Medical Image Captioning.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.14","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"nvidiaTeslaT4","dataSources":[{"sourceId":951996,"sourceType":"datasetVersion","datasetId":516716},{"sourceId":9853467,"sourceType":"datasetVersion","datasetId":6017914}],"dockerImageVersionId":30786,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":true}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"import numpy as np\nimport pandas as pd\nimport os\nimport matplotlib.pyplot as plt\nfrom PIL import Image\nimport torch\nimport torch.nn as nn\nimport torch.nn.functional as F\nimport torch.optim as optim\nimport torchvision\nfrom torchvision import transforms\nimport transformers\nfrom transformers import AutoProcessor, Blip2ForConditionalGeneration, BitsAndBytesConfig\nfrom torch.utils.data import Dataset, DataLoader, random_split\nfrom sklearn.model_selection import train_test_split\nfrom tqdm import tqdm\n#from peft import LoraConfig, get_peft_model","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:34:09.885788Z","iopub.execute_input":"2024-11-10T05:34:09.886747Z","iopub.status.idle":"2024-11-10T05:34:26.425098Z","shell.execute_reply.started":"2024-11-10T05:34:09.886698Z","shell.execute_reply":"2024-11-10T05:34:26.424282Z"}},"outputs":[],"execution_count":26},{"cell_type":"code","source":"!pip install nltk","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:52:19.192479Z","iopub.execute_input":"2024-11-10T05:52:19.192926Z","iopub.status.idle":"2024-11-10T05:52:32.150396Z","shell.execute_reply.started":"2024-11-10T05:52:19.192885Z","shell.execute_reply":"2024-11-10T05:52:32.149311Z"}},"outputs":[{"name":"stdout","text":"Collecting nltk\n  Using cached nltk-3.9.1-py3-none-any.whl.metadata (2.9 kB)\nRequirement already satisfied: click in /opt/conda/lib/python3.10/site-packages (from nltk) (8.1.7)\nRequirement already satisfied: joblib in /opt/conda/lib/python3.10/site-packages (from nltk) (1.4.2)\nRequirement already satisfied: regex>=2021.8.3 in /opt/conda/lib/python3.10/site-packages (from nltk) (2024.5.15)\nRequirement already satisfied: tqdm in /opt/conda/lib/python3.10/site-packages (from nltk) (4.66.4)\nUsing cached nltk-3.9.1-py3-none-any.whl (1.5 MB)\nInstalling collected packages: nltk\n\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.\npreprocessing 0.1.13 requires nltk==3.2.4, but you have nltk 3.9.1 which is incompatible.\u001b[0m\u001b[31m\n\u001b[0mSuccessfully installed nltk-3.9.1\n","output_type":"stream"}],"execution_count":78},{"cell_type":"code","source":"import nltk","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:52:34.992253Z","iopub.execute_input":"2024-11-10T05:52:34.993091Z","iopub.status.idle":"2024-11-10T05:52:34.998868Z","shell.execute_reply.started":"2024-11-10T05:52:34.993030Z","shell.execute_reply":"2024-11-10T05:52:34.997435Z"}},"outputs":[],"execution_count":79},{"cell_type":"code","source":"nltk.download('wordnet')\nnltk.download('omw-1.4')\nnltk.download('punkt')","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:52:37.273805Z","iopub.execute_input":"2024-11-10T05:52:37.274607Z","iopub.status.idle":"2024-11-10T05:52:37.282425Z","shell.execute_reply.started":"2024-11-10T05:52:37.274564Z","shell.execute_reply":"2024-11-10T05:52:37.281401Z"}},"outputs":[{"name":"stdout","text":"[nltk_data] Downloading package wordnet to /usr/share/nltk_data...\n[nltk_data]   Package wordnet is already up-to-date!\n[nltk_data] Downloading package omw-1.4 to /usr/share/nltk_data...\n[nltk_data]   Package omw-1.4 is already up-to-date!\n[nltk_data] Downloading package punkt to /usr/share/nltk_data...\n[nltk_data]   Package punkt is already up-to-date!\n","output_type":"stream"},{"execution_count":80,"output_type":"execute_result","data":{"text/plain":"True"},"metadata":{}}],"execution_count":80},{"cell_type":"code","source":"from nltk.translate.meteor_score import meteor_score\nfrom nltk import word_tokenize","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:52:39.997952Z","iopub.execute_input":"2024-11-10T05:52:39.998339Z","iopub.status.idle":"2024-11-10T05:52:40.002869Z","shell.execute_reply.started":"2024-11-10T05:52:39.998299Z","shell.execute_reply":"2024-11-10T05:52:40.001933Z"}},"outputs":[],"execution_count":81},{"cell_type":"code","source":"dir_path = \"/kaggle/input/chest-xrays-indiana-university\"\ndevice = \"cuda\" if torch.cuda.is_available() else \"cpu\"\nbatch_size = 2\nlr = 5e-4\nepoch = 0\nload_model = True\nsave_model =True","metadata":{"execution":{"iopub.status.busy":"2024-11-10T05:34:38.290528Z","iopub.execute_input":"2024-11-10T05:34:38.290890Z","iopub.status.idle":"2024-11-10T05:34:38.357158Z","shell.execute_reply.started":"2024-11-10T05:34:38.290854Z","shell.execute_reply":"2024-11-10T05:34:38.355652Z"},"trusted":true},"outputs":[],"execution_count":31},{"cell_type":"code","source":"xray_df = pd.read_csv(\"/kaggle/input/chest-xrays-indiana-university/indiana_projections.csv\")\nxray_df =  xray_df[xray_df['projection'] == \"Frontal\"]\nxray_df.reset_index(drop=True, inplace=True)","metadata":{"execution":{"iopub.status.busy":"2024-11-10T05:34:38.358590Z","iopub.execute_input":"2024-11-10T05:34:38.359007Z","iopub.status.idle":"2024-11-10T05:34:38.413743Z","shell.execute_reply.started":"2024-11-10T05:34:38.358961Z","shell.execute_reply":"2024-11-10T05:34:38.412766Z"},"trusted":true},"outputs":[],"execution_count":32},{"cell_type":"code","source":"xray_df","metadata":{"execution":{"iopub.status.busy":"2024-11-10T05:34:38.415013Z","iopub.execute_input":"2024-11-10T05:34:38.415343Z","iopub.status.idle":"2024-11-10T05:34:38.430927Z","shell.execute_reply.started":"2024-11-10T05:34:38.415310Z","shell.execute_reply":"2024-11-10T05:34:38.429991Z"},"trusted":true},"outputs":[{"execution_count":33,"output_type":"execute_result","data":{"text/plain":"       uid                   filename projection\n0        1     1_IM-0001-4001.dcm.png    Frontal\n1        2     2_IM-0652-1001.dcm.png    Frontal\n2        3     3_IM-1384-1001.dcm.png    Frontal\n3        4     4_IM-2050-1001.dcm.png    Frontal\n4        5  5_IM-2117-1003002.dcm.png    Frontal\n...    ...                        ...        ...\n3813  3995  3995_IM-2046-1001.dcm.png    Frontal\n3814  3996  3996_IM-2047-1001.dcm.png    Frontal\n3815  3997  3997_IM-2048-1001.dcm.png    Frontal\n3816  3998  3998_IM-2048-1001.dcm.png    Frontal\n3817  3999  3999_IM-2049-1001.dcm.png    Frontal\n\n[3818 rows x 3 columns]","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>uid</th>\n      <th>filename</th>\n      <th>projection</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>1_IM-0001-4001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>2_IM-0652-1001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3</td>\n      <td>3_IM-1384-1001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>4</td>\n      <td>4_IM-2050-1001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>5</td>\n      <td>5_IM-2117-1003002.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>3813</th>\n      <td>3995</td>\n      <td>3995_IM-2046-1001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>3814</th>\n      <td>3996</td>\n      <td>3996_IM-2047-1001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>3815</th>\n      <td>3997</td>\n      <td>3997_IM-2048-1001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>3816</th>\n      <td>3998</td>\n      <td>3998_IM-2048-1001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n    <tr>\n      <th>3817</th>\n      <td>3999</td>\n      <td>3999_IM-2049-1001.dcm.png</td>\n      <td>Frontal</td>\n    </tr>\n  </tbody>\n</table>\n<p>3818 rows × 3 columns</p>\n</div>"},"metadata":{}}],"execution_count":33},{"cell_type":"code","source":"report_df = pd.read_csv(\"/kaggle/input/chest-xrays-indiana-university/indiana_reports.csv\")\nreport_df['report'] = report_df['findings'] + ' ' + report_df['impression']\nreport_df.drop(['MeSH', 'Problems', 'image', 'indication', 'comparison', 'findings', 'impression'], axis=1, inplace=True)\nreport_df.dropna(inplace=True)\nreport_df.reset_index(drop=True, inplace=True)","metadata":{"execution":{"iopub.status.busy":"2024-11-10T05:34:38.432028Z","iopub.execute_input":"2024-11-10T05:34:38.432373Z","iopub.status.idle":"2024-11-10T05:34:38.502004Z","shell.execute_reply.started":"2024-11-10T05:34:38.432316Z","shell.execute_reply":"2024-11-10T05:34:38.501172Z"},"trusted":true},"outputs":[],"execution_count":34},{"cell_type":"code","source":"report_df","metadata":{"execution":{"iopub.status.busy":"2024-11-10T05:34:38.503171Z","iopub.execute_input":"2024-11-10T05:34:38.503525Z","iopub.status.idle":"2024-11-10T05:34:38.514074Z","shell.execute_reply.started":"2024-11-10T05:34:38.503489Z","shell.execute_reply":"2024-11-10T05:34:38.512996Z"},"trusted":true},"outputs":[{"execution_count":35,"output_type":"execute_result","data":{"text/plain":"       uid                                             report\n0        1  The cardiac silhouette and mediastinum size ar...\n1        2  Borderline cardiomegaly. Midline sternotomy XX...\n2        4  There are diffuse bilateral interstitial and a...\n3        5  The cardiomediastinal silhouette and pulmonary...\n4        6  Heart size and mediastinal contour are within ...\n...    ...                                                ...\n3326  3993  The heart is mildly enlarged. Left hemidiaphra...\n3327  3994  Similar mild cardiomegaly. Of the pulmonary va...\n3328  3995  The cardiomediastinal silhouette and pulmonary...\n3329  3996  The lungs are clear. Heart size is normal. No ...\n3330  3997  Heart size within normal limits. Small, nodula...\n\n[3331 rows x 2 columns]","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>uid</th>\n      <th>report</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>The cardiac silhouette and mediastinum size ar...</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>Borderline cardiomegaly. Midline sternotomy XX...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>4</td>\n      <td>There are diffuse bilateral interstitial and a...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>5</td>\n      <td>The cardiomediastinal silhouette and pulmonary...</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>6</td>\n      <td>Heart size and mediastinal contour are within ...</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>3326</th>\n      <td>3993</td>\n      <td>The heart is mildly enlarged. Left hemidiaphra...</td>\n    </tr>\n    <tr>\n      <th>3327</th>\n      <td>3994</td>\n      <td>Similar mild cardiomegaly. Of the pulmonary va...</td>\n    </tr>\n    <tr>\n      <th>3328</th>\n      <td>3995</td>\n      <td>The cardiomediastinal silhouette and pulmonary...</td>\n    </tr>\n    <tr>\n      <th>3329</th>\n      <td>3996</td>\n      <td>The lungs are clear. Heart size is normal. No ...</td>\n    </tr>\n    <tr>\n      <th>3330</th>\n      <td>3997</td>\n      <td>Heart size within normal limits. Small, nodula...</td>\n    </tr>\n  </tbody>\n</table>\n<p>3331 rows × 2 columns</p>\n</div>"},"metadata":{}}],"execution_count":35},{"cell_type":"code","source":"df = pd.merge(xray_df, report_df, on='uid', how='inner')\ndf.drop(['projection'], axis=1, inplace=True)","metadata":{"execution":{"iopub.status.busy":"2024-11-10T05:34:38.515468Z","iopub.execute_input":"2024-11-10T05:34:38.515834Z","iopub.status.idle":"2024-11-10T05:34:38.532130Z","shell.execute_reply.started":"2024-11-10T05:34:38.515789Z","shell.execute_reply":"2024-11-10T05:34:38.531315Z"},"trusted":true},"outputs":[],"execution_count":36},{"cell_type":"code","source":"df","metadata":{"execution":{"iopub.status.busy":"2024-11-10T05:34:38.533502Z","iopub.execute_input":"2024-11-10T05:34:38.533870Z","iopub.status.idle":"2024-11-10T05:34:38.544567Z","shell.execute_reply.started":"2024-11-10T05:34:38.533818Z","shell.execute_reply":"2024-11-10T05:34:38.543681Z"},"trusted":true},"outputs":[{"execution_count":37,"output_type":"execute_result","data":{"text/plain":"       uid                   filename  \\\n0        1     1_IM-0001-4001.dcm.png   \n1        2     2_IM-0652-1001.dcm.png   \n2        4     4_IM-2050-1001.dcm.png   \n3        5  5_IM-2117-1003002.dcm.png   \n4        6     6_IM-2192-1001.dcm.png   \n...    ...                        ...   \n3296  3993  3993_IM-2044-1001.dcm.png   \n3297  3994  3994_IM-2045-2001.dcm.png   \n3298  3995  3995_IM-2046-1001.dcm.png   \n3299  3996  3996_IM-2047-1001.dcm.png   \n3300  3997  3997_IM-2048-1001.dcm.png   \n\n                                                 report  \n0     The cardiac silhouette and mediastinum size ar...  \n1     Borderline cardiomegaly. Midline sternotomy XX...  \n2     There are diffuse bilateral interstitial and a...  \n3     The cardiomediastinal silhouette and pulmonary...  \n4     Heart size and mediastinal contour are within ...  \n...                                                 ...  \n3296  The heart is mildly enlarged. Left hemidiaphra...  \n3297  Similar mild cardiomegaly. Of the pulmonary va...  \n3298  The cardiomediastinal silhouette and pulmonary...  \n3299  The lungs are clear. Heart size is normal. No ...  \n3300  Heart size within normal limits. Small, nodula...  \n\n[3301 rows x 3 columns]","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>uid</th>\n      <th>filename</th>\n      <th>report</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>1_IM-0001-4001.dcm.png</td>\n      <td>The cardiac silhouette and mediastinum size ar...</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>2_IM-0652-1001.dcm.png</td>\n      <td>Borderline cardiomegaly. Midline sternotomy XX...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>4</td>\n      <td>4_IM-2050-1001.dcm.png</td>\n      <td>There are diffuse bilateral interstitial and a...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>5</td>\n      <td>5_IM-2117-1003002.dcm.png</td>\n      <td>The cardiomediastinal silhouette and pulmonary...</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>6</td>\n      <td>6_IM-2192-1001.dcm.png</td>\n      <td>Heart size and mediastinal contour are within ...</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>3296</th>\n      <td>3993</td>\n      <td>3993_IM-2044-1001.dcm.png</td>\n      <td>The heart is mildly enlarged. Left hemidiaphra...</td>\n    </tr>\n    <tr>\n      <th>3297</th>\n      <td>3994</td>\n      <td>3994_IM-2045-2001.dcm.png</td>\n      <td>Similar mild cardiomegaly. Of the pulmonary va...</td>\n    </tr>\n    <tr>\n      <th>3298</th>\n      <td>3995</td>\n      <td>3995_IM-2046-1001.dcm.png</td>\n      <td>The cardiomediastinal silhouette and pulmonary...</td>\n    </tr>\n    <tr>\n      <th>3299</th>\n      <td>3996</td>\n      <td>3996_IM-2047-1001.dcm.png</td>\n      <td>The lungs are clear. Heart size is normal. No ...</td>\n    </tr>\n    <tr>\n      <th>3300</th>\n      <td>3997</td>\n      <td>3997_IM-2048-1001.dcm.png</td>\n      <td>Heart size within normal limits. Small, nodula...</td>\n    </tr>\n  </tbody>\n</table>\n<p>3301 rows × 3 columns</p>\n</div>"},"metadata":{}}],"execution_count":37},{"cell_type":"code","source":"def load_image(file_path):\n    try:\n        image = Image.open(f\"{dir_path}/images/images_normalized/{file_path}\")\n        return image\n    except Exception as e:\n        print(f\"Error loading image: {e}\")\n        return None\n\ndf['image'] = df['filename'].apply(load_image)","metadata":{"execution":{"iopub.status.busy":"2024-11-10T05:34:38.545734Z","iopub.execute_input":"2024-11-10T05:34:38.546038Z","iopub.status.idle":"2024-11-10T05:35:33.291480Z","shell.execute_reply.started":"2024-11-10T05:34:38.546006Z","shell.execute_reply":"2024-11-10T05:35:33.290662Z"},"trusted":true},"outputs":[],"execution_count":38},{"cell_type":"code","source":"df.drop([\"filename\"], axis=1, inplace=True)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:35:33.295624Z","iopub.execute_input":"2024-11-10T05:35:33.295932Z","iopub.status.idle":"2024-11-10T05:35:33.301615Z","shell.execute_reply.started":"2024-11-10T05:35:33.295900Z","shell.execute_reply":"2024-11-10T05:35:33.300665Z"}},"outputs":[],"execution_count":39},{"cell_type":"code","source":"df","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:35:33.302871Z","iopub.execute_input":"2024-11-10T05:35:33.303212Z","iopub.status.idle":"2024-11-10T05:35:33.320823Z","shell.execute_reply.started":"2024-11-10T05:35:33.303138Z","shell.execute_reply":"2024-11-10T05:35:33.319694Z"}},"outputs":[{"execution_count":40,"output_type":"execute_result","data":{"text/plain":"       uid                                             report  \\\n0        1  The cardiac silhouette and mediastinum size ar...   \n1        2  Borderline cardiomegaly. Midline sternotomy XX...   \n2        4  There are diffuse bilateral interstitial and a...   \n3        5  The cardiomediastinal silhouette and pulmonary...   \n4        6  Heart size and mediastinal contour are within ...   \n...    ...                                                ...   \n3296  3993  The heart is mildly enlarged. Left hemidiaphra...   \n3297  3994  Similar mild cardiomegaly. Of the pulmonary va...   \n3298  3995  The cardiomediastinal silhouette and pulmonary...   \n3299  3996  The lungs are clear. Heart size is normal. No ...   \n3300  3997  Heart size within normal limits. Small, nodula...   \n\n                                                  image  \n0     <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n1     <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n2     <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n3     <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n4     <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n...                                                 ...  \n3296  <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n3297  <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n3298  <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n3299  <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n3300  <PIL.PngImagePlugin.PngImageFile image mode=L ...  \n\n[3301 rows x 3 columns]","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>uid</th>\n      <th>report</th>\n      <th>image</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>The cardiac silhouette and mediastinum size ar...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>Borderline cardiomegaly. Midline sternotomy XX...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>4</td>\n      <td>There are diffuse bilateral interstitial and a...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>5</td>\n      <td>The cardiomediastinal silhouette and pulmonary...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>6</td>\n      <td>Heart size and mediastinal contour are within ...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>3296</th>\n      <td>3993</td>\n      <td>The heart is mildly enlarged. Left hemidiaphra...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>3297</th>\n      <td>3994</td>\n      <td>Similar mild cardiomegaly. Of the pulmonary va...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>3298</th>\n      <td>3995</td>\n      <td>The cardiomediastinal silhouette and pulmonary...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>3299</th>\n      <td>3996</td>\n      <td>The lungs are clear. Heart size is normal. No ...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n    <tr>\n      <th>3300</th>\n      <td>3997</td>\n      <td>Heart size within normal limits. Small, nodula...</td>\n      <td>&lt;PIL.PngImagePlugin.PngImageFile image mode=L ...</td>\n    </tr>\n  </tbody>\n</table>\n<p>3301 rows × 3 columns</p>\n</div>"},"metadata":{}}],"execution_count":40},{"cell_type":"code","source":"processor = AutoProcessor.from_pretrained(\"Salesforce/blip2-opt-2.7b\")\nmodel = Blip2ForConditionalGeneration.from_pretrained(\"ybelkada/blip2-opt-2.7b-fp16-sharded\", \n                                                      device_map=\"auto\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:35:33.322000Z","iopub.execute_input":"2024-11-10T05:35:33.322328Z","iopub.status.idle":"2024-11-10T05:37:56.283125Z","shell.execute_reply.started":"2024-11-10T05:35:33.322295Z","shell.execute_reply":"2024-11-10T05:37:56.282288Z"}},"outputs":[{"output_type":"display_data","data":{"text/plain":"preprocessor_config.json:   0%|          | 0.00/432 [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"84daa9e2f16c405f9b895d3c6da2b80a"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"tokenizer_config.json:   0%|          | 0.00/904 [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"c4f1309513df45cfb187297aa667a674"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"vocab.json:   0%|          | 0.00/798k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"014dd2030bb646999e3cb491f991c6bd"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"merges.txt:   0%|          | 0.00/456k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"2271b489e3dd4c51ab04e19d1f0c85a1"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"tokenizer.json:   0%|          | 0.00/2.11M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"db24e9d5b1da4d8f8c787d105517dfce"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"special_tokens_map.json:   0%|          | 0.00/548 [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"4d89b9f17eea43dd88fb42fdb6a13069"}},"metadata":{}},{"name":"stderr","text":"/opt/conda/lib/python3.10/site-packages/transformers/tokenization_utils_base.py:1617: FutureWarning: `clean_up_tokenization_spaces` was not set. It will be set to `True` by default. This behavior will be deprecated in transformers v4.45, and will be then set to `False` by default. For more details check this issue: https://github.com/huggingface/transformers/issues/31884\n  warnings.warn(\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"config.json:   0%|          | 0.00/7.05k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"a29a6f86613d47f59d9c1300d88de5dc"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model.bin.index.json:   0%|          | 0.00/122k [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"75c9d86213ba4c47ade7b6c8fc505f44"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Downloading shards:   0%|          | 0/8 [00:00<?, ?it/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"65ac173814534e23bac8f3059e479000"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model-00001-of-00008.bin:   0%|          | 0.00/995M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"f051341948dc4a36937e65d97f0f888c"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model-00002-of-00008.bin:   0%|          | 0.00/999M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"1bf152e3ee1b428ea86a4a8c7b72e6fc"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model-00003-of-00008.bin:   0%|          | 0.00/985M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"68131de564234217b1e9edd56212d9e8"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model-00004-of-00008.bin:   0%|          | 0.00/997M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"93c72151b41e4b3eaddbaafa084a8e10"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model-00005-of-00008.bin:   0%|          | 0.00/997M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"7fc9e4b090104c44b816bf9ef14ebc9c"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model-00006-of-00008.bin:   0%|          | 0.00/997M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"9f98175411474878952238b851490f2c"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model-00007-of-00008.bin:   0%|          | 0.00/997M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"e22dd986b2b345619432b1d12c102bce"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"pytorch_model-00008-of-00008.bin:   0%|          | 0.00/782M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"e76517af0fd6407a9e4e6c420b065d2f"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"Loading checkpoint shards:   0%|          | 0/8 [00:00<?, ?it/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"8220db8ece20428f93c431a0097c14de"}},"metadata":{}}],"execution_count":41},{"cell_type":"code","source":"class MedCapDataset(Dataset):\n    def __init__(self, df, processor):\n        self.df = df\n        self.len_df = df.shape[0]\n        self.processor = processor\n    \n    def __len__(self):\n        return self.len_df\n    \n    def __getitem__(self, idx):\n        item = self.df.iloc[idx]\n        encoding = self.processor(images=item[\"image\"], \n                                  padding=\"max_length\", \n                                  return_tensors=\"pt\")\n        encoding = {k: v.squeeze() for k, v in encoding.items()}\n        encoding[\"report\"] = item[\"report\"]\n        return encoding","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:37:56.284261Z","iopub.execute_input":"2024-11-10T05:37:56.284584Z","iopub.status.idle":"2024-11-10T05:37:56.293260Z","shell.execute_reply.started":"2024-11-10T05:37:56.284551Z","shell.execute_reply":"2024-11-10T05:37:56.292285Z"}},"outputs":[],"execution_count":42},{"cell_type":"code","source":"def collate_fn(batch):\n    # pad the input_ids and attention_mask\n    processed_batch = {}\n    for key in batch[0].keys():\n        if key != \"report\":\n            processed_batch[key] = torch.stack([example[key] for example in batch])\n        else:\n            text_inputs = processor.tokenizer(\n                [example[\"report\"] for example in batch], padding=True, return_tensors=\"pt\"\n            )\n            processed_batch[\"input_ids\"] = text_inputs[\"input_ids\"]\n            processed_batch[\"attention_mask\"] = text_inputs[\"attention_mask\"]\n    return processed_batch","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:37:56.294655Z","iopub.execute_input":"2024-11-10T05:37:56.294931Z","iopub.status.idle":"2024-11-10T05:37:56.304089Z","shell.execute_reply.started":"2024-11-10T05:37:56.294899Z","shell.execute_reply":"2024-11-10T05:37:56.303275Z"}},"outputs":[],"execution_count":43},{"cell_type":"code","source":"def freeze_model_except_qformer(model):\n    \"\"\"Freeze all parameters except Q-Former\"\"\"\n    # Freeze vision encoder\n    for param in model.vision_model.parameters():\n        param.requires_grad = False\n    \n    # Freeze language model\n    for param in model.language_model.parameters():\n        param.requires_grad = False\n    \n    # Only keep Q-Former parameters trainable\n    for param in model.qformer.parameters():\n        param.requires_grad = True","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:37:56.305460Z","iopub.execute_input":"2024-11-10T05:37:56.305847Z","iopub.status.idle":"2024-11-10T05:37:56.314036Z","shell.execute_reply.started":"2024-11-10T05:37:56.305801Z","shell.execute_reply":"2024-11-10T05:37:56.313184Z"}},"outputs":[],"execution_count":44},{"cell_type":"code","source":"def save_checkpoint(model, optimizer, epoch, loss, filename='checkpoint.pth'):\n    checkpoint = {\n        'epoch': epoch,\n        'model_state_dict': model.state_dict(),\n        'optimizer_state_dict': optimizer.state_dict(),\n        'loss': loss,\n    }\n    torch.save(checkpoint, filename)\n    print(f\"Checkpoint saved to {filename}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:37:56.315375Z","iopub.execute_input":"2024-11-10T05:37:56.316203Z","iopub.status.idle":"2024-11-10T05:37:56.323275Z","shell.execute_reply.started":"2024-11-10T05:37:56.316152Z","shell.execute_reply":"2024-11-10T05:37:56.322387Z"}},"outputs":[],"execution_count":45},{"cell_type":"code","source":"def load_checkpoint(model, optimizer, lr, filename='checkpoint.pth'):\n    checkpoint = torch.load(filename)\n    model.load_state_dict(checkpoint['model_state_dict'])\n    optimizer.load_state_dict(checkpoint['optimizer_state_dict'])\n    epoch = checkpoint['epoch']\n    loss = checkpoint['loss']\n    for param_group in optimizer.param_groups:\n        param_group[\"lr\"] = lr\n    print(f\"Checkpoint loaded from {filename}, epoch {epoch}, loss {loss}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:37:56.324560Z","iopub.execute_input":"2024-11-10T05:37:56.324879Z","iopub.status.idle":"2024-11-10T05:37:56.335966Z","shell.execute_reply.started":"2024-11-10T05:37:56.324848Z","shell.execute_reply":"2024-11-10T05:37:56.335129Z"}},"outputs":[],"execution_count":46},{"cell_type":"code","source":"autocast = torch.amp.autocast('cuda',enabled=True, dtype=torch.half)\nscaler = torch.amp.GradScaler('cuda', enabled=True, init_scale=4096)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:37:56.337163Z","iopub.execute_input":"2024-11-10T05:37:56.337649Z","iopub.status.idle":"2024-11-10T05:37:56.346058Z","shell.execute_reply.started":"2024-11-10T05:37:56.337606Z","shell.execute_reply":"2024-11-10T05:37:56.345170Z"}},"outputs":[],"execution_count":47},{"cell_type":"code","source":"def train(model, processor, optimizer, train_dl, epoch):\n    if load_model:\n        load_checkpoint(model, optimizer, lr, \"/kaggle/input/medcapcheckpoints/checkpoint-2.pth\")\n        model.float()\n    model.train()\n    for epoch in range(epoch):\n        print(\"Epoch: \", epoch+1)\n        total_loss = 0\n        loop =tqdm(train_dl, leave=True)\n        for idx, batch in enumerate(loop):\n            input_ids = batch[\"input_ids\"].to(model.device)\n            pixel_vals = batch[\"pixel_values\"].to(model.device, torch.float16)\n            optimizer.zero_grad()\n            with autocast:\n                outputs = model(input_ids=input_ids, pixel_values=pixel_vals, \n                                labels=input_ids)\n                loss = outputs.loss\n            loop.set_description(f\"Loss: {loss.item():.4f}\")\n            total_loss += loss.item()\n            \n            scaler.scale(loss).backward()\n            scaler.step(optimizer)\n            scaler.update() \n            \n        # Average the loss over all batches\n        avg_loss = total_loss / len(train_dl)  # Dividing by number of batches\n        print(\"Loss:\", avg_loss)\n        \n        if save_model and (epoch+1)%10==0:\n            save_checkpoint(model.half(), optimizer, epoch+1, avg_loss, f\"checkpoint.pth\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T07:03:29.844521Z","iopub.execute_input":"2024-11-10T07:03:29.845399Z","iopub.status.idle":"2024-11-10T07:03:29.854653Z","shell.execute_reply.started":"2024-11-10T07:03:29.845358Z","shell.execute_reply":"2024-11-10T07:03:29.853701Z"}},"outputs":[],"execution_count":117},{"cell_type":"code","source":"def main():\n    train_ds = MedCapDataset(df, processor)\n    train_dl = DataLoader(train_ds, shuffle=True, batch_size=batch_size, collate_fn=collate_fn)\n    \n    freeze_model_except_qformer(model)\n    \n    optimizer = torch.optim.AdamW([p for p in model.parameters() if p.requires_grad], lr=lr)\n    \n    train(model, processor, optimizer, train_dl, epoch)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T07:03:31.997831Z","iopub.execute_input":"2024-11-10T07:03:31.998214Z","iopub.status.idle":"2024-11-10T07:03:32.004381Z","shell.execute_reply.started":"2024-11-10T07:03:31.998171Z","shell.execute_reply":"2024-11-10T07:03:32.003401Z"}},"outputs":[],"execution_count":118},{"cell_type":"code","source":"if __name__ == \"__main__\":\n    main()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T07:03:34.552625Z","iopub.execute_input":"2024-11-10T07:03:34.553024Z","iopub.status.idle":"2024-11-10T07:04:18.934061Z","shell.execute_reply.started":"2024-11-10T07:03:34.552985Z","shell.execute_reply":"2024-11-10T07:04:18.933067Z"}},"outputs":[{"name":"stderr","text":"/tmp/ipykernel_30/66246431.py:2: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.\n  checkpoint = torch.load(filename)\n","output_type":"stream"},{"name":"stdout","text":"Checkpoint loaded from /kaggle/input/medcapcheckpoints/checkpoint-2.pth, epoch 15, loss 1.8834556452004856\n","output_type":"stream"}],"execution_count":119},{"cell_type":"code","source":"example = df.iloc[550]\nimage = example[\"image\"]\ninputs = processor(images=image, return_tensors=\"pt\").to(device, torch.float16)\npixel_values = inputs.pixel_values\n\ngenerated_ids = model.generate(pixel_values=pixel_values, \n                               max_length=150,\n                               num_beams=3,              # Number of beams to use (try 3-5 for better results)\n                               no_repeat_ngram_size=3,   # Avoid repeating n-grams\n                               length_penalty=2.0,       # Longer sequences get bonus (>1.0 favors longer sequences)\n                               early_stopping=True,      # Stop when all beams reach EOS\n                               temperature=0.7,          # Lower temperature for more focused sampling\n                               top_k=100,                # Limit vocabulary to top k tokens\n                               top_p=0.95)\ngenerated_caption = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]\nprint(generated_caption)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T07:06:43.750253Z","iopub.execute_input":"2024-11-10T07:06:43.751162Z","iopub.status.idle":"2024-11-10T07:06:53.962634Z","shell.execute_reply.started":"2024-11-10T07:06:43.751118Z","shell.execute_reply":"2024-11-10T07:06:53.961713Z"}},"outputs":[{"name":"stderr","text":"The `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\n","output_type":"stream"},{"name":"stdout","text":"The cardiomediastinal silhouette is within normal limits. There is no evidence of cardiomegaly, pneumothorax, or pleural effusion. The lungs are normal. No evidence of acute cardiopulmonary disease. No acute cardiovasculatory abnormality. 1. Normal cardiometabolic parameters. 2. No cardiovascular abnormality or acute cardiomyopathy. 3. Abnormalities of the thoracic aorta. 4. Abnormality of the aortic arch. 5. Abdominal pain. 6. Abundant pleural fluid. 7. No pulmonary embolism. 8. No abnormality of pulmonary function. 9. No indication of acute\n","output_type":"stream"}],"execution_count":126},{"cell_type":"code","source":"plt.imshow(image)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T07:09:54.688281Z","iopub.execute_input":"2024-11-10T07:09:54.689175Z","iopub.status.idle":"2024-11-10T07:09:55.515999Z","shell.execute_reply.started":"2024-11-10T07:09:54.689132Z","shell.execute_reply":"2024-11-10T07:09:55.515024Z"}},"outputs":[{"execution_count":127,"output_type":"execute_result","data":{"text/plain":"<matplotlib.image.AxesImage at 0x7b40841ae170>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 640x480 with 1 Axes>","image/png":"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"},"metadata":{}}],"execution_count":127},{"cell_type":"code","source":"example[\"report\"]","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T07:05:33.012717Z","iopub.execute_input":"2024-11-10T07:05:33.013115Z","iopub.status.idle":"2024-11-10T07:05:33.019322Z","shell.execute_reply.started":"2024-11-10T07:05:33.013076Z","shell.execute_reply":"2024-11-10T07:05:33.018422Z"}},"outputs":[{"execution_count":123,"output_type":"execute_result","data":{"text/plain":"'Heart size and pulmonary vascularity appear within normal limits. There is mild tortuosity to the descending thoracic aorta. The lungs are free of focal airspace disease. No pleural effusion or pneumothorax is seen. No discrete nodules or adenopathy are noted. Degenerative changes are present in the spine. No evidence of active disease.'"},"metadata":{}}],"execution_count":123},{"cell_type":"code","source":"from nltk.translate.bleu_score import sentence_bleu\n\nscore = sentence_bleu([example[\"report\"]], generated_caption, weights = [0.5, 0.5, 0, 0])\nprint(score)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:45:48.918974Z","iopub.execute_input":"2024-11-10T05:45:48.919732Z","iopub.status.idle":"2024-11-10T05:45:48.927065Z","shell.execute_reply.started":"2024-11-10T05:45:48.919689Z","shell.execute_reply":"2024-11-10T05:45:48.926124Z"}},"outputs":[{"name":"stdout","text":"0.4698010349900459\n","output_type":"stream"}],"execution_count":66},{"cell_type":"code","source":"from rouge_score import rouge_scorer\n\n# Example sentences\nreference = \"The cat is on the mat.\"\nhypothesis = \"There is a cat on the mat.\"\n\n# Initialize scorer\nscorer = rouge_scorer.RougeScorer(['rouge1', 'rouge2', 'rougeL'], use_stemmer=True)\nscores = scorer.score(reference, hypothesis)\nprint(scores['rouge1'])","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:58:05.641202Z","iopub.execute_input":"2024-11-10T05:58:05.641966Z","iopub.status.idle":"2024-11-10T05:58:05.648894Z","shell.execute_reply.started":"2024-11-10T05:58:05.641923Z","shell.execute_reply":"2024-11-10T05:58:05.647776Z"}},"outputs":[{"name":"stdout","text":"Score(precision=0.7142857142857143, recall=0.8333333333333334, fmeasure=0.7692307692307692)\n","output_type":"stream"}],"execution_count":92},{"cell_type":"code","source":"nltk.download('wordnet')","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:40:27.963984Z","iopub.execute_input":"2024-11-10T05:40:27.964436Z","iopub.status.idle":"2024-11-10T05:40:27.971691Z","shell.execute_reply.started":"2024-11-10T05:40:27.964395Z","shell.execute_reply":"2024-11-10T05:40:27.970721Z"}},"outputs":[{"name":"stdout","text":"[nltk_data] Downloading package wordnet to /usr/share/nltk_data...\n[nltk_data]   Package wordnet is already up-to-date!\n","output_type":"stream"},{"execution_count":56,"output_type":"execute_result","data":{"text/plain":"True"},"metadata":{}}],"execution_count":56},{"cell_type":"code","source":"hypothesis = []\nreferences = []\nrscorel = 0\nrscore1 = 0\nrscore2 = 0\n#mscore = 0\nfor i in range(329):\n    example = df.iloc[2971+i]\n    image = example[\"image\"]\n    inputs = processor(images=image, return_tensors=\"pt\").to(device, torch.float16)\n    pixel_values = inputs.pixel_values\n    generated_ids = model.generate(pixel_values=pixel_values, \n                               max_length=100,\n                               num_beams=3,              # Number of beams to use (try 3-5 for better results)\n                               no_repeat_ngram_size=2,   # Avoid repeating n-grams\n                               length_penalty=2.0,       # Longer sequences get bonus (>1.0 favors longer sequences)\n                               early_stopping=True,      # Stop when all beams reach EOS\n                               temperature=0.7,          # Lower temperature for more focused sampling\n                               top_k=50,                # Limit vocabulary to top k tokens\n                               top_p=0.95)\n    generated_caption = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]\n    scorer = rouge_scorer.RougeScorer(['rouge1', 'rouge2', 'rougeL'], use_stemmer=True)\n    scores = scorer.score(example[\"report\"], generated_caption)\n    #mscore += meteor_score([word_tokenize(example[\"report\"])], word_tokenize(generated_caption))\n    rscorel += scores['rougeL'].fmeasure\n    rscore1 += scores['rouge1'].fmeasure\n    rscore2 += scores['rouge2'].fmeasure\n    hypothesis.append(generated_caption)\n    references.append([example[\"report\"]])","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T06:03:45.548883Z","iopub.execute_input":"2024-11-10T06:03:45.549882Z","iopub.status.idle":"2024-11-10T06:40:48.355764Z","shell.execute_reply.started":"2024-11-10T06:03:45.549837Z","shell.execute_reply":"2024-11-10T06:40:48.354721Z"},"collapsed":true,"jupyter":{"outputs_hidden":true}},"outputs":[{"name":"stderr","text":"The `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\nThe `language_model` is not in the `hf_device_map` dictionary and you are running your script in a multi-GPU environment. this may lead to unexpected behavior when using `accelerate`. Please pass a `device_map` that contains `language_model` to remove this warning. Please refer to https://github.com/huggingface/blog/blob/main/accelerate-large-models.md for more details on creating a `device_map` for large models.\n","output_type":"stream"}],"execution_count":93},{"cell_type":"code","source":"rscorel/330","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T06:42:07.131701Z","iopub.execute_input":"2024-11-10T06:42:07.132132Z","iopub.status.idle":"2024-11-10T06:42:07.138556Z","shell.execute_reply.started":"2024-11-10T06:42:07.132091Z","shell.execute_reply":"2024-11-10T06:42:07.137676Z"}},"outputs":[{"execution_count":94,"output_type":"execute_result","data":{"text/plain":"0.24079259722149274"},"metadata":{}}],"execution_count":94},{"cell_type":"code","source":"rscore1/330","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T06:42:18.127381Z","iopub.execute_input":"2024-11-10T06:42:18.128305Z","iopub.status.idle":"2024-11-10T06:42:18.134428Z","shell.execute_reply.started":"2024-11-10T06:42:18.128252Z","shell.execute_reply":"2024-11-10T06:42:18.133290Z"}},"outputs":[{"execution_count":95,"output_type":"execute_result","data":{"text/plain":"0.3534435558595618"},"metadata":{}}],"execution_count":95},{"cell_type":"code","source":"rscore2/330","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T06:42:26.336764Z","iopub.execute_input":"2024-11-10T06:42:26.337179Z","iopub.status.idle":"2024-11-10T06:42:26.343683Z","shell.execute_reply.started":"2024-11-10T06:42:26.337140Z","shell.execute_reply":"2024-11-10T06:42:26.342685Z"}},"outputs":[{"execution_count":96,"output_type":"execute_result","data":{"text/plain":"0.13852719690886062"},"metadata":{}}],"execution_count":96},{"cell_type":"code","source":"from nltk.translate.bleu_score import corpus_bleu, sentence_bleu\n\nbleu1 = corpus_bleu(references, hypothesis, weights=(1.0, 0, 0, 0))\nbleu2 = corpus_bleu(references, hypothesis, weights=(0.5, 0.5, 0, 0))\nbleu3 = corpus_bleu(references, hypothesis, weights=(0.33, 0.33, 0.33, 0))\nbleu4 = corpus_bleu(references, hypothesis, weights=(0.25, 0.25, 0.25, 0.25))\n    \nprint(f\"BLEU-1: {bleu1}\")\nprint(f\"BLEU-2: {bleu2}\")\nprint(f\"BLEU-3: {bleu3}\")\nprint(f\"BLEU-4: {bleu4}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T06:42:33.487728Z","iopub.execute_input":"2024-11-10T06:42:33.488144Z","iopub.status.idle":"2024-11-10T06:42:35.826227Z","shell.execute_reply.started":"2024-11-10T06:42:33.488102Z","shell.execute_reply":"2024-11-10T06:42:35.825232Z"}},"outputs":[{"name":"stdout","text":"BLEU-1: 0.6579938560170963\nBLEU-2: 0.5544542413130831\nBLEU-3: 0.4627318887549421\nBLEU-4: 0.39410560298673497\n","output_type":"stream"}],"execution_count":97},{"cell_type":"code","source":"FileLink(r'checkpoint.pth')","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2024-11-10T05:38:56.703310Z","iopub.status.idle":"2024-11-10T05:38:56.703807Z","shell.execute_reply.started":"2024-11-10T05:38:56.703563Z","shell.execute_reply":"2024-11-10T05:38:56.703589Z"}},"outputs":[],"execution_count":null}]}