":"code","source":"dataset1 = dataset\ndataset1","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:17:32.776983Z","iopub.execute_input":"2022-11-19T03:17:32.777381Z","iopub.status.idle":"2022-11-19T03:17:32.785496Z","shell.execute_reply.started":"2022-11-19T03:17:32.777327Z","shell.execute_reply":"2022-11-19T03:17:32.784552Z"},"trusted":true},"execution_count":7,"outputs":[{"execution_count":7,"output_type":"execute_result","data":{"text/plain":"DatasetDict({\n    test: Dataset({\n        features: ['image_id', 'image_url', 'caption', 'story_id', 'album_id', 'license', 'original_bloom_language_tag', 'index_in_story', 'image_path'],\n        num_rows: 52\n    })\n    validation: Dataset({\n        features: ['image_id', 'image_url', 'caption', 'story_id', 'album_id', 'license', 'original_bloom_language_tag', 'index_in_story', 'image_path'],\n        num_rows: 52\n    })\n    train: Dataset({\n        features: ['image_id', 'image_url', 'caption', 'story_id', 'album_id', 'license', 'original_bloom_language_tag', 'index_in_story', 'image_path'],\n        num_rows: 1761\n    })\n})"},"metadata":{}}]},{"cell_type":"code","source":"import pandas as pd\ndf_train = pd.DataFrame.from_dict(dataset1['train'])\ndf_val = pd.DataFrame.from_dict(dataset1['validation'])","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:17:36.961707Z","iopub.execute_input":"2022-11-19T03:17:36.962063Z","iopub.status.idle":"2022-11-19T03:17:37.178809Z","shell.execute_reply.started":"2022-11-19T03:17:36.962034Z","shell.execute_reply":"2022-11-19T03:17:37.177893Z"},"trusted":true},"execution_count":8,"outputs":[]},{"cell_type":"code","source":"from multiprocessing import Pool, cpu_count\n\ndef translate_to_en(data):\n  result = translator.translate(data, dest='en').text\n  return result\n# For Training data:\nwith Pool(processes= cpu_count() ) as p:\n  ret = p.map(translate_to_en, [cap for cap in df_train['caption']])\n  df_train['en_caption'] = ret\n\n# For Validation data:\nwith Pool(processes= cpu_count() ) as p:\n  ret = p.map(translate_to_en, [cap for cap in df_val['caption']])\n  df_val['en_caption'] = ret\n","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:17:37.345789Z","iopub.execute_input":"2022-11-19T03:17:37.346405Z","iopub.status.idle":"2022-11-19T03:20:01.223420Z","shell.execute_reply.started":"2022-11-19T03:17:37.346365Z","shell.execute_reply":"2022-11-19T03:20:01.221910Z"},"trusted":true},"execution_count":9,"outputs":[]},{"cell_type":"code","source":"df_train.head()","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:15.375800Z","iopub.execute_input":"2022-11-19T03:20:15.376183Z","iopub.status.idle":"2022-11-19T03:20:15.398214Z","shell.execute_reply.started":"2022-11-19T03:20:15.376150Z","shell.execute_reply":"2022-11-19T03:20:15.397367Z"},"trusted":true},"execution_count":10,"outputs":[{"execution_count":10,"output_type":"execute_result","data":{"text/plain":"                               image_id  \\\n0  5e7e2ab6-493f-4430-a635-695fbff76cf0   \n1  04229e2f-45cd-4d5f-a356-033f0e65a853   \n2  fdd362c6-a321-4f6c-94ea-3835f34bba27   \n3  3efbe441-0c46-45d6-bdc5-dbdaf4c2b1fe   \n4  aa9cf976-d82a-4c00-8cbb-89f9d08d98a6   \n\n                                           image_url  \\\n0  https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...   \n1  https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...   \n2  https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...   \n3  https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...   \n4  https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...   \n\n                                             caption  \\\n0  Lokacinan almajiran suna tuƙa jirgin ruwansu, ...   \n1  Sai Yesu ya ƙare addu’a, kuma ya tafi wurin al...   \n2  Almajiran suka ji tsoro sarai da suka ga Yesu,...   \n3  Sai Bitrus ya cewa Yesu, “Malam, in kaine, umu...   \n4  Sai Bitrus ya fita daga jirgin ruwa, kuma ya f...   \n\n                               story_id                              album_id  \\\n0  cd17125d-66c6-467c-b6c3-7463929faff9  a3074fc4-b88f-4769-a6de-dc952fdb35f0   \n1  cd17125d-66c6-467c-b6c3-7463929faff9  a3074fc4-b88f-4769-a6de-dc952fdb35f0   \n2  cd17125d-66c6-467c-b6c3-7463929faff9  a3074fc4-b88f-4769-a6de-dc952fdb35f0   \n3  cd17125d-66c6-467c-b6c3-7463929faff9  a3074fc4-b88f-4769-a6de-dc952fdb35f0   \n4  cd17125d-66c6-467c-b6c3-7463929faff9  a3074fc4-b88f-4769-a6de-dc952fdb35f0   \n\n    license original_bloom_language_tag  index_in_story  \\\n0  cc-by-nc                          ha               0   \n1  cc-by-nc                          ha               1   \n2  cc-by-nc                          ha               2   \n3  cc-by-nc                          ha               3   \n4  cc-by-nc                          ha               4   \n\n                                        image_path  \\\n0  images/36a43711-e6a1-4535-a6db-02df5f70d3a1.jpg   \n1  images/547a6e8e-8f28-42df-85f1-6896d619052b.jpg   \n2  images/a8d10ac5-6e59-400b-80e9-a32878aaef8a.jpg   \n3  images/c07f8ece-4dfe-46d0-be17-f5d325cc0119.jpg   \n4  images/5575aeae-a502-46aa-a9b0-5f6b175a71e1.jpg   \n\n                                          en_caption  \n0  The disciples were sailing their boat, but in ...  \n1  Then Jesus finished praying and went to the di...  \n2  The disciples were very afraid when they saw J...  \n3  Then Peter said to Jesus, \"Teacher, if I am al...  \n4  Then Peter got out of the boat and began to wa...  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>image_id</th>\n      <th>image_url</th>\n      <th>caption</th>\n      <th>story_id</th>\n      <th>album_id</th>\n      <th>license</th>\n      <th>original_bloom_language_tag</th>\n      <th>index_in_story</th>\n      <th>image_path</th>\n      <th>en_caption</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>5e7e2ab6-493f-4430-a635-695fbff76cf0</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...</td>\n      <td>Lokacinan almajiran suna tuƙa jirgin ruwansu, ...</td>\n      <td>cd17125d-66c6-467c-b6c3-7463929faff9</td>\n      <td>a3074fc4-b88f-4769-a6de-dc952fdb35f0</td>\n      <td>cc-by-nc</td>\n      <td>ha</td>\n      <td>0</td>\n      <td>images/36a43711-e6a1-4535-a6db-02df5f70d3a1.jpg</td>\n      <td>The disciples were sailing their boat, but in ...</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>04229e2f-45cd-4d5f-a356-033f0e65a853</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...</td>\n      <td>Sai Yesu ya ƙare addu’a, kuma ya tafi wurin al...</td>\n      <td>cd17125d-66c6-467c-b6c3-7463929faff9</td>\n      <td>a3074fc4-b88f-4769-a6de-dc952fdb35f0</td>\n      <td>cc-by-nc</td>\n      <td>ha</td>\n      <td>1</td>\n      <td>images/547a6e8e-8f28-42df-85f1-6896d619052b.jpg</td>\n      <td>Then Jesus finished praying and went to the di...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>fdd362c6-a321-4f6c-94ea-3835f34bba27</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...</td>\n      <td>Almajiran suka ji tsoro sarai da suka ga Yesu,...</td>\n      <td>cd17125d-66c6-467c-b6c3-7463929faff9</td>\n      <td>a3074fc4-b88f-4769-a6de-dc952fdb35f0</td>\n      <td>cc-by-nc</td>\n      <td>ha</td>\n      <td>2</td>\n      <td>images/a8d10ac5-6e59-400b-80e9-a32878aaef8a.jpg</td>\n      <td>The disciples were very afraid when they saw J...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>3efbe441-0c46-45d6-bdc5-dbdaf4c2b1fe</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...</td>\n      <td>Sai Bitrus ya cewa Yesu, “Malam, in kaine, umu...</td>\n      <td>cd17125d-66c6-467c-b6c3-7463929faff9</td>\n      <td>a3074fc4-b88f-4769-a6de-dc952fdb35f0</td>\n      <td>cc-by-nc</td>\n      <td>ha</td>\n      <td>3</td>\n      <td>images/c07f8ece-4dfe-46d0-be17-f5d325cc0119.jpg</td>\n      <td>Then Peter said to Jesus, \"Teacher, if I am al...</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>aa9cf976-d82a-4c00-8cbb-89f9d08d98a6</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%A4%AF%...</td>\n      <td>Sai Bitrus ya fita daga jirgin ruwa, kuma ya f...</td>\n      <td>cd17125d-66c6-467c-b6c3-7463929faff9</td>\n      <td>a3074fc4-b88f-4769-a6de-dc952fdb35f0</td>\n      <td>cc-by-nc</td>\n      <td>ha</td>\n      <td>4</td>\n      <td>images/5575aeae-a502-46aa-a9b0-5f6b175a71e1.jpg</td>\n      <td>Then Peter got out of the boat and began to wa...</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"from datasets import Dataset, DatasetDict\ntdf = Dataset.from_pandas(df_train)\nvdf = Dataset.from_pandas(df_val)\n\ndf = DatasetDict()\n\ndf['train'] =tdf\ndf['validation'] = vdf","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:15.639369Z","iopub.execute_input":"2022-11-19T03:20:15.639645Z","iopub.status.idle":"2022-11-19T03:20:15.658629Z","shell.execute_reply.started":"2022-11-19T03:20:15.639619Z","shell.execute_reply":"2022-11-19T03:20:15.657794Z"},"trusted":true},"execution_count":11,"outputs":[]},{"cell_type":"code","source":"dataset = df","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:16.524284Z","iopub.execute_input":"2022-11-19T03:20:16.526767Z","iopub.status.idle":"2022-11-19T03:20:16.531473Z","shell.execute_reply.started":"2022-11-19T03:20:16.526731Z","shell.execute_reply":"2022-11-19T03:20:16.530375Z"},"trusted":true},"execution_count":12,"outputs":[]},{"cell_type":"code","source":"import os\nimport numpy as np\nimport h5py\nimport json\nimport torch\nfrom imageio import imread\nfrom skimage.transform import resize\nfrom tqdm import tqdm\nfrom collections import Counter\nfrom random import seed, choice, sample","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:16.719297Z","iopub.execute_input":"2022-11-19T03:20:16.719632Z","iopub.status.idle":"2022-11-19T03:20:18.576973Z","shell.execute_reply.started":"2022-11-19T03:20:16.719605Z","shell.execute_reply":"2022-11-19T03:20:18.576007Z"},"trusted":true},"execution_count":13,"outputs":[]},{"cell_type":"code","source":"def create_input_files(dataset,database,image_folder,captions_per_image,min_word_freq,output_folder,max_len = 100):\n  train_image_paths = []\n  train_image_captions = []\n\n  # test_image_paths = []\n  # test_image_captions = []\n\n  val_image_paths = []\n  val_image_captions = []\n\n  word_freq = Counter()\n\n  punc = \"!\\\"'#$%&\\(\\)\\*\\+.,-/:;=?@\\[\\\\\\]^_`{|}~<>–\"\n\n  for img_path,img_caption in zip(database['train']['image_path'],database['train']['en_caption']):\n    captions = []\n    \n    # temp_caption = img_caption.lower().strip()\n    temp_caption = img_caption.strip()\n    temp_caption = \" \".join(temp_caption.split())\n\n    # Removing punctuations from the input captions\n    for ele in temp_caption:\n      if ele in punc:\n          temp_caption = temp_caption.replace(ele, \"\")\n    \n    temp_caption = temp_caption.replace('\\n', ' ')#.replace('\\xa0', '')\n    img_caption = temp_caption\n    tokens = img_caption.split(\" \")\n    # print(\"\\n\",tokens)\n    word_freq.update(tokens)\n    if len(tokens) <= max_len:         # Why????\n      captions.append(tokens)\n    \n    if len(captions) == 0:\n      continue\n\n    train_image_paths.append(img_path)\n    train_image_captions.append(captions)\n\n  # for img_path,img_caption in zip(database['test']['image_path'],database['test']['caption']):\n  #   captions = []\n    \n  #   # temp_caption = img_caption.lower().strip()\n  #   temp_caption = img_caption.strip()\n  #   temp_caption = \" \".join(temp_caption.split())\n\n  #   for ele in temp_caption:\n  #     if ele in punc:\n  #         temp_caption = temp_caption.replace(ele, \"\")\n    \n  #   temp_caption = temp_caption.replace('\\n', ' ')#.replace('\\xa0', '')\n  #   img_caption = temp_caption\n  #   tokens = img_caption.split(\" \")\n  #   # print(\"\\n\",tokens)\n  #   word_freq.update(tokens)\n  #   if len(tokens) <= max_len:\n  #     captions.append(tokens)\n    \n  #   if len(captions) == 0:\n  #     continue\n  #   test_image_paths.append(img_path)\n  #   test_image_captions.append(captions)\n\n  for img_path,img_caption in zip(database['validation']['image_path'],database['validation']['en_caption']):\n    captions = []\n    \n    # temp_caption = img_caption.lower().strip()\n    temp_caption = img_caption.strip()\n    temp_caption = \" \".join(temp_caption.split())\n\n    for ele in temp_caption:\n      if ele in punc:\n          temp_caption = temp_caption.replace(ele, \"\")\n    \n    temp_caption = temp_caption.replace('\\n', ' ').replace('\\xa0', '')\n    img_caption = temp_caption\n    tokens = img_caption.split(\" \")\n    # print(\"\\n\",tokens)\n    word_freq.update(tokens)\n    if len(tokens) <= max_len:\n      captions.append(tokens)\n    \n    if len(captions) == 0:\n      continue\n\n    val_image_paths.append(img_path)\n    val_image_captions.append(captions)\n\n  words = [w for w in word_freq.keys() if word_freq[w] > min_word_freq]\n  word_map = {k: v + 1 for v, k in enumerate(words)}\n  word_map['<unk>'] = len(word_map) + 1\n  word_map['<start>'] = len(word_map) + 1\n  word_map['<end>'] = len(word_map) + 1\n  word_map['<pad>'] = 0\n\n  base_filename = dataset + '_' + str(captions_per_image) + '_cap_per_img_' + str(min_word_freq) + '_min_word_freq'\n\n  with open(os.path.join(output_folder, 'WORDMAP_' + base_filename + '.json'), 'w') as j:\n    json.dump(word_map, j)\n  \n  seed(123)\n  for impaths, imcaps, split in [(train_image_paths, train_image_captions, 'TRAIN'),\n                                  (val_image_paths, val_image_captions, 'VAL')]:\n\n      with h5py.File(os.path.join(output_folder, split + '_IMAGES_' + base_filename + '.hdf5'), 'a') as h:\n          \n          h.attrs['captions_per_image'] = captions_per_image\n\n          \n          images = h.create_dataset('images', (len(impaths), 3, 256, 256), dtype='uint8')\n\n          print(\"\\nReading %s images and captions, storing to file...\\n\" % split)\n\n          enc_captions = []\n          caplens = []\n\n          for i, path in enumerate(tqdm(impaths)):\n\n              \n              if len(imcaps[i]) < captions_per_image:\n                  captions = imcaps[i] + [choice(imcaps[i]) for _ in range(captions_per_image - len(imcaps[i]))]\n              else:\n                  captions = sample(imcaps[i], k=captions_per_image)\n\n              \n              assert len(captions) == captions_per_image\n\n              \n              img = imread(impaths[i])\n              if len(img.shape) == 2:\n                  img = img[:, :, np.newaxis]\n                  img = np.concatenate([img, img, img], axis=2)\n              img = resize(img, (256, 256))\n              img = img.transpose(2, 0, 1)\n              assert img.shape == (3, 256, 256)\n              assert np.max(img) <= 255\n\n              \n              images[i] = img\n\n              for j, c in enumerate(captions):\n                  \n                  enc_c = [word_map['<start>']] + [word_map.get(word, word_map['<unk>']) for word in c] + [\n                      word_map['<end>']] + [word_map['<pad>']] * (max_len - len(c))\n\n                  \n                  c_len = len(c) + 2\n\n                  enc_captions.append(enc_c)\n                  caplens.append(c_len)\n          assert images.shape[0] * captions_per_image == len(enc_captions) == len(caplens)\n\n          with open(os.path.join(output_folder, split + '_CAPTIONS_' + base_filename + '.json'), 'w') as j:\n              json.dump(enc_captions, j)\n\n          with open(os.path.join(output_folder, split + '_CAPLENS_' + base_filename + '.json'), 'w') as j:\n              json.dump(caplens, j)\n  return train_image_paths,train_image_captions,val_image_paths,val_image_captions,word_freq","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:18.579224Z","iopub.execute_input":"2022-11-19T03:20:18.579630Z","iopub.status.idle":"2022-11-19T03:20:18.604903Z","shell.execute_reply.started":"2022-11-19T03:20:18.579592Z","shell.execute_reply":"2022-11-19T03:20:18.603656Z"},"trusted":true},"execution_count":14,"outputs":[]},{"cell_type":"code","source":"def init_embedding(embeddings):\n    bias = np.sqrt(3.0 / embeddings.size(1))\n    torch.nn.init.uniform_(embeddings, -bias, bias)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:18.606703Z","iopub.execute_input":"2022-11-19T03:20:18.607354Z","iopub.status.idle":"2022-11-19T03:20:18.620669Z","shell.execute_reply.started":"2022-11-19T03:20:18.607305Z","shell.execute_reply":"2022-11-19T03:20:18.619769Z"},"trusted":true},"execution_count":15,"outputs":[]},{"cell_type":"code","source":"def load_embeddings(emb_file, word_map):\n    with open(emb_file, 'r') as f:\n        emb_dim = len(f.readline().split(' ')) - 1\n\n    vocab = set(word_map.keys())\n    embeddings = torch.FloatTensor(len(vocab), emb_dim)\n    init_embedding(embeddings)\n    print(\"\\nLoading embeddings...\")\n    for line in open(emb_file, 'r'):\n        line = line.split(' ')\n\n        emb_word = line[0]\n        embedding = list(map(lambda t: float(t), filter(lambda n: n and not n.isspace(), line[1:])))\n        if emb_word not in vocab:\n            continue\n\n        embeddings[word_map[emb_word]] = torch.FloatTensor(embedding)\n\n    return embeddings, emb_dim","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:18.622793Z","iopub.execute_input":"2022-11-19T03:20:18.623400Z","iopub.status.idle":"2022-11-19T03:20:18.631509Z","shell.execute_reply.started":"2022-11-19T03:20:18.623361Z","shell.execute_reply":"2022-11-19T03:20:18.630534Z"},"trusted":true},"execution_count":16,"outputs":[]},{"cell_type":"code","source":"def clip_gradient(optimizer, grad_clip):\n    for group in optimizer.param_groups:\n        for param in group['params']:\n            if param.grad is not None:\n                param.grad.data.clamp_(-grad_clip, grad_clip)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:18.691710Z","iopub.execute_input":"2022-11-19T03:20:18.692488Z","iopub.status.idle":"2022-11-19T03:20:18.698118Z","shell.execute_reply.started":"2022-11-19T03:20:18.692453Z","shell.execute_reply":"2022-11-19T03:20:18.696457Z"},"trusted":true},"execution_count":17,"outputs":[]},{"cell_type":"code","source":"def save_checkpoint(data_name, epoch, epochs_since_improvement, encoder, decoder, encoder_optimizer, decoder_optimizer,\n                    bleu4, is_best):\n    state = {'epoch': epoch,\n             'epochs_since_improvement': epochs_since_improvement,\n             'bleu-4': bleu4,\n             'encoder': encoder,\n             'decoder': decoder,\n             'encoder_optimizer': encoder_optimizer,\n             'decoder_optimizer': decoder_optimizer}\n    filename = 'checkpoint_' + data_name + '.pth.tar'\n    torch.save(state, filename)\n    if is_best:\n        torch.save(state, 'BEST_' + filename)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:19.174140Z","iopub.execute_input":"2022-11-19T03:20:19.174520Z","iopub.status.idle":"2022-11-19T03:20:19.183220Z","shell.execute_reply.started":"2022-11-19T03:20:19.174487Z","shell.execute_reply":"2022-11-19T03:20:19.182186Z"},"trusted":true},"execution_count":18,"outputs":[]},{"cell_type":"code","source":"class AverageMeter(object):\n    def __init__(self):\n        self.reset()\n\n    def reset(self):\n        self.val = 0\n        self.avg = 0\n        self.sum = 0\n        self.count = 0\n\n    def update(self, val, n=1):\n        self.val = val\n        self.sum += val * n\n        self.count += n\n        self.avg = self.sum / self.count","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:19.552195Z","iopub.execute_input":"2022-11-19T03:20:19.553108Z","iopub.status.idle":"2022-11-19T03:20:19.559869Z","shell.execute_reply.started":"2022-11-19T03:20:19.553076Z","shell.execute_reply":"2022-11-19T03:20:19.558650Z"},"trusted":true},"execution_count":19,"outputs":[]},{"cell_type":"code","source":"def adjust_learning_rate(optimizer, shrink_factor):\n    print(\"\\nDECAYING learning rate.\")\n    for param_group in optimizer.param_groups:\n        param_group['lr'] = param_group['lr'] * shrink_factor\n    print(\"The new learning rate is %f\\n\" % (optimizer.param_groups[0]['lr'],))","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:19.823381Z","iopub.execute_input":"2022-11-19T03:20:19.823669Z","iopub.status.idle":"2022-11-19T03:20:19.830448Z","shell.execute_reply.started":"2022-11-19T03:20:19.823643Z","shell.execute_reply":"2022-11-19T03:20:19.829406Z"},"trusted":true},"execution_count":20,"outputs":[]},{"cell_type":"code","source":"def accuracy(scores, targets, k):\n    batch_size = targets.size(0)\n    _, ind = scores.topk(k, 1, True, True)\n    correct = ind.eq(targets.view(-1, 1).expand_as(ind))\n    correct_total = correct.view(-1).float().sum()\n    return correct_total.item() * (100.0 / batch_size)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:20.218134Z","iopub.execute_input":"2022-11-19T03:20:20.219173Z","iopub.status.idle":"2022-11-19T03:20:20.225535Z","shell.execute_reply.started":"2022-11-19T03:20:20.219117Z","shell.execute_reply":"2022-11-19T03:20:20.224431Z"},"trusted":true},"execution_count":21,"outputs":[]},{"cell_type":"code","source":"train_image_paths, train_image_captions, val_image_paths, val_image_captions, word_freq = create_input_files(\n    dataset = 'haudata',\n    database = dataset,\n    image_folder = './images/',\n    captions_per_image = 1,\n    min_word_freq = 5,\n    output_folder = './images/',\n    max_len = 50\n)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:20:20.605692Z","iopub.execute_input":"2022-11-19T03:20:20.606000Z","iopub.status.idle":"2022-11-19T03:28:35.018706Z","shell.execute_reply.started":"2022-11-19T03:20:20.605974Z","shell.execute_reply":"2022-11-19T03:28:35.017557Z"},"trusted":true},"execution_count":22,"outputs":[{"name":"stdout","text":"\nReading TRAIN images and captions, storing to file...\n\n","output_type":"stream"},{"name":"stderr","text":"  0%|          | 0/1683 [00:00<?, ?it/s]/opt/conda/lib/python3.7/site-packages/ipykernel_launcher.py:130: DeprecationWarning: Starting with ImageIO v3 the behavior of this function will switch to that of iio.v3.imread. To keep the current behavior (and make this warning dissapear) use `import imageio.v2 as imageio` or call `imageio.v2.imread` directly.\n100%|██████████| 1683/1683 [08:11<00:00,  3.43it/s] \n","output_type":"stream"},{"name":"stdout","text":"\nReading VAL images and captions, storing to file...\n\n","output_type":"stream"},{"name":"stderr","text":"100%|██████████| 47/47 [00:02<00:00, 15.84it/s]\n","output_type":"stream"}]},{"cell_type":"code","source":"import torch\nfrom torch.utils.data import Dataset\nimport h5py\nimport json\nimport os","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:28:45.724885Z","iopub.execute_input":"2022-11-19T03:28:45.725265Z","iopub.status.idle":"2022-11-19T03:28:45.730798Z","shell.execute_reply.started":"2022-11-19T03:28:45.725233Z","shell.execute_reply":"2022-11-19T03:28:45.729316Z"},"trusted":true},"execution_count":23,"outputs":[]},{"cell_type":"code","source":"class CaptionDataset(Dataset):\n  def __init__(self,data_folder,data_name,split,transform = None):\n    self.split = split\n    assert self.split in {\"TRAIN\",\"VAL\"}\n\n    self.h = h5py.File(os.path.join(data_folder, self.split + '_IMAGES_' + data_name + '.hdf5'), 'r')\n    self.imgs = self.h['images']\n    self.cpi = self.h.attrs['captions_per_image']\n    with open(os.path.join(data_folder, self.split + '_CAPTIONS_' + data_name + '.json'), 'r') as j:\n        self.captions = json.load(j)\n    with open(os.path.join(data_folder, self.split + '_CAPLENS_' + data_name + '.json'), 'r') as j:\n        self.caplens = json.load(j)\n    self.transform = transform\n\n    self.dataset_size = len(self.captions)\n  \n  def __getitem__(self, i):\n    img = torch.FloatTensor(self.imgs[i // self.cpi] / 255.)\n    if self.transform is not None:\n        img = self.transform(img)\n\n    caption = torch.LongTensor(self.captions[i])\n\n    caplen = torch.LongTensor([self.caplens[i]])\n\n    if self.split is 'TRAIN':\n        return img, caption, caplen\n    else:\n        all_captions = torch.LongTensor(\n            self.captions[((i // self.cpi) * self.cpi):(((i // self.cpi) * self.cpi) + self.cpi)])\n        return img, caption, caplen, all_captions\n\n  def __len__(self):\n      return self.dataset_size","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:28:45.913262Z","iopub.execute_input":"2022-11-19T03:28:45.913564Z","iopub.status.idle":"2022-11-19T03:28:45.924686Z","shell.execute_reply.started":"2022-11-19T03:28:45.913539Z","shell.execute_reply":"2022-11-19T03:28:45.923326Z"},"trusted":true},"execution_count":24,"outputs":[]},{"cell_type":"code","source":"import torch\nfrom torch import nn\nimport torchvision","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:28:46.389245Z","iopub.execute_input":"2022-11-19T03:28:46.389880Z","iopub.status.idle":"2022-11-19T03:28:46.600856Z","shell.execute_reply.started":"2022-11-19T03:28:46.389849Z","shell.execute_reply":"2022-11-19T03:28:46.599951Z"},"trusted":true},"execution_count":25,"outputs":[]},{"cell_type":"code","source":"device = torch.device(\n    \"cuda\" if torch.cuda.is_available()\n    else \"cpu\"\n)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:28:46.612431Z","iopub.execute_input":"2022-11-19T03:28:46.612716Z","iopub.status.idle":"2022-11-19T03:28:46.679828Z","shell.execute_reply.started":"2022-11-19T03:28:46.612690Z","shell.execute_reply":"2022-11-19T03:28:46.678731Z"},"trusted":true},"execution_count":26,"outputs":[]},{"cell_type":"code","source":"class Encoder(nn.Module):\n  def __init__(self,encoded_image_size = 14):\n    super(Encoder,self).__init__()\n    self.enc_image_size = encoded_image_size\n    resnet = torchvision.models.resnet101(pretrained = True)\n    modules = list(resnet.children())[:-2]\n    self.resnet = nn.Sequential(*modules)\n    self.addaptive_pool = nn.AdaptiveAvgPool2d((encoded_image_size,encoded_image_size))\n    self.fine_tune()\n  \n  def forward(self,images):\n    out = self.resnet(images)\n    out = self.addaptive_pool(out)\n    out = out.permute(0,2,3,1)\n    return out\n  \n  def fine_tune(self,fine_tune = True):\n    for p in self.resnet.parameters():\n      p.requires_grad = False\n    for c in list(self.resnet.children())[5:]:\n      for p in c.parameters():\n        p.required_grad = fine_tune","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:28:46.813020Z","iopub.execute_input":"2022-11-19T03:28:46.813901Z","iopub.status.idle":"2022-11-19T03:28:46.822230Z","shell.execute_reply.started":"2022-11-19T03:28:46.813862Z","shell.execute_reply":"2022-11-19T03:28:46.821355Z"},"trusted":true},"execution_count":27,"outputs":[]},{"cell_type":"code","source":"class Attention(nn.Module):\n  def __init__(self,encoder_dim,decoder_dim,attention_dim):\n    super(Attention,self).__init__()\n    self.encoder_att = nn.Linear(encoder_dim,attention_dim)\n    self.decoder_att = nn.Linear(decoder_dim,attention_dim)\n    self.full_att = nn.Linear(attention_dim,1)\n    self.relu = nn.ReLU()\n    self.softmax = nn.Softmax(dim = 1)\n  \n  def forward(self,encoder_out,decoder_hidden):\n    att1 = self.encoder_att(encoder_out)\n    att2 = self.decoder_att(decoder_hidden)\n    att = self.full_att(self.relu(att1 + att2.unsqueeze(1))).squeeze(2)\n    alpha = self.softmax(att)\n    attention_weighted_encoding = (encoder_out * alpha.unsqueeze(2)).sum(dim = 1)\n    return attention_weighted_encoding, alpha","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:28:46.989924Z","iopub.execute_input":"2022-11-19T03:28:46.990193Z","iopub.status.idle":"2022-11-19T03:28:46.997681Z","shell.execute_reply.started":"2022-11-19T03:28:46.990168Z","shell.execute_reply":"2022-11-19T03:28:46.996397Z"},"trusted":true},"execution_count":28,"outputs":[]},{"cell_type":"code","source":"class DecoderWithAttention(nn.Module):\n\n    def __init__(self, attention_dim, embed_dim, decoder_dim, vocab_size, encoder_dim=2048, dropout=0.5):\n        super(DecoderWithAttention, self).__init__()\n\n        self.encoder_dim = encoder_dim\n        self.attention_dim = attention_dim\n        self.embed_dim = embed_dim\n        self.decoder_dim = decoder_dim\n        self.vocab_size = vocab_size\n        self.dropout = dropout\n\n        self.attention = Attention(encoder_dim, decoder_dim, attention_dim)\n\n        self.embedding = nn.Embedding(vocab_size, embed_dim)\n        self.dropout = nn.Dropout(p=self.dropout)\n        self.decode_step = nn.LSTMCell(embed_dim + encoder_dim, decoder_dim, bias=True)\n        self.init_h = nn.Linear(encoder_dim, decoder_dim)\n        self.init_c = nn.Linear(encoder_dim, decoder_dim)\n        self.f_beta = nn.Linear(decoder_dim, encoder_dim)\n        self.sigmoid = nn.Sigmoid()\n        self.fc = nn.Linear(decoder_dim, vocab_size)\n        self.init_weights()\n\n    def init_weights(self):\n        self.embedding.weight.data.uniform_(-0.1, 0.1)\n        self.fc.bias.data.fill_(0)\n        self.fc.weight.data.uniform_(-0.1, 0.1)\n\n    def load_pretrained_embeddings(self, embeddings):\n        self.embedding.weight = nn.Parameter(embeddings)\n\n    def fine_tune_embeddings(self, fine_tune=True):\n        for p in self.embedding.parameters():\n            p.requires_grad = fine_tune\n\n    def init_hidden_state(self, encoder_out):\n        mean_encoder_out = encoder_out.mean(dim=1)\n        h = self.init_h(mean_encoder_out)\n        c = self.init_c(mean_encoder_out)\n        return h, c\n\n    def forward(self, encoder_out, encoded_captions, caption_lengths):\n        batch_size = encoder_out.size(0)\n        encoder_dim = encoder_out.size(-1)\n        vocab_size = self.vocab_size\n\n        encoder_out = encoder_out.view(batch_size, -1, encoder_dim)\n        num_pixels = encoder_out.size(1)\n\n        caption_lengths, sort_ind = caption_lengths.squeeze(1).sort(dim=0, descending=True)\n        encoder_out = encoder_out[sort_ind]\n        encoded_captions = encoded_captions[sort_ind]\n\n        embeddings = self.embedding(encoded_captions) \n        h, c = self.init_hidden_state(encoder_out)\n        decode_lengths = (caption_lengths - 1).tolist()\n\n        predictions = torch.zeros(batch_size, max(decode_lengths), vocab_size).to(device)\n        alphas = torch.zeros(batch_size, max(decode_lengths), num_pixels).to(device)\n        for t in range(max(decode_lengths)):\n            batch_size_t = sum([l > t for l in decode_lengths])\n            attention_weighted_encoding, alpha = self.attention(encoder_out[:batch_size_t],\n                                                                h[:batch_size_t])\n            gate = self.sigmoid(self.f_beta(h[:batch_size_t]))  \n            attention_weighted_encoding = gate * attention_weighted_encoding\n            h, c = self.decode_step(\n                torch.cat([embeddings[:batch_size_t, t, :], attention_weighted_encoding], dim=1),\n                (h[:batch_size_t], c[:batch_size_t]))\n            preds = self.fc(self.dropout(h))\n            predictions[:batch_size_t, t, :] = preds\n            alphas[:batch_size_t, t, :] = alpha\n\n        return predictions, encoded_captions, decode_lengths, alphas, sort_ind","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:28:47.192794Z","iopub.execute_input":"2022-11-19T03:28:47.193449Z","iopub.status.idle":"2022-11-19T03:28:47.211747Z","shell.execute_reply.started":"2022-11-19T03:28:47.193410Z","shell.execute_reply":"2022-11-19T03:28:47.210451Z"},"trusted":true},"execution_count":29,"outputs":[]},{"cell_type":"code","source":"import time\nimport torch.backends.cudnn as cudnn\nimport torch.optim\nimport torch.utils.data\nimport torchvision.transforms as transforms\nfrom torch import nn\nfrom torch.nn.utils.rnn import pack_padded_sequence\nfrom nltk.translate.bleu_score import corpus_bleu","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:28:47.581706Z","iopub.execute_input":"2022-11-19T03:28:47.581990Z","iopub.status.idle":"2022-11-19T03:28:48.093141Z","shell.execute_reply.started":"2022-11-19T03:28:47.581964Z","shell.execute_reply":"2022-11-19T03:28:48.092195Z"},"trusted":true},"execution_count":30,"outputs":[]},{"cell_type":"code","source":"data_folder = './images'\ndata_name = 'haudata_1_cap_per_img_5_min_word_freq'\n\nemb_dim = 512 \nattention_dim = 1024\ndecoder_dim = 512\ndropout = 0.4\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\ncudnn.benchmark = True \n\n\nstart_epoch = 0\nepochs = 15\nepochs_since_improvement = 0\nbatch_size = 8\nworkers = 2\nencoder_lr = 1e-4\ndecoder_lr = 4e-4\ngrad_clip = 5.\nalpha_c = 1. \nbest_bleu4 = 0. \nprint_freq = 100\nfine_tune_encoder = False \ncheckpoint = None","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:51:00.395874Z","iopub.execute_input":"2022-11-19T03:51:00.396478Z","iopub.status.idle":"2022-11-19T03:51:00.405530Z","shell.execute_reply.started":"2022-11-19T03:51:00.396430Z","shell.execute_reply":"2022-11-19T03:51:00.404427Z"},"trusted":true},"execution_count":49,"outputs":[]},{"cell_type":"code","source":"def main():\n    global best_bleu4, epochs_since_improvement, checkpoint, start_epoch, fine_tune_encoder, data_name, word_map\n\n    word_map_file = os.path.join(data_folder, 'WORDMAP_' + data_name + '.json')\n    with open(word_map_file, 'r') as j:\n        word_map = json.load(j)\n\n    if checkpoint is None:\n        decoder = DecoderWithAttention(attention_dim=attention_dim,\n                                       embed_dim=emb_dim,\n                                       decoder_dim=decoder_dim,\n                                       vocab_size=len(word_map),\n                                       dropout=dropout)\n        decoder_optimizer = torch.optim.Adam(params=filter(lambda p: p.requires_grad, decoder.parameters()),\n                                             lr=decoder_lr)\n        encoder = Encoder()\n        encoder.fine_tune(fine_tune_encoder)\n        encoder_optimizer = torch.optim.Adam(params=filter(lambda p: p.requires_grad, encoder.parameters()),\n                                             lr=encoder_lr) if fine_tune_encoder else None\n\n    else:\n        checkpoint = torch.load(checkpoint)\n        start_epoch = checkpoint['epoch'] + 1\n        epochs_since_improvement = checkpoint['epochs_since_improvement']\n        best_bleu4 = checkpoint['bleu-4']\n        decoder = checkpoint['decoder']\n        decoder_optimizer = checkpoint['decoder_optimizer']\n        encoder = checkpoint['encoder']\n        encoder_optimizer = checkpoint['encoder_optimizer']\n        if fine_tune_encoder is True and encoder_optimizer is None:\n            encoder.fine_tune(fine_tune_encoder)\n            encoder_optimizer = torch.optim.Adam(params=filter(lambda p: p.requires_grad, encoder.parameters()),\n                                             lr=encoder_lr)\n\n    decoder = decoder.to(device)\n    encoder = encoder.to(device)\n\n    criterion = nn.CrossEntropyLoss().to(device)\n\n    normalize = transforms.Normalize(mean=[0.77364918,0.7688241 ,0.73459606],\n                                     std=[0.35354225,0.35658083,0.37686874])\n    train_loader = torch.utils.data.DataLoader(\n        CaptionDataset(data_folder, data_name, 'TRAIN', transform=transforms.Compose([normalize])),\n        batch_size=batch_size, shuffle=True, num_workers=workers, pin_memory=True)\n    val_loader = torch.utils.data.DataLoader(\n        CaptionDataset(data_folder, data_name, 'VAL', transform=transforms.Compose([normalize])),\n        batch_size=batch_size, shuffle=True, num_workers=workers, pin_memory=True)\n\n    for epoch in range(start_epoch, epochs):\n\n        \n        if epochs_since_improvement == 50:\n            break\n        if epochs_since_improvement > 0 and epochs_since_improvement % 8 == 0:\n            adjust_learning_rate(decoder_optimizer, 0.8)\n            if fine_tune_encoder:\n                adjust_learning_rate(encoder_optimizer, 0.8)\n\n        \n        train(train_loader=train_loader,\n              encoder=encoder,\n              decoder=decoder,\n              criterion=criterion,\n              encoder_optimizer=encoder_optimizer,\n              decoder_optimizer=decoder_optimizer,\n              epoch=epoch)\n\n        \n        recent_bleu4 = validate(val_loader=val_loader,\n                                encoder=encoder,\n                                decoder=decoder,\n                                criterion=criterion)\n\n        \n        is_best = recent_bleu4 > best_bleu4\n        best_bleu4 = max(recent_bleu4, best_bleu4)\n        if not is_best:\n            epochs_since_improvement += 1\n            print(\"\\nEpochs since last improvement: %d\\n\" % (epochs_since_improvement,))\n        else:\n            epochs_since_improvement = 0\n\n        \n        save_checkpoint(data_name, epoch, epochs_since_improvement, encoder, decoder, encoder_optimizer,\n                        decoder_optimizer, recent_bleu4, is_best)\n\n\ndef train(train_loader, encoder, decoder, criterion, encoder_optimizer, decoder_optimizer, epoch):\n\n\n    decoder.train()\n    encoder.train()\n\n    batch_time = AverageMeter()\n    data_time = AverageMeter()\n    losses = AverageMeter() \n    top5accs = AverageMeter()\n\n    start = time.time()\n\n\n    for i, (imgs, caps, caplens) in enumerate(train_loader):\n        data_time.update(time.time() - start)\n\n      \n        imgs = imgs.to(device)\n        caps = caps.to(device)\n        caplens = caplens.to(device)\n\n\n        imgs = encoder(imgs)\n        scores, caps_sorted, decode_lengths, alphas, sort_ind = decoder(imgs, caps, caplens)\n\n        \n        targets = caps_sorted[:, 1:]\n\n        scores = pack_padded_sequence(scores, decode_lengths, batch_first=True)[0]\n        targets = pack_padded_sequence(targets, decode_lengths, batch_first=True)[0]\n\n        \n        loss = criterion(scores, targets)\n\n        \n        loss += alpha_c * ((1. - alphas.sum(dim=1)) ** 2).mean()\n\n\n        decoder_optimizer.zero_grad()\n        if encoder_optimizer is not None:\n            encoder_optimizer.zero_grad()\n        loss.backward()\n\n        if grad_clip is not None:\n            clip_gradient(decoder_optimizer, grad_clip)\n            if encoder_optimizer is not None:\n                clip_gradient(encoder_optimizer, grad_clip)\n\n\n        decoder_optimizer.step()\n        if encoder_optimizer is not None:\n            encoder_optimizer.step()\n\n        top5 = accuracy(scores, targets, 5)\n        losses.update(loss.item(), sum(decode_lengths))\n        top5accs.update(top5, sum(decode_lengths))\n        batch_time.update(time.time() - start)\n\n        start = time.time()\n\n        if i % print_freq == 0:\n            print('Epoch: [{0}][{1}/{2}]\\t'\n                  'Batch Time {batch_time.val:.3f} ({batch_time.avg:.3f})\\t'\n                  'Data Load Time {data_time.val:.3f} ({data_time.avg:.3f})\\t'\n                  'Loss {loss.val:.4f} ({loss.avg:.4f})\\t'\n                  'Top-5 Accuracy {top5.val:.3f} ({top5.avg:.3f})'.format(epoch, i, len(train_loader),\n                                                                          batch_time=batch_time,\n                                                                          data_time=data_time, loss=losses,\n                                                                          top5=top5accs))\n\n\ndef validate(val_loader, encoder, decoder, criterion):\n\n    decoder.eval()\n    if encoder is not None:\n        encoder.eval()\n\n    batch_time = AverageMeter()\n    losses = AverageMeter()\n    top5accs = AverageMeter()\n\n    start = time.time()\n\n    references = list()  \n    hypotheses = list()  \n\n\n    with torch.no_grad():\n        \n        for i, (imgs, caps, caplens, allcaps) in enumerate(val_loader):\n\n        \n            imgs = imgs.to(device)\n            caps = caps.to(device)\n            caplens = caplens.to(device)\n\n        \n            if encoder is not None:\n                imgs = encoder(imgs)\n            scores, caps_sorted, decode_lengths, alphas, sort_ind = decoder(imgs, caps, caplens)\n\n        \n            targets = caps_sorted[:, 1:]\n\n        \n            scores_copy = scores.clone()\n            scores = pack_padded_sequence(scores, decode_lengths, batch_first=True)[0]\n            targets = pack_padded_sequence(targets, decode_lengths, batch_first=True)[0]\n\n        \n            loss = criterion(scores, targets)\n            loss += alpha_c * ((1. - alphas.sum(dim=1)) ** 2).mean()\n\n            \n            losses.update(loss.item(), sum(decode_lengths))\n            top5 = accuracy(scores, targets, 5)\n            top5accs.update(top5, sum(decode_lengths))\n            batch_time.update(time.time() - start)\n\n            start = time.time()\n\n            if i % print_freq == 0:\n                print('Validation: [{0}/{1}]\\t'\n                      'Batch Time {batch_time.val:.3f} ({batch_time.avg:.3f})\\t'\n                      'Loss {loss.val:.4f} ({loss.avg:.4f})\\t'\n                      'Top-5 Accuracy {top5.val:.3f} ({top5.avg:.3f})\\t'.format(i, len(val_loader), batch_time=batch_time,\n                                                                                loss=losses, top5=top5accs))\n            allcaps = allcaps[sort_ind] \n            for j in range(allcaps.shape[0]):\n                img_caps = allcaps[j].tolist()\n                img_captions = list(\n                    map(lambda c: [w for w in c if w not in {word_map['<start>'], word_map['<pad>']}],\n                        img_caps))\n                references.append(img_captions)\n\n            _, preds = torch.max(scores_copy, dim=2)\n            preds = preds.tolist()\n            temp_preds = list()\n            for j, p in enumerate(preds):\n                temp_preds.append(preds[j][:decode_lengths[j]])\n            preds = temp_preds\n            hypotheses.extend(preds)\n\n            assert len(references) == len(hypotheses)\n\n        # Calculate BLEU-4 scores\n        bleu4 = corpus_bleu(references, hypotheses)\n\n        print(\n            '\\n * LOSS - {loss.avg:.3f}, TOP-5 ACCURACY - {top5.avg:.3f}, BLEU-4 - {bleu}\\n'.format(\n                loss=losses,\n                top5=top5accs,\n                bleu=bleu4))\n\n    return bleu4","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:51:01.539450Z","iopub.execute_input":"2022-11-19T03:51:01.539826Z","iopub.status.idle":"2022-11-19T03:51:01.573762Z","shell.execute_reply.started":"2022-11-19T03:51:01.539796Z","shell.execute_reply":"2022-11-19T03:51:01.572739Z"},"trusted":true},"execution_count":50,"outputs":[]},{"cell_type":"code","source":"main()","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:51:01.856806Z","iopub.execute_input":"2022-11-19T03:51:01.857438Z","iopub.status.idle":"2022-11-19T03:58:26.452507Z","shell.execute_reply.started":"2022-11-19T03:51:01.857398Z","shell.execute_reply":"2022-11-19T03:58:26.451291Z"},"trusted":true},"execution_count":51,"outputs":[{"name":"stdout","text":"Epoch: [0][0/211]\tBatch Time 0.308 (0.308)\tData Load Time 0.150 (0.150)\tLoss 7.7096 (7.7096)\tTop-5 Accuracy 0.000 (0.000)\nEpoch: [0][100/211]\tBatch Time 0.118 (0.132)\tData Load Time 0.000 (0.002)\tLoss 6.0991 (6.2365)\tTop-5 Accuracy 33.333 (30.257)\nEpoch: [0][200/211]\tBatch Time 0.096 (0.129)\tData Load Time 0.000 (0.001)\tLoss 6.2207 (6.1172)\tTop-5 Accuracy 31.068 (31.305)\nValidation: [0/6]\tBatch Time 0.286 (0.286)\tLoss 5.8290 (5.8290)\tTop-5 Accuracy 29.795 (29.795)\t\n\n * LOSS - 5.844, TOP-5 ACCURACY - 30.342, BLEU-4 - 0.014022871302856095\n\nEpoch: [1][0/211]\tBatch Time 0.329 (0.329)\tData Load Time 0.171 (0.171)\tLoss 5.9406 (5.9406)\tTop-5 Accuracy 34.351 (34.351)\nEpoch: [1][100/211]\tBatch Time 0.092 (0.132)\tData Load Time 0.000 (0.002)\tLoss 6.2309 (5.7842)\tTop-5 Accuracy 28.000 (34.033)\nEpoch: [1][200/211]\tBatch Time 0.115 (0.129)\tData Load Time 0.000 (0.001)\tLoss 5.2959 (5.7083)\tTop-5 Accuracy 44.348 (35.238)\nValidation: [0/6]\tBatch Time 0.243 (0.243)\tLoss 5.3678 (5.3678)\tTop-5 Accuracy 37.433 (37.433)\t\n\n * LOSS - 5.566, TOP-5 ACCURACY - 37.093, BLEU-4 - 0.016464747981673047\n\nEpoch: [2][0/211]\tBatch Time 0.375 (0.375)\tData Load Time 0.225 (0.225)\tLoss 5.3809 (5.3809)\tTop-5 Accuracy 42.727 (42.727)\nEpoch: [2][100/211]\tBatch Time 0.142 (0.132)\tData Load Time 0.000 (0.003)\tLoss 5.0045 (5.4356)\tTop-5 Accuracy 46.995 (39.149)\nEpoch: [2][200/211]\tBatch Time 0.158 (0.130)\tData Load Time 0.000 (0.001)\tLoss 5.3973 (5.3788)\tTop-5 Accuracy 36.416 (39.566)\nValidation: [0/6]\tBatch Time 0.242 (0.242)\tLoss 5.3383 (5.3383)\tTop-5 Accuracy 38.725 (38.725)\t\n\n * LOSS - 5.372, TOP-5 ACCURACY - 40.111, BLEU-4 - 0.019551277202883355\n\nEpoch: [3][0/211]\tBatch Time 0.284 (0.284)\tData Load Time 0.161 (0.161)\tLoss 5.2065 (5.2065)\tTop-5 Accuracy 41.007 (41.007)\nEpoch: [3][100/211]\tBatch Time 0.136 (0.133)\tData Load Time 0.000 (0.002)\tLoss 5.3776 (5.1436)\tTop-5 Accuracy 41.711 (42.414)\nEpoch: [3][200/211]\tBatch Time 0.118 (0.130)\tData Load Time 0.000 (0.001)\tLoss 5.0545 (5.1272)\tTop-5 Accuracy 45.652 (42.411)\nValidation: [0/6]\tBatch Time 0.255 (0.255)\tLoss 5.1799 (5.1799)\tTop-5 Accuracy 44.706 (44.706)\t\n\n * LOSS - 5.222, TOP-5 ACCURACY - 42.415, BLEU-4 - 0.024743789612362677\n\nEpoch: [4][0/211]\tBatch Time 0.334 (0.334)\tData Load Time 0.175 (0.175)\tLoss 5.0122 (5.0122)\tTop-5 Accuracy 42.391 (42.391)\nEpoch: [4][100/211]\tBatch Time 0.102 (0.132)\tData Load Time 0.000 (0.003)\tLoss 4.7978 (4.9593)\tTop-5 Accuracy 52.066 (44.301)\nEpoch: [4][200/211]\tBatch Time 0.118 (0.131)\tData Load Time 0.000 (0.002)\tLoss 4.7681 (4.9472)\tTop-5 Accuracy 45.763 (44.809)\nValidation: [0/6]\tBatch Time 0.247 (0.247)\tLoss 5.0870 (5.0870)\tTop-5 Accuracy 42.400 (42.400)\t\n\n * LOSS - 5.079, TOP-5 ACCURACY - 44.162, BLEU-4 - 0.027480562248338357\n\nEpoch: [5][0/211]\tBatch Time 0.296 (0.296)\tData Load Time 0.165 (0.165)\tLoss 4.5396 (4.5396)\tTop-5 Accuracy 52.206 (52.206)\nEpoch: [5][100/211]\tBatch Time 0.146 (0.136)\tData Load Time 0.000 (0.002)\tLoss 4.4685 (4.7997)\tTop-5 Accuracy 53.390 (46.310)\nEpoch: [5][200/211]\tBatch Time 0.156 (0.133)\tData Load Time 0.000 (0.001)\tLoss 4.5302 (4.7731)\tTop-5 Accuracy 49.537 (46.927)\nValidation: [0/6]\tBatch Time 0.245 (0.245)\tLoss 4.9471 (4.9471)\tTop-5 Accuracy 47.027 (47.027)\t\n\n * LOSS - 5.022, TOP-5 ACCURACY - 44.480, BLEU-4 - 0.023523687384274412\n\n\nEpochs since last improvement: 1\n\nEpoch: [6][0/211]\tBatch Time 0.297 (0.297)\tData Load Time 0.149 (0.149)\tLoss 4.6578 (4.6578)\tTop-5 Accuracy 47.097 (47.097)\nEpoch: [6][100/211]\tBatch Time 0.118 (0.133)\tData Load Time 0.000 (0.002)\tLoss 4.6569 (4.6318)\tTop-5 Accuracy 48.462 (48.950)\nEpoch: [6][200/211]\tBatch Time 0.155 (0.131)\tData Load Time 0.000 (0.001)\tLoss 4.5305 (4.6352)\tTop-5 Accuracy 49.550 (48.899)\nValidation: [0/6]\tBatch Time 0.285 (0.285)\tLoss 4.9810 (4.9810)\tTop-5 Accuracy 47.368 (47.368)\t\n\n * LOSS - 4.978, TOP-5 ACCURACY - 45.671, BLEU-4 - 0.03579302748923495\n\nEpoch: [7][0/211]\tBatch Time 0.271 (0.271)\tData Load Time 0.146 (0.146)\tLoss 4.2141 (4.2141)\tTop-5 Accuracy 56.452 (56.452)\nEpoch: [7][100/211]\tBatch Time 0.121 (0.137)\tData Load Time 0.000 (0.002)\tLoss 4.4603 (4.4894)\tTop-5 Accuracy 52.941 (50.909)\nEpoch: [7][200/211]\tBatch Time 0.095 (0.132)\tData Load Time 0.000 (0.001)\tLoss 4.3999 (4.4835)\tTop-5 Accuracy 51.613 (51.284)\nValidation: [0/6]\tBatch Time 0.246 (0.246)\tLoss 4.7026 (4.7026)\tTop-5 Accuracy 52.105 (52.105)\t\n\n * LOSS - 4.917, TOP-5 ACCURACY - 46.783, BLEU-4 - 0.04166026769385804\n\nEpoch: [8][0/211]\tBatch Time 0.307 (0.307)\tData Load Time 0.149 (0.149)\tLoss 4.5009 (4.5009)\tTop-5 Accuracy 52.830 (52.830)\nEpoch: [8][100/211]\tBatch Time 0.125 (0.136)\tData Load Time 0.000 (0.003)\tLoss 4.2156 (4.3493)\tTop-5 Accuracy 55.280 (53.169)\nEpoch: [8][200/211]\tBatch Time 0.138 (0.131)\tData Load Time 0.000 (0.001)\tLoss 4.5598 (4.3489)\tTop-5 Accuracy 48.322 (53.376)\nValidation: [0/6]\tBatch Time 0.252 (0.252)\tLoss 4.7143 (4.7143)\tTop-5 Accuracy 48.402 (48.402)\t\n\n * LOSS - 4.895, TOP-5 ACCURACY - 47.021, BLEU-4 - 0.026696193068540975\n\n\nEpochs since last improvement: 1\n\nEpoch: [9][0/211]\tBatch Time 0.304 (0.304)\tData Load Time 0.143 (0.143)\tLoss 3.9263 (3.9263)\tTop-5 Accuracy 61.932 (61.932)\nEpoch: [9][100/211]\tBatch Time 0.165 (0.133)\tData Load Time 0.007 (0.002)\tLoss 4.2902 (4.1760)\tTop-5 Accuracy 51.634 (55.898)\nEpoch: [9][200/211]\tBatch Time 0.140 (0.132)\tData Load Time 0.000 (0.001)\tLoss 4.1136 (4.2137)\tTop-5 Accuracy 58.667 (55.418)\nValidation: [0/6]\tBatch Time 0.253 (0.253)\tLoss 4.9210 (4.9210)\tTop-5 Accuracy 49.091 (49.091)\t\n\n * LOSS - 4.885, TOP-5 ACCURACY - 48.213, BLEU-4 - 0.04021128450716708\n\n\nEpochs since last improvement: 2\n\nEpoch: [10][0/211]\tBatch Time 0.316 (0.316)\tData Load Time 0.152 (0.152)\tLoss 3.8964 (3.8964)\tTop-5 Accuracy 55.769 (55.769)\nEpoch: [10][100/211]\tBatch Time 0.124 (0.134)\tData Load Time 0.000 (0.002)\tLoss 3.8476 (4.0563)\tTop-5 Accuracy 60.588 (57.722)\nEpoch: [10][200/211]\tBatch Time 0.147 (0.131)\tData Load Time 0.000 (0.001)\tLoss 4.2724 (4.0825)\tTop-5 Accuracy 50.769 (57.551)\nValidation: [0/6]\tBatch Time 0.276 (0.276)\tLoss 4.6922 (4.6922)\tTop-5 Accuracy 49.115 (49.115)\t\n\n * LOSS - 4.885, TOP-5 ACCURACY - 48.848, BLEU-4 - 0.027820895019976453\n\n\nEpochs since last improvement: 3\n\nEpoch: [11][0/211]\tBatch Time 0.294 (0.294)\tData Load Time 0.149 (0.149)\tLoss 3.8355 (3.8355)\tTop-5 Accuracy 59.494 (59.494)\nEpoch: [11][100/211]\tBatch Time 0.125 (0.134)\tData Load Time 0.000 (0.002)\tLoss 4.4980 (3.9356)\tTop-5 Accuracy 49.296 (60.093)\nEpoch: [11][200/211]\tBatch Time 0.117 (0.132)\tData Load Time 0.000 (0.001)\tLoss 4.0429 (3.9528)\tTop-5 Accuracy 56.757 (59.856)\nValidation: [0/6]\tBatch Time 0.252 (0.252)\tLoss 5.1949 (5.1949)\tTop-5 Accuracy 43.925 (43.925)\t\n\n * LOSS - 4.893, TOP-5 ACCURACY - 47.339, BLEU-4 - 0.03382954765458924\n\n\nEpochs since last improvement: 4\n\nEpoch: [12][0/211]\tBatch Time 0.278 (0.278)\tData Load Time 0.151 (0.151)\tLoss 3.8003 (3.8003)\tTop-5 Accuracy 63.248 (63.248)\nEpoch: [12][100/211]\tBatch Time 0.130 (0.127)\tData Load Time 0.000 (0.002)\tLoss 3.7837 (3.7772)\tTop-5 Accuracy 62.000 (62.908)\nEpoch: [12][200/211]\tBatch Time 0.120 (0.131)\tData Load Time 0.000 (0.001)\tLoss 3.8158 (3.8216)\tTop-5 Accuracy 62.329 (62.100)\nValidation: [0/6]\tBatch Time 0.251 (0.251)\tLoss 4.9399 (4.9399)\tTop-5 Accuracy 49.010 (49.010)\t\n\n * LOSS - 4.917, TOP-5 ACCURACY - 46.863, BLEU-4 - 0.032003639568270174\n\n\nEpochs since last improvement: 5\n\nEpoch: [13][0/211]\tBatch Time 0.330 (0.330)\tData Load Time 0.151 (0.151)\tLoss 3.6706 (3.6706)\tTop-5 Accuracy 63.743 (63.743)\nEpoch: [13][100/211]\tBatch Time 0.154 (0.131)\tData Load Time 0.000 (0.002)\tLoss 3.8211 (3.6629)\tTop-5 Accuracy 61.084 (64.874)\nEpoch: [13][200/211]\tBatch Time 0.120 (0.131)\tData Load Time 0.000 (0.001)\tLoss 3.9737 (3.6875)\tTop-5 Accuracy 62.698 (64.356)\nValidation: [0/6]\tBatch Time 0.252 (0.252)\tLoss 4.9450 (4.9450)\tTop-5 Accuracy 43.969 (43.969)\t\n\n * LOSS - 4.909, TOP-5 ACCURACY - 46.465, BLEU-4 - 0.030394516903027528\n\n\nEpochs since last improvement: 6\n\nEpoch: [14][0/211]\tBatch Time 0.279 (0.279)\tData Load Time 0.143 (0.143)\tLoss 3.6254 (3.6254)\tTop-5 Accuracy 62.411 (62.411)\nEpoch: [14][100/211]\tBatch Time 0.145 (0.132)\tData Load Time 0.007 (0.002)\tLoss 3.7614 (3.5299)\tTop-5 Accuracy 61.905 (67.269)\nEpoch: [14][200/211]\tBatch Time 0.144 (0.132)\tData Load Time 0.000 (0.001)\tLoss 3.6952 (3.5609)\tTop-5 Accuracy 62.564 (66.827)\nValidation: [0/6]\tBatch Time 0.271 (0.271)\tLoss 4.9005 (4.9005)\tTop-5 Accuracy 43.974 (43.974)\t\n\n * LOSS - 4.913, TOP-5 ACCURACY - 47.021, BLEU-4 - 0.04594417718940004\n\n","output_type":"stream"}]},{"cell_type":"code","source":"import torch\nimport torch.nn.functional as F\nimport numpy as np\nimport json\nimport torchvision.transforms as transforms\nimport matplotlib.pyplot as plt\nimport matplotlib.cm as cm\nimport skimage.transform\nfrom imageio import imread\nimport os\nfrom skimage.transform import resize\nfrom PIL import Image\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n\ndef caption_image_beam_search(encoder, decoder, image_path, word_map, beam_size=3):\n    k = beam_size\n    vocab_size = len(word_map)\n\n    img = imread(image_path)\n    if len(img.shape) == 2:\n        img = img[:, :, np.newaxis]\n        img = np.concatenate([img, img, img], axis=2)\n    img = resize(img, (256, 256))\n    img = img.transpose(2, 0, 1)\n    img = img / 255.\n    img = torch.FloatTensor(img).to(device)\n    normalize = transforms.Normalize(mean=[0.77364918,0.7688241 ,0.73459606],\n                                     std=[0.35354225,0.35658083,0.37686874])\n    transform = transforms.Compose([normalize])\n    image = transform(img)\n    image = image.unsqueeze(0)  \n    encoder_out = encoder(image)\n    enc_image_size = encoder_out.size(1)\n    encoder_dim = encoder_out.size(3)\n\n    \n    encoder_out = encoder_out.view(1, -1, encoder_dim) \n    num_pixels = encoder_out.size(1)\n\n  \n    encoder_out = encoder_out.expand(k, num_pixels, encoder_dim) \n\n    k_prev_words = torch.LongTensor([[word_map['<start>']]] * k).to(device)\n\n    seqs = k_prev_words \n    top_k_scores = torch.zeros(k, 1).to(device)  \n    seqs_alpha = torch.ones(k, 1, enc_image_size, enc_image_size).to(device) \n\n    complete_seqs = list()\n    complete_seqs_alpha = list()\n    complete_seqs_scores = list()\n\n    step = 1\n    h, c = decoder.init_hidden_state(encoder_out)\n\n    while True:\n\n        embeddings = decoder.embedding(k_prev_words).squeeze(1) \n\n        awe, alpha = decoder.attention(encoder_out, h) \n\n        alpha = alpha.view(-1, enc_image_size, enc_image_size) \n\n        gate = decoder.sigmoid(decoder.f_beta(h)) \n        awe = gate * awe\n\n        h, c = decoder.decode_step(torch.cat([embeddings, awe], dim=1), (h, c)) \n\n        scores = decoder.fc(h)  \n        scores = F.log_softmax(scores, dim=1)\n\n        \n        scores = top_k_scores.expand_as(scores) + scores \n\n        if step == 1:\n            top_k_scores, top_k_words = scores[0].topk(k, 0, True, True)\n        else:\n            top_k_scores, top_k_words = scores.view(-1).topk(k, 0, True, True)\n\n        prev_word_inds = top_k_words / vocab_size \n        next_word_inds = top_k_words % vocab_size \n        seqs = torch.cat([seqs[prev_word_inds.long()], next_word_inds.unsqueeze(1)], dim=1)\n        seqs_alpha = torch.cat([seqs_alpha[prev_word_inds.long()], alpha[prev_word_inds.long()].unsqueeze(1)],\n                               dim=1)\n\n        incomplete_inds = [ind for ind, next_word in enumerate(next_word_inds) if\n                           next_word != word_map['<end>']]\n        complete_inds = list(set(range(len(next_word_inds))) - set(incomplete_inds))\n\n\n        if len(complete_inds) > 0:\n            complete_seqs.extend(seqs[complete_inds].tolist())\n            complete_seqs_alpha.extend(seqs_alpha[complete_inds].tolist())\n            complete_seqs_scores.extend(top_k_scores[complete_inds])\n        k -= len(complete_inds)\n\n        if k == 0:\n            break\n        seqs = seqs[incomplete_inds]\n        seqs_alpha = seqs_alpha[incomplete_inds]\n        h = h[prev_word_inds[incomplete_inds].long()]\n        c = c[prev_word_inds[incomplete_inds].long()]\n        encoder_out = encoder_out[prev_word_inds[incomplete_inds].long()]\n        top_k_scores = top_k_scores[incomplete_inds].unsqueeze(1)\n        k_prev_words = next_word_inds[incomplete_inds].unsqueeze(1)\n\n        if step > 50:\n            break\n        step += 1\n\n    i = complete_seqs_scores.index(max(complete_seqs_scores))\n    seq = complete_seqs[i]\n    alphas = complete_seqs_alpha[i]\n\n    return seq, alphas\n\n\ndef visualize_att(image_path, seq, alphas, rev_word_map, smooth=True):\n\n    image = Image.open(image_path)\n    image = image.resize([14 * 24, 14 * 24], Image.LANCZOS)\n\n    words = [rev_word_map[ind] for ind in seq]\n\n    for t in range(len(words)):\n        if t > 50:\n            break\n        plt.subplot(np.ceil(len(words) / 5.), 5, t + 1)\n\n        plt.text(0, 1, '%s' % (words[t]), color='black', backgroundcolor='white', fontsize=12)\n        plt.imshow(image)\n        current_alpha = alphas[t, :]\n        if smooth:\n            alpha = skimage.transform.pyramid_expand(current_alpha.numpy(), upscale=24, sigma=8)\n        else:\n            alpha = skimage.transform.resize(current_alpha.numpy(), [14 * 24, 14 * 24])\n        if t == 0:\n            plt.imshow(alpha, alpha=0)\n        else:\n            plt.imshow(alpha, alpha=0.8)\n        plt.set_cmap(cm.Greys_r)\n        plt.axis('off')\n    plt.show()","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:58:41.066435Z","iopub.execute_input":"2022-11-19T03:58:41.066985Z","iopub.status.idle":"2022-11-19T03:58:41.093309Z","shell.execute_reply.started":"2022-11-19T03:58:41.066951Z","shell.execute_reply":"2022-11-19T03:58:41.092279Z"},"trusted":true},"execution_count":54,"outputs":[]},{"cell_type":"code","source":"os.listdir('./images')[10:20]","metadata":{"execution":{"iopub.status.busy":"2022-11-19T03:58:42.292842Z","iopub.execute_input":"2022-11-19T03:58:42.293636Z","iopub.status.idle":"2022-11-19T03:58:42.302956Z","shell.execute_reply.started":"2022-11-19T03:58:42.293602Z","shell.execute_reply":"2022-11-19T03:58:42.302002Z"},"trusted":true},"execution_count":55,"outputs":[{"execution_count":55,"output_type":"execute_result","data":{"text/plain":"['f0d75449-dcd3-485d-965e-c8270c4f5ef4.jpg',\n '840e5442-7fbf-40f5-ad67-7ad1b81d297e.jpg',\n '2c451b8e-8e9e-4527-b451-55d9bf6dedd1.jpg',\n 'fc199ab2-b436-49d3-80db-360f7df3368f.jpg',\n '5a22757e-54fe-41cb-a6fc-1887d47a7f75.jpg',\n 'fbc5670b-f897-424e-9836-7232f8f7fdf4.jpg',\n 'a08d3dbf-56d5-4c74-9f6b-50311f2d70c5.jpg',\n '6d2623b1-3e69-4ad3-9b69-37f892deac53.jpg',\n '150c247e-c89b-4de1-8b54-cf1151b6aa74.jpg',\n '33895118-3220-45de-959a-f0271923a73c.jpg']"},"metadata":{}}]},{"cell_type":"code","source":"checkpoint = torch.load('./BEST_checkpoint_haudata_1_cap_per_img_5_min_word_freq.pth.tar', map_location=str(device))\ndecoder = checkpoint['decoder']\ndecoder = decoder.to(device)\ndecoder.eval()\nencoder = checkpoint['encoder']\nencoder = encoder.to(device)\nencoder.eval()\n\nwith open('./images/WORDMAP_haudata_1_cap_per_img_5_min_word_freq.json', 'r') as j:\n    word_map = json.load(j)\nrev_word_map = {v: k for k, v in word_map.items()}\n\nimg_path = './images/840e5442-7fbf-40f5-ad67-7ad1b81d297e.jpg'\n\nseq, alphas = caption_image_beam_search(encoder, decoder,img_path, word_map, 35)\nalphas = torch.FloatTensor(alphas)\n# visualize_att(img_path, seq, alphas, rev_word_map, True)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:01:38.094289Z","iopub.execute_input":"2022-11-19T04:01:38.095058Z","iopub.status.idle":"2022-11-19T04:01:39.215684Z","shell.execute_reply.started":"2022-11-19T04:01:38.095021Z","shell.execute_reply":"2022-11-19T04:01:39.214619Z"},"trusted":true},"execution_count":67,"outputs":[{"name":"stderr","text":"/opt/conda/lib/python3.7/site-packages/ipykernel_launcher.py:19: DeprecationWarning: Starting with ImageIO v3 the behavior of this function will switch to that of iio.v3.imread. To keep the current behavior (and make this warning dissapear) use `import imageio.v2 as imageio` or call `imageio.v2.imread` directly.\n","output_type":"stream"}]},{"cell_type":"code","source":"seq\nwords = [rev_word_map[ind] for ind in seq]\nwords","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:01:40.376842Z","iopub.execute_input":"2022-11-19T04:01:40.377222Z","iopub.status.idle":"2022-11-19T04:01:40.387668Z","shell.execute_reply.started":"2022-11-19T04:01:40.377190Z","shell.execute_reply":"2022-11-19T04:01:40.385215Z"},"trusted":true},"execution_count":68,"outputs":[{"execution_count":68,"output_type":"execute_result","data":{"text/plain":"['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']"},"metadata":{}}]},{"cell_type":"code","source":"import matplotlib.pyplot as plt\nimg = plt.imread('./images/840e5442-7fbf-40f5-ad67-7ad1b81d297e.jpg')\nplt.imshow(img)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:01:47.801939Z","iopub.execute_input":"2022-11-19T04:01:47.802310Z","iopub.status.idle":"2022-11-19T04:01:48.038196Z","shell.execute_reply.started":"2022-11-19T04:01:47.802280Z","shell.execute_reply":"2022-11-19T04:01:48.037235Z"},"trusted":true},"execution_count":69,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"import pandas as pd\n\ntest_data = pd.read_csv('../input/purdue-test-dataset/test.csv')\ntest_data.head()","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:04:16.074651Z","iopub.execute_input":"2022-11-19T04:04:16.075267Z","iopub.status.idle":"2022-11-19T04:04:16.112417Z","shell.execute_reply.started":"2022-11-19T04:04:16.075223Z","shell.execute_reply":"2022-11-19T04:04:16.111505Z"},"trusted":true},"execution_count":70,"outputs":[{"execution_count":70,"output_type":"execute_result","data":{"text/plain":"                                         Id  \\\n0  0293a8c7-b69e-4c58-8caf-4a58e17bbacb_kir   \n1  02d89130-e6e5-4aea-88ed-99e100aafe84_tha   \n2  04763763-d79a-4a97-a529-20c5178d7d2d_tha   \n3  0478f1ca-3db4-4025-a838-255d45b2c603_hau   \n4  04a00291-ef0f-4bb5-8b37-75d300ceffaf_kir   \n\n                                            ImageURL ISO639-3  \n0  https://bloom-vist.s3.amazonaws.com/%D0%A6%D0%...      kir  \n1  https://bloom-vist.s3.amazonaws.com/%E0%B9%82%...      tha  \n2  https://bloom-vist.s3.amazonaws.com/%E0%B9%80%...      tha  \n3  https://bloom-vist.s3.amazonaws.com/Gallina%20...      hau  \n4  https://bloom-vist.s3.amazonaws.com/%D0%9C%D0%...      kir  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>Id</th>\n      <th>ImageURL</th>\n      <th>ISO639-3</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0293a8c7-b69e-4c58-8caf-4a58e17bbacb_kir</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%D0%A6%D0%...</td>\n      <td>kir</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>02d89130-e6e5-4aea-88ed-99e100aafe84_tha</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%B9%82%...</td>\n      <td>tha</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>04763763-d79a-4a97-a529-20c5178d7d2d_tha</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%B9%80%...</td>\n      <td>tha</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>0478f1ca-3db4-4025-a838-255d45b2c603_hau</td>\n      <td>https://bloom-vist.s3.amazonaws.com/Gallina%20...</td>\n      <td>hau</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>04a00291-ef0f-4bb5-8b37-75d300ceffaf_kir</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%D0%9C%D0%...</td>\n      <td>kir</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"id_list = test_data.loc[test_data['ISO639-3'] == 'hau']['Id'].to_list()","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:22:07.365456Z","iopub.execute_input":"2022-11-19T04:22:07.365848Z","iopub.status.idle":"2022-11-19T04:22:07.374196Z","shell.execute_reply.started":"2022-11-19T04:22:07.365818Z","shell.execute_reply":"2022-11-19T04:22:07.371276Z"},"trusted":true},"execution_count":96,"outputs":[]},{"cell_type":"code","source":"test_data['Predicted'] = ''","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:22:08.379707Z","iopub.execute_input":"2022-11-19T04:22:08.380071Z","iopub.status.idle":"2022-11-19T04:22:08.385074Z","shell.execute_reply.started":"2022-11-19T04:22:08.380039Z","shell.execute_reply":"2022-11-19T04:22:08.384046Z"},"trusted":true},"execution_count":97,"outputs":[]},{"cell_type":"code","source":"for id in id_list:\n    image_url = test_data.loc[test_data['Id'] == id]['ImageURL'].to_list()[0]\n    idx = test_data.loc[test_data['Id'] == id].index[0]\n    image_path = fetch_single_image(image_url)\n    checkpoint = torch.load('./BEST_checkpoint_haudata_1_cap_per_img_5_min_word_freq.pth.tar', map_location=str(device))\n    decoder = checkpoint['decoder']\n    decoder = decoder.to(device)\n    decoder.eval()\n    encoder = checkpoint['encoder']\n    encoder = encoder.to(device)\n    encoder.eval()\n\n    with open('./images/WORDMAP_haudata_1_cap_per_img_5_min_word_freq.json', 'r') as j:\n        word_map = json.load(j)\n    rev_word_map = {v: k for k, v in word_map.items()}\n\n    seq, alphas = caption_image_beam_search(encoder, decoder,image_path, word_map, 35)\n    words = [rev_word_map[ind] for ind in seq]\n    print(words)\n    new_words = [i for i in words if i != '<start>']\n    new_words_1 = [i for i in new_words if i != '<end>']\n    new_words_2 = [i for i in new_words_1 if i != '<unk>']\n    caption = ' '.join(word for word in new_words_2)\n#     print(caption)\n    final_caption = translator.translate(caption,dest = 'ha').text\n#     print(final_caption)\n    \n    test_data.iloc[idx]['Predicted'] = final_caption","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:22:13.331584Z","iopub.execute_input":"2022-11-19T04:22:13.331953Z","iopub.status.idle":"2022-11-19T04:23:46.440848Z","shell.execute_reply.started":"2022-11-19T04:22:13.331922Z","shell.execute_reply":"2022-11-19T04:23:46.439787Z"},"trusted":true},"execution_count":98,"outputs":[{"name":"stderr","text":"/opt/conda/lib/python3.7/site-packages/ipykernel_launcher.py:19: DeprecationWarning: Starting with ImageIO v3 the behavior of this function will switch to that of iio.v3.imread. To keep the current behavior (and make this warning dissapear) use `import imageio.v2 as imageio` or call `imageio.v2.imread` directly.\n","output_type":"stream"},{"name":"stdout","text":"['<start>', 'The', '<unk>', 'is', '<unk>', '<unk>', '<end>']\n['<start>', 'Then', 'the', 'Ant', 'Girl', 'was', 'coming', '<end>']\n['<start>', 'There', 'are', '<unk>', '<unk>', 'on', 'the', 'ground', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'Please', 'pay', 'attention', 'Pay', 'attention', 'to', 'the', 'wedding', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']\n['<start>', 'Then', 'the', 'Ant', 'Girl', 'was', 'very', 'happy', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<unk>', '<end>']\n['<start>', 'There', 'are', '<unk>', '<unk>', 'on', 'the', '<unk>', '<end>']\n['<start>', 'There', 'are', '<unk>', '<unk>', 'on', 'the', 'ground', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']\n['<start>', 'The', '<unk>', '<unk>', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'Then', 'the', 'Ant', 'Girl', 'was', 'coming', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']\n['<start>', 'Here', 'are', 'the', '<unk>', 'Ado', 'try', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', '<unk>', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<unk>', '<end>']\n['<start>', 'Then', 'the', 'Ant', 'Girl', 'got', 'some', 'advice', '<end>']\n['<start>', 'Plants', 'have', 'a', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'has', 'a', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'of', 'the', '<unk>', 'is', 'not', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', 'used', 'to', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'There', 'are', '<unk>', '<unk>', 'on', 'the', 'ground', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'Then', 'the', 'Ant', 'Girl', 'got', 'some', 'advice', '<end>']\n['<start>', 'The', '<unk>', '<unk>', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'Then', 'the', 'Ant', 'Girl', 'was', 'very', 'happy', '<end>']\n['<start>', '<unk>', 'Fatima', 'passed', 'a', '<unk>', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'has', 'a', '<unk>', '<unk>', '<end>']\n['<start>', 'The', '<unk>', '<unk>', '<unk>', '<end>']\n['<start>', '<unk>', 'parts', 'include', 'the', '<unk>', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', 'Modi', 'will', 'fix', 'it', '<end>']\n['<start>', 'The', '<unk>', 'has', '<unk>', '<unk>', '<end>']\n['<start>', 'The', 'rich', 'man', 'was', 'very', 'happy', '<end>']\n['<start>', 'The', '<unk>', 'was', '<unk>', 'with', 'a', '<unk>', '<end>']\n['<start>', 'The', 'rich', 'man', 'was', 'very', 'happy', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<unk>', 'are', '<unk>', '<end>']\n['<start>', 'The', 'rich', 'man', 'was', 'very', 'happy', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<unk>', '<end>']\n['<start>', 'Then', 'the', 'Ant', 'Girl', 'was', 'very', 'happy', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']\n['<start>', 'Then', 'the', 'Ant', 'Girl', 'got', 'some', 'advice', '<end>']\n['<start>', 'The', '<unk>', 'is', '<unk>', '<end>']\n['<start>', 'The', 'rich', 'man', 'was', 'very', 'happy', '<end>']\n","output_type":"stream"}]},{"cell_type":"code","source":"test_data.sample(10)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:23:54.180004Z","iopub.execute_input":"2022-11-19T04:23:54.180505Z","iopub.status.idle":"2022-11-19T04:23:54.197479Z","shell.execute_reply.started":"2022-11-19T04:23:54.180464Z","shell.execute_reply":"2022-11-19T04:23:54.196214Z"},"trusted":true},"execution_count":99,"outputs":[{"execution_count":99,"output_type":"execute_result","data":{"text/plain":"                                           Id  \\\n26   20daeb3b-d1f3-43ce-a1ee-b80008e3f8be_hau   \n139  ce1c599d-8fa4-4fcc-b62f-e816095eb757_kir   \n47   490f3269-0a81-4643-bbbe-f13703bda8de_hau   \n5    095b71cc-ef78-4861-af3f-5353c1bfcc59_tha   \n7    09bb3c0d-e00b-4c6d-978f-218874717a72_tha   \n178  69698865-cbc8-45c0-b6df-90e2b3337c8a_kir   \n168  8d2e6ebf-97fd-45f3-9665-1cc3431c93ad_tha   \n46   48be99d1-a017-4aff-af69-8a5fa6d3442e_hau   \n145  d783941b-149d-4078-b02a-2c48de172f6c_hau   \n17   1278528d-1cfe-4516-ad37-f700823444dd_hau   \n\n                                              ImageURL ISO639-3  \\\n26   https://bloom-vist.s3.amazonaws.com/01/image10...      hau   \n139  https://bloom-vist.s3.amazonaws.com/%D0%9A%D0%...      kir   \n47   https://bloom-vist.s3.amazonaws.com/Kyauta%20m...      hau   \n5    https://bloom-vist.s3.amazonaws.com/%E0%B8%AB%...      tha   \n7    https://bloom-vist.s3.amazonaws.com/%E0%B9%80%...      tha   \n178  https://bloom-vist.s3.amazonaws.com/test/testk...      kir   \n168  https://bloom-vist.s3.amazonaws.com/test/testt...      tha   \n46   https://bloom-vist.s3.amazonaws.com/Gallina%20...      hau   \n145  https://bloom-vist.s3.amazonaws.com/01/image8.jpg      hau   \n17   https://bloom-vist.s3.amazonaws.com/Gallina%20...      hau   \n\n                             Predicted  \n26   Sai Budurwar Ant tayi murna sosai  \n139                                     \n47               Modi ne zai gyara shi  \n5                                       \n7                                       \n178                                     \n168                                     \n46                                 The  \n145                   Fatima ta wuce a  \n17                           The shine  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>Id</th>\n      <th>ImageURL</th>\n      <th>ISO639-3</th>\n      <th>Predicted</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>26</th>\n      <td>20daeb3b-d1f3-43ce-a1ee-b80008e3f8be_hau</td>\n      <td>https://bloom-vist.s3.amazonaws.com/01/image10...</td>\n      <td>hau</td>\n      <td>Sai Budurwar Ant tayi murna sosai</td>\n    </tr>\n    <tr>\n      <th>139</th>\n      <td>ce1c599d-8fa4-4fcc-b62f-e816095eb757_kir</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%D0%9A%D0%...</td>\n      <td>kir</td>\n      <td></td>\n    </tr>\n    <tr>\n      <th>47</th>\n      <td>490f3269-0a81-4643-bbbe-f13703bda8de_hau</td>\n      <td>https://bloom-vist.s3.amazonaws.com/Kyauta%20m...</td>\n      <td>hau</td>\n      <td>Modi ne zai gyara shi</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>095b71cc-ef78-4861-af3f-5353c1bfcc59_tha</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%B8%AB%...</td>\n      <td>tha</td>\n      <td></td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>09bb3c0d-e00b-4c6d-978f-218874717a72_tha</td>\n      <td>https://bloom-vist.s3.amazonaws.com/%E0%B9%80%...</td>\n      <td>tha</td>\n      <td></td>\n    </tr>\n    <tr>\n      <th>178</th>\n      <td>69698865-cbc8-45c0-b6df-90e2b3337c8a_kir</td>\n      <td>https://bloom-vist.s3.amazonaws.com/test/testk...</td>\n      <td>kir</td>\n      <td></td>\n    </tr>\n    <tr>\n      <th>168</th>\n      <td>8d2e6ebf-97fd-45f3-9665-1cc3431c93ad_tha</td>\n      <td>https://bloom-vist.s3.amazonaws.com/test/testt...</td>\n      <td>tha</td>\n      <td></td>\n    </tr>\n    <tr>\n      <th>46</th>\n      <td>48be99d1-a017-4aff-af69-8a5fa6d3442e_hau</td>\n      <td>https://bloom-vist.s3.amazonaws.com/Gallina%20...</td>\n      <td>hau</td>\n      <td>The</td>\n    </tr>\n    <tr>\n      <th>145</th>\n      <td>d783941b-149d-4078-b02a-2c48de172f6c_hau</td>\n      <td>https://bloom-vist.s3.amazonaws.com/01/image8.jpg</td>\n      <td>hau</td>\n      <td>Fatima ta wuce a</td>\n    </tr>\n    <tr>\n      <th>17</th>\n      <td>1278528d-1cfe-4516-ad37-f700823444dd_hau</td>\n      <td>https://bloom-vist.s3.amazonaws.com/Gallina%20...</td>\n      <td>hau</td>\n      <td>The shine</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"test_data.to_csv(r'submission.csv',index = False)","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:24:37.364218Z","iopub.execute_input":"2022-11-19T04:24:37.364704Z","iopub.status.idle":"2022-11-19T04:24:37.380563Z","shell.execute_reply.started":"2022-11-19T04:24:37.364664Z","shell.execute_reply":"2022-11-19T04:24:37.379631Z"},"trusted":true},"execution_count":100,"outputs":[]},{"cell_type":"code","source":"from IPython.display import FileLink\nFileLink(r'submission.csv')","metadata":{"execution":{"iopub.status.busy":"2022-11-19T04:24:46.631234Z","iopub.execute_input":"2022-11-19T04:24:46.631934Z","iopub.status.idle":"2022-11-19T04:24:46.638515Z","shell.execute_reply.started":"2022-11-19T04:24:46.631899Z","shell.execute_reply":"2022-11-19T04:24:46.637498Z"},"trusted":true},"execution_count":101,"outputs":[{"execution_count":101,"output_type":"execute_result","data":{"text/plain":"/kaggle/working/submission.csv","text/html":"<a href='submission.csv' target='_blank'>submission.csv</a><br>"},"metadata":{}}]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}