{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.10","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"**This notebook is a baseline for fine-tuning the `esm2` model.**\n\n- I used a `Colab Pro+ V100 high-memory instance` for training and inference.\n- I made some modifications to the evaluation code provided by the competition host (such as minor adjustments like float to float16).\n- The hyperparameter settings are arbitrary.\n\n\n**Reference code & discussions:**\n- https://www.kaggle.com/code/gusthema/cafa-5-protein-function-with-tensorflow\n- https://www.kaggle.com/code/henriupton/proteinet-pytorch-ems2-t5-protbert-embeddings\n- https://www.kaggle.com/code/sergeifironov/validate-ridge\n- https://www.kaggle.com/competitions/cafa-5-protein-function-prediction/discussion/407220\n- https://www.kaggle.com/code/yasufuminakama/fb3-deberta-v3-base-baseline-train (NLP pipeline)","metadata":{}},{"cell_type":"code","source":"# make submission (run only this cell)\nimport pandas as pd\nsubmission = pd.read_csv(\"/kaggle/input/cafa5-exp001/submission.tsv\", sep=\"\\t\", header=None)\nsubmission.to_csv(\"submission.tsv\",sep=\"\\t\", header=False, index=False)","metadata":{"execution":{"iopub.status.busy":"2026-08-14T11:11:04.902079Z","iopub.execute_input":"2026-08-14T11:11:04.902423Z","iopub.status.idle":"2026-08-14T11:11:05.183139Z","shell.execute_reply.started":"2026-08-14T11:11:04.902401Z","shell.execute_reply":"2026-08-14T11:11:05.181739Z"},"trusted":true},"outputs":[{"traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mFileNotFoundError\u001b[0m Traceback (most recent call last)","Cell \u001b[0;32mIn[134], line 3\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[38;5;66;03m# make submission (run only this cell)\u001b[39;00m\n\u001b[1;32m 2\u001b[0m \u001b[38;5;28;01mimport\u001b[39;00m \u001b[38;5;21;01mpandas\u001b[39;00m \u001b[38;5;28;01mas\u001b[39;00m \u001b[38;5;21;01mpd\u001b[39;00m\n\u001b[0;32m----> 3\u001b[0m submission \u001b[38;5;241m=\u001b[39m \u001b[43mpd\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mread_csv\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43m/kaggle/input/cafa5-exp001/submission.tsv\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msep\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;130;43;01m\\t\u001b[39;49;00m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mheader\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m)\u001b[49m\n\u001b[1;32m 4\u001b[0m submission\u001b[38;5;241m.\u001b[39mto_csv(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124msubmission.tsv\u001b[39m\u001b[38;5;124m\"\u001b[39m,sep\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;130;01m\\t\u001b[39;00m\u001b[38;5;124m\"\u001b[39m, header\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mFalse\u001b[39;00m, index\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mFalse\u001b[39;00m)\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/pandas/util/_decorators.py:211\u001b[0m, in \u001b[0;36mdeprecate_kwarg.._deprecate_kwarg..wrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m 209\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 210\u001b[0m kwargs[new_arg_name] \u001b[38;5;241m=\u001b[39m new_arg_value\n\u001b[0;32m--> 211\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfunc\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/pandas/util/_decorators.py:331\u001b[0m, in \u001b[0;36mdeprecate_nonkeyword_arguments..decorate..wrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m 325\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mlen\u001b[39m(args) \u001b[38;5;241m>\u001b[39m num_allow_args:\n\u001b[1;32m 326\u001b[0m warnings\u001b[38;5;241m.\u001b[39mwarn(\n\u001b[1;32m 327\u001b[0m msg\u001b[38;5;241m.\u001b[39mformat(arguments\u001b[38;5;241m=\u001b[39m_format_argument_list(allow_args)),\n\u001b[1;32m 328\u001b[0m \u001b[38;5;167;01mFutureWarning\u001b[39;00m,\n\u001b[1;32m 329\u001b[0m stacklevel\u001b[38;5;241m=\u001b[39mfind_stack_level(),\n\u001b[1;32m 330\u001b[0m )\n\u001b[0;32m--> 331\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfunc\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/pandas/io/parsers/readers.py:950\u001b[0m, in \u001b[0;36mread_csv\u001b[0;34m(filepath_or_buffer, sep, delimiter, header, names, index_col, usecols, squeeze, prefix, mangle_dupe_cols, dtype, engine, converters, true_values, false_values, skipinitialspace, skiprows, skipfooter, nrows, na_values, keep_default_na, na_filter, verbose, skip_blank_lines, parse_dates, infer_datetime_format, keep_date_col, date_parser, dayfirst, cache_dates, iterator, chunksize, compression, thousands, decimal, lineterminator, quotechar, quoting, doublequote, escapechar, comment, encoding, encoding_errors, dialect, error_bad_lines, warn_bad_lines, on_bad_lines, delim_whitespace, low_memory, memory_map, float_precision, storage_options)\u001b[0m\n\u001b[1;32m 935\u001b[0m kwds_defaults \u001b[38;5;241m=\u001b[39m _refine_defaults_read(\n\u001b[1;32m 936\u001b[0m dialect,\n\u001b[1;32m 937\u001b[0m delimiter,\n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 946\u001b[0m defaults\u001b[38;5;241m=\u001b[39m{\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mdelimiter\u001b[39m\u001b[38;5;124m\"\u001b[39m: \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m,\u001b[39m\u001b[38;5;124m\"\u001b[39m},\n\u001b[1;32m 947\u001b[0m )\n\u001b[1;32m 948\u001b[0m kwds\u001b[38;5;241m.\u001b[39mupdate(kwds_defaults)\n\u001b[0;32m--> 950\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43m_read\u001b[49m\u001b[43m(\u001b[49m\u001b[43mfilepath_or_buffer\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mkwds\u001b[49m\u001b[43m)\u001b[49m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/pandas/io/parsers/readers.py:605\u001b[0m, in \u001b[0;36m_read\u001b[0;34m(filepath_or_buffer, kwds)\u001b[0m\n\u001b[1;32m 602\u001b[0m _validate_names(kwds\u001b[38;5;241m.\u001b[39mget(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mnames\u001b[39m\u001b[38;5;124m\"\u001b[39m, \u001b[38;5;28;01mNone\u001b[39;00m))\n\u001b[1;32m 604\u001b[0m \u001b[38;5;66;03m# Create the parser.\u001b[39;00m\n\u001b[0;32m--> 605\u001b[0m parser \u001b[38;5;241m=\u001b[39m \u001b[43mTextFileReader\u001b[49m\u001b[43m(\u001b[49m\u001b[43mfilepath_or_buffer\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwds\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 607\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m chunksize \u001b[38;5;129;01mor\u001b[39;00m iterator:\n\u001b[1;32m 608\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m parser\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/pandas/io/parsers/readers.py:1442\u001b[0m, in \u001b[0;36mTextFileReader.__init__\u001b[0;34m(self, f, engine, **kwds)\u001b[0m\n\u001b[1;32m 1439\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39moptions[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mhas_index_names\u001b[39m\u001b[38;5;124m\"\u001b[39m] \u001b[38;5;241m=\u001b[39m kwds[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mhas_index_names\u001b[39m\u001b[38;5;124m\"\u001b[39m]\n\u001b[1;32m 1441\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mhandles: IOHandles \u001b[38;5;241m|\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[0;32m-> 1442\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_engine \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_make_engine\u001b[49m\u001b[43m(\u001b[49m\u001b[43mf\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mengine\u001b[49m\u001b[43m)\u001b[49m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/pandas/io/parsers/readers.py:1735\u001b[0m, in \u001b[0;36mTextFileReader._make_engine\u001b[0;34m(self, f, engine)\u001b[0m\n\u001b[1;32m 1733\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mb\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;129;01min\u001b[39;00m mode:\n\u001b[1;32m 1734\u001b[0m mode \u001b[38;5;241m+\u001b[39m\u001b[38;5;241m=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mb\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[0;32m-> 1735\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mhandles \u001b[38;5;241m=\u001b[39m \u001b[43mget_handle\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 1736\u001b[0m \u001b[43m \u001b[49m\u001b[43mf\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1737\u001b[0m \u001b[43m \u001b[49m\u001b[43mmode\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1738\u001b[0m \u001b[43m \u001b[49m\u001b[43mencoding\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43moptions\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mget\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mencoding\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1739\u001b[0m \u001b[43m \u001b[49m\u001b[43mcompression\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43moptions\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mget\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mcompression\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1740\u001b[0m \u001b[43m \u001b[49m\u001b[43mmemory_map\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43moptions\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mget\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mmemory_map\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1741\u001b[0m \u001b[43m \u001b[49m\u001b[43mis_text\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mis_text\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1742\u001b[0m \u001b[43m \u001b[49m\u001b[43merrors\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43moptions\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mget\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mencoding_errors\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mstrict\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1743\u001b[0m \u001b[43m \u001b[49m\u001b[43mstorage_options\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43moptions\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mget\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mstorage_options\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1744\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1745\u001b[0m \u001b[38;5;28;01massert\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mhandles \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m 1746\u001b[0m f \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mhandles\u001b[38;5;241m.\u001b[39mhandle\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/pandas/io/common.py:856\u001b[0m, in \u001b[0;36mget_handle\u001b[0;34m(path_or_buf, mode, encoding, compression, memory_map, is_text, errors, storage_options)\u001b[0m\n\u001b[1;32m 851\u001b[0m \u001b[38;5;28;01melif\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(handle, \u001b[38;5;28mstr\u001b[39m):\n\u001b[1;32m 852\u001b[0m \u001b[38;5;66;03m# Check whether the filename is to be opened in binary mode.\u001b[39;00m\n\u001b[1;32m 853\u001b[0m \u001b[38;5;66;03m# Binary mode does not support 'encoding' and 'newline'.\u001b[39;00m\n\u001b[1;32m 854\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m ioargs\u001b[38;5;241m.\u001b[39mencoding \u001b[38;5;129;01mand\u001b[39;00m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mb\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;129;01min\u001b[39;00m ioargs\u001b[38;5;241m.\u001b[39mmode:\n\u001b[1;32m 855\u001b[0m \u001b[38;5;66;03m# Encoding\u001b[39;00m\n\u001b[0;32m--> 856\u001b[0m handle \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mopen\u001b[39;49m\u001b[43m(\u001b[49m\n\u001b[1;32m 857\u001b[0m \u001b[43m \u001b[49m\u001b[43mhandle\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 858\u001b[0m \u001b[43m \u001b[49m\u001b[43mioargs\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mmode\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 859\u001b[0m \u001b[43m \u001b[49m\u001b[43mencoding\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mioargs\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mencoding\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 860\u001b[0m \u001b[43m \u001b[49m\u001b[43merrors\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43merrors\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 861\u001b[0m \u001b[43m \u001b[49m\u001b[43mnewline\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m 862\u001b[0m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 863\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m 864\u001b[0m \u001b[38;5;66;03m# Binary mode\u001b[39;00m\n\u001b[1;32m 865\u001b[0m handle \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mopen\u001b[39m(handle, ioargs\u001b[38;5;241m.\u001b[39mmode)\n","\u001b[0;31mFileNotFoundError\u001b[0m: [Errno 2] No such file or directory: '/kaggle/input/cafa5-exp001/submission.tsv'"],"ename":"FileNotFoundError","evalue":"[Errno 2] No such file or directory: '/kaggle/input/cafa5-exp001/submission.tsv'","output_type":"error"}],"execution_count":134},{"cell_type":"markdown","source":"## Install","metadata":{}},{"cell_type":"code","source":"! pip install biopython -q\n! pip install polars\n! pip install transformers -q\n! pip install wandb -q\n! pip install torchmetrics -q","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T11:11:19.523053Z","iopub.execute_input":"2026-08-14T11:11:19.52342Z"}},"outputs":[{"name":"stdout","text":"\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n\u001b[0mRequirement already satisfied: polars in /opt/conda/lib/python3.10/site-packages (0.17.15)\n\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n\u001b[0m\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n\u001b[0m\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n\u001b[0m\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n\u001b[0m","output_type":"stream"}],"execution_count":null},{"cell_type":"code","source":"import sys\nimport os\nimport json\nimport logging\nimport datetime\nimport requests\nimport gc\nimport joblib\nimport tempfile\nfrom pathlib import Path\n\nfrom Bio import SeqIO\nimport pandas as pd\nimport numpy as np\nimport polars as pl\nfrom tqdm import tqdm\nimport torch\nimport torch.nn as nn\nimport wandb\n\nfrom sklearn.model_selection import KFold\n\nfrom transformers import AutoTokenizer, AutoModel, AutoConfig\nfrom torch.utils.data import DataLoader, Dataset\nfrom torch.cuda.amp import autocast, GradScaler\nfrom torch.optim import Adam, AdamW\nfrom torch.optim.lr_scheduler import (\n CosineAnnealingWarmRestarts,\n CosineAnnealingLR,\n MultiStepLR,\n ReduceLROnPlateau,\n)\n\nfrom transformers import get_linear_schedule_with_warmup, get_cosine_schedule_with_warmup\nfrom transformers import DataCollatorWithPadding\nfrom torchmetrics.classification import MultilabelF1Score\n","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:17:20.771534Z","iopub.execute_input":"2026-08-14T13:17:20.772324Z","iopub.status.idle":"2026-08-14T13:17:20.777663Z","shell.execute_reply.started":"2026-08-14T13:17:20.772301Z","shell.execute_reply":"2026-08-14T13:17:20.776868Z"}},"outputs":[],"execution_count":306},{"cell_type":"markdown","source":"## Config","metadata":{}},{"cell_type":"code","source":"class Config:\n train = True\n evaluation = True\n inference = True\n submission = True\n\n max_label = 512\n n_fold = 5\n trn_fold = [0]\n\n max_epochs = 8\n train_batch_size = 24\n valid_batch_size = 64\n num_workers = 4\n seed = 8823\n name = \"exp001\"\n competition = \"cafa-5-protein-function-prediction\"\n debug = False\n model = \"facebook/esm2_t6_8M_UR50D\"\n max_len = 512\n gradient_checkpointing = False\n gradient_accumulation_steps = 1\n clip_grad_norm = 1000\n\n optimizer = dict(\n optimizer_name=\"AdamW\",\n lr=2e-5,\n weight_decay=1e-2,\n eps=1e-6,\n beta=(0.9, 0.999),\n encoder_lr=2e-5,\n decoder_lr=2e-5,\n )\n\n scheduler = dict(\n scheduler_name=\"cosine_restarts\",\n first_cycle_steps_ratio=0.5,\n cycle_mult=1.0,\n max_lr=2e-5,\n min_lr=1e-7,\n warmup_steps=100,\n gamma=0.8,\n )\n batch_scheduler = True\n\n\nif Config.debug:\n Config.max_label = 100\n Config.max_epochs = 2\n Config.max_len = 100\n Config.n_fold = 2\n Config.trn_fold = [0, 1]","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:53:48.011709Z","iopub.execute_input":"2026-08-14T12:53:48.012127Z","iopub.status.idle":"2026-08-14T12:53:48.019822Z","shell.execute_reply.started":"2026-08-14T12:53:48.012105Z","shell.execute_reply":"2026-08-14T12:53:48.018738Z"}},"outputs":[],"execution_count":240},{"cell_type":"code","source":"config = Config()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:53:52.403035Z","iopub.execute_input":"2026-08-14T12:53:52.403395Z","iopub.status.idle":"2026-08-14T12:53:52.409929Z","shell.execute_reply.started":"2026-08-14T12:53:52.40337Z","shell.execute_reply":"2026-08-14T12:53:52.408861Z"}},"outputs":[],"execution_count":241},{"cell_type":"code","source":"# constants\n# HOME改为Kaggle工作目录\nHOME = Path(\"/kaggle/working\")\n# API这一行是Colab谷歌网盘路径，Kaggle不需要，注释掉\n# API = Path(\"/content/drive/MyDrive/competition/kaggle.json\")\n\n# 下面这行是Colab内网接口，Kaggle无法运行，直接注释\n# NOTEBOOK_NAME = requests.get(\"http://172.28.0.12:9000/api/sessions\").json()[0][\"name\"][:-6]\n# 手动指定实验名称，名字可以自己修改\nEXP_NAME = Config.name if Config.name is not None else \"cafa5_exp01\"\n\nINPUTS = HOME / \"inputs\"\nOUTPUTS = HOME / \"outputs\"\nINTERMEDIATES = HOME / \"intermediates\" # intermediate outputs\nSUBMISSIONS = HOME / \"submissions\"\nOUTPUTS_EXP = OUTPUTS / EXP_NAME\nEXP_MODELS = OUTPUTS_EXP / \"models\"\nEXP_REPORTS = OUTPUTS_EXP / \"reports\"\nEXP_PREDS = OUTPUTS_EXP / \"predictions\"\n\nSCRIPTS = HOME / \"scripts\"\nCAFA_EVALUATOR = SCRIPTS / \"CAFA_evaluator\"","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:53:55.28356Z","iopub.execute_input":"2026-08-14T12:53:55.283947Z","iopub.status.idle":"2026-08-14T12:53:55.289922Z","shell.execute_reply.started":"2026-08-14T12:53:55.283926Z","shell.execute_reply":"2026-08-14T12:53:55.289106Z"}},"outputs":[],"execution_count":242},{"cell_type":"markdown","source":"## Setup","metadata":{}},{"cell_type":"code","source":"def setup(Config):\n for d in [\n HOME,\n INPUTS,\n SUBMISSIONS,\n EXP_MODELS,\n EXP_REPORTS,\n EXP_PREDS,\n INTERMIDIATES,\n SCRIPTS,\n CAFA_EVALUATOR,\n ]:\n d.mkdir(parents=True, exist_ok=True)\n\n \n if not (CAFA_EVALUATOR / \"src\").is_dir():\n import subprocess\n subprocess.run([\"git\", \"clone\", \"https://github.com/BioComputingUP/CAFA-evaluator.git\", str(CAFA_EVALUATOR)])\n\n\n","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:53:59.183294Z","iopub.execute_input":"2026-08-14T12:53:59.183608Z","iopub.status.idle":"2026-08-14T12:53:59.189027Z","shell.execute_reply.started":"2026-08-14T12:53:59.183573Z","shell.execute_reply":"2026-08-14T12:53:59.188016Z"}},"outputs":[],"execution_count":243},{"cell_type":"code","source":"from pathlib import Path\nimport sys\nimport torch\n\nHOME = Path(\"/kaggle/working\")\nINPUTS = HOME / \"../input\"\n\n# 全部路径一次性定义\nINTERMEDIATES = HOME / \"intermediates\"\nSUBMISSIONS = HOME / \"submission\"\nEXP_MODELS = HOME / \"exp_models\"\nEXP_REPORTS = HOME / \"exp_reports\"\nEXP_PREDS = HOME / \"exp_preds\"\nSCRIPTS = HOME / \"scripts\"\nCAFA_EVALUATOR = SCRIPTS / \"CAFA_evaluator\"\n\n# 批量建文件夹\nfor p in [INTERMEDIATES, SUBMISSIONS, EXP_MODELS, EXP_REPORTS, EXP_PREDS, SCRIPTS, CAFA_EVALUATOR]:\n p.mkdir(parents=True, exist_ok=True)\n\n# Config类，不再二次重复赋值\nclass Config:\n competition = \"cafa5\"\n max_label = 1500\n max_seq_len = 1024\n batch_size = 32\n epochs = 5\n lr = 1e-4\n model = \"facebook/esm2-t6-35M-ur50S\"\n seed = 42\n debug = False\n\ncfg = Config()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:06.983314Z","iopub.execute_input":"2026-08-14T12:54:06.983922Z","iopub.status.idle":"2026-08-14T12:54:06.990886Z","shell.execute_reply.started":"2026-08-14T12:54:06.983897Z","shell.execute_reply":"2026-08-14T12:54:06.989871Z"}},"outputs":[],"execution_count":246},{"cell_type":"code","source":"def setup(cfg):\n import subprocess\n if not (SCRIPTS / \"CAFA_evaluator\").exists():\n subprocess.run([\n \"git\",\n \"clone\",\n \"https://github.com/BioPandas/CAFA‑evaluator.git\",\n str(CAFA_EVALUATOR)\n ])","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:11.536569Z","iopub.execute_input":"2026-08-14T12:54:11.536919Z","iopub.status.idle":"2026-08-14T12:54:11.542319Z","shell.execute_reply.started":"2026-08-14T12:54:11.536898Z","shell.execute_reply":"2026-08-14T12:54:11.541063Z"}},"outputs":[],"execution_count":247},{"cell_type":"code","source":"setup(Config)\n\n# LOGGER = Logger(OUTPUTS_EXP.as_posix())\n# 等你把Logger类的定义放到前面之后，再取消上一行注释\n\nDEVICE = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n# wandb.login()\nsys.path.append((CAFA_EVALUATOR / \"src\").as_posix())","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:15.955916Z","iopub.execute_input":"2026-08-14T12:54:15.956254Z","iopub.status.idle":"2026-08-14T12:54:15.961328Z","shell.execute_reply.started":"2026-08-14T12:54:15.956234Z","shell.execute_reply":"2026-08-14T12:54:15.96054Z"}},"outputs":[],"execution_count":248},{"cell_type":"markdown","source":"## Metrics","metadata":{}},{"cell_type":"code","source":"# competition metrics\ndef get_score(gt_df, pred_df, th_step=0.05, max_terms=None, use_cache=False, n_cpu=1):\n from parser import obo_parser, gt_parser, pred_parser, ia_parser\n from graph import Graph\n from evaluation import evaluate_prediction\n\n pred_df = pred_df.astype({\"preds\": np.float16})\n gt_df = gt_df[[\"EntryID\", \"term\"]]\n\n with tempfile.TemporaryDirectory() as tmp_dir:\n dirpath = Path(tmp_dir)\n\n # set filepath and parameter\n gt_file, pred_file = dirpath / \"gt.tsv\", dirpath / \"pred.tsv\"\n ia = INPUTS / \"IA.txt\"\n obo_file = INPUTS / \"Train\" / \"go-basic.obo\"\n no_orphans = False\n prop = \"max\"\n norm = \"cafa\"\n parsed_gt_file = INTERMIDIATES / \"parsed_gt.pkl\"\n\n # save evaluator inputs to tmpdir\n gt_df.to_csv(gt_file, header=False, index=False, sep=\"\\t\")\n pred_df.to_csv(pred_file, header=False, index=False, sep=\"\\t\")\n\n # main\n ia_dict = None if ia is None else ia_parser(ia)\n tau_arr = np.arange(0.01, 1, th_step)\n ontologies = [\n Graph(ns, terms_dict, ia_dict, not no_orphans)\n for ns, terms_dict in obo_parser(obo_file).items()\n ]\n\n if use_cache:\n if not parsed_gt_file.is_file():\n # to save time\n gt = gt_parser(gt_file, ontologies)\n joblib.dump(gt, parsed_gt_file)\n gt = joblib.load(parsed_gt_file)\n else:\n gt = gt_parser(gt_file, ontologies)\n\n df_pred = evaluate_prediction( # too heavy and slow...\n prediction=pred_parser(pred_file, ontologies, gt, prop_mode=prop, max_terms=max_terms),\n gt=gt,\n ontologies=ontologies,\n tau_arr=tau_arr,\n normalization=norm,\n n_cpu=n_cpu,\n )\n score = df_pred.groupby(\"ns\").agg({\"f\": \"max\"}).values.mean()\n return score, df_pred.reset_index(drop=True)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:19.643855Z","iopub.execute_input":"2026-08-14T12:54:19.64425Z","iopub.status.idle":"2026-08-14T12:54:19.653646Z","shell.execute_reply.started":"2026-08-14T12:54:19.644228Z","shell.execute_reply":"2026-08-14T12:54:19.652647Z"}},"outputs":[],"execution_count":249},{"cell_type":"markdown","source":"## Load Data","metadata":{}},{"cell_type":"code","source":"import polars as pl\n\ntrain_label_df_wide = train_label_df.pivot(\n columns=\"term\",\n index=\"EntryID\",\n values=\"count\",\n aggregate_function=\"sum\"\n).fill_null(0)\n\ntrain_label_df_pd = train_label_df_wide.to_pandas().reset_index()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:24.95243Z","iopub.execute_input":"2026-08-14T12:54:24.953185Z","iopub.status.idle":"2026-08-14T12:54:25.001293Z","shell.execute_reply.started":"2026-08-14T12:54:24.953163Z","shell.execute_reply":"2026-08-14T12:54:25.00001Z"}},"outputs":[],"execution_count":250},{"cell_type":"code","source":"train_df, target_columns = make_train_dataset(train_seq_df, train_label_df_pd, max_label=Config.max_label)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:29.59197Z","iopub.execute_input":"2026-08-14T12:54:29.592325Z","iopub.status.idle":"2026-08-14T12:54:29.685792Z","shell.execute_reply.started":"2026-08-14T12:54:29.592302Z","shell.execute_reply":"2026-08-14T12:54:29.684791Z"}},"outputs":[],"execution_count":251},{"cell_type":"code","source":"print(train_label_df_pd.shape)\nprint(train_label_df_pd.columns[:5].tolist())\nprint(train_label_df_pd.dtypes)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:32.790397Z","iopub.execute_input":"2026-08-14T12:54:32.790698Z","iopub.status.idle":"2026-08-14T12:54:32.798642Z","shell.execute_reply.started":"2026-08-14T12:54:32.790679Z","shell.execute_reply":"2026-08-14T12:54:32.797276Z"}},"outputs":[{"name":"stdout","text":"(69, 762)\n['index', 'EntryID', 'GO:0008152', 'GO:0072523', 'GO:0044237']\nindex int64\nEntryID object\nGO:0008152 uint32\nGO:0072523 uint32\nGO:0044237 uint32\n ... \nGO:1903514 uint32\nGO:0008654 uint32\nGO:0006663 uint32\nGO:0046474 uint32\nGO:0018904 uint32\nLength: 762, dtype: object\n","output_type":"stream"}],"execution_count":252},{"cell_type":"code","source":"train_label_df_pd = train_label_df_pd.drop(columns=[\"index\"])","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:39.644383Z","iopub.execute_input":"2026-08-14T12:54:39.644731Z","iopub.status.idle":"2026-08-14T12:54:39.651792Z","shell.execute_reply.started":"2026-08-14T12:54:39.644706Z","shell.execute_reply":"2026-08-14T12:54:39.650126Z"}},"outputs":[],"execution_count":253},{"cell_type":"code","source":"print(train_label_df_pd.columns[:5].tolist())","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:54:43.262206Z","iopub.execute_input":"2026-08-14T12:54:43.262518Z","iopub.status.idle":"2026-08-14T12:54:43.267488Z","shell.execute_reply.started":"2026-08-14T12:54:43.262498Z","shell.execute_reply":"2026-08-14T12:54:43.266317Z"}},"outputs":[{"name":"stdout","text":"['EntryID', 'GO:0008152', 'GO:0072523', 'GO:0044237', 'GO:1901360']\n","output_type":"stream"}],"execution_count":254},{"cell_type":"code","source":"import os\nfrom pathlib import Path\nfilepath = INTERMEDIATES / \"sorted_GO_list.pkl\"\nif filepath.exists():\n os.remove(str(filepath))\n\ntrain_df, target_columns = make_train_dataset(train_seq_df, train_label_df_pd, max_label=Config.max_label)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:57:07.892577Z","iopub.execute_input":"2026-08-14T12:57:07.892931Z","iopub.status.idle":"2026-08-14T12:57:07.993259Z","shell.execute_reply.started":"2026-08-14T12:57:07.892911Z","shell.execute_reply":"2026-08-14T12:57:07.992426Z"}},"outputs":[],"execution_count":259},{"cell_type":"code","source":"import pandas as pd\nimport polars as pl\nimport joblib\nimport gc\nfrom pathlib import Path\n\nINTERMIDIATES = Path(\"./intermediates\")\n\ndef make_train_terms_crosstab(train_terms_df):\n filepath = INTERMIDIATES / \"train_terms_crosstab.pkl\"\n\n if not filepath.is_file():\n # train_terms_df 如果是pl.DataFrame先转pandas，避免pivot行为差异\n if isinstance(train_terms_df, pl.DataFrame):\n train_terms_df = train_terms_df.to_pandas()\n\n crosstab_df = train_terms_df.pivot(\n values=\"aspect\", index=\"EntryID\", columns=\"term\", aggregatefunc=\"count\"\n ).fillna(0)\n crosstab_df = crosstab_df.astype(bool).reset_index()\n\n joblib.dump(crosstab_df, filepath)\n del crosstab_df\n gc.collect()\n\n return joblib.load(filepath).sort_values(\"EntryID\").reset_index(drop=True)\n\n\ndef make_sequence_df(input_fasta):\n sequence_df = (\n pd.DataFrame([[x.name, x.seq] for x in input_fasta], columns=[\"EntryID\", \"seq\"])\n .sort_values(\"EntryID\")\n .reset_index(drop=True)\n )\n return sequence_df\n\ndef make_train_dataset(train_seq_df, train_label_df, max_label=None):\n import pandas as pd\n import joblib\n from pathlib import Path\n INTERMEDIATES = Path(\"./intermediates\")\n INTERMEDIATES.mkdir(exist_ok=True)\n\n # 如果传进来的是polars DataFrame，转成pandas\n if hasattr(train_label_df, \"to_pandas\"):\n train_label_df = train_label_df.to_pandas()\n\n if max_label is None:\n max_label = train_label_df.shape[1]\n\n filepath = INTERMEDIATES / \"sorted_GO_list.pkl\"\n if not filepath.is_file():\n label_cols = [c for c in train_label_df.columns if c != \"EntryID\"]\n col_sum = train_label_df[label_cols].sum(axis=0)\n target_columns = col_sum.sort_values(ascending=False).index.tolist()\n joblib.dump(target_columns, filepath)\n\n # 下面这一段必须在if外面，无论文件是否存在都会执行\n target_columns = joblib.load(filepath)[:max_label]\n target_columns = [c for c in target_columns if c in train_label_df.columns]\n\n # 替换原来concat，改用merge按EntryID合并\n train_df = pd.merge(\n train_seq_df,\n train_label_df[[\"EntryID\"] + target_columns],\n on=\"EntryID\",\n how=\"inner\"\n )\n\n return train_df, target_columns\n\n \n\n \n","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:57:12.256687Z","iopub.execute_input":"2026-08-14T12:57:12.257812Z","iopub.status.idle":"2026-08-14T12:57:12.26966Z","shell.execute_reply.started":"2026-08-14T12:57:12.257778Z","shell.execute_reply":"2026-08-14T12:57:12.268609Z"}},"outputs":[],"execution_count":260},{"cell_type":"code","source":"\nimport polars as pl\nfrom pathlib import Path\n\ndef make_train_terms_crosstab(train_terms_df: pl.DataFrame, intermediates_dir: Path):\n filepath = intermediates_dir / \"train_terms_sparse.parquet\"\n if not filepath.is_file():\n crosstab_df = (\n train_terms_df\n .group_by([\"EntryID\", \"term\"])\n .agg(pl.count(\"*\").alias(\"count\"))\n )\n crosstab_df.write_parquet(filepath)\n else:\n crosstab_df = pl.read_parquet(filepath)\n return crosstab_df\n","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:57:19.629908Z","iopub.execute_input":"2026-08-14T12:57:19.630279Z","iopub.status.idle":"2026-08-14T12:57:19.635313Z","shell.execute_reply.started":"2026-08-14T12:57:19.630253Z","shell.execute_reply":"2026-08-14T12:57:19.634562Z"}},"outputs":[],"execution_count":261},{"cell_type":"code","source":"import os\npkl_path = str(INTERMEDIATES / \"sorted_GO_list.pkl\")\nif os.path.exists(pkl_path):\n os.remove(pkl_path)\n print(\"缓存文件已删除\")\nelse:\n print(\"缓存文件本来就不存在\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:57:24.626163Z","iopub.execute_input":"2026-08-14T12:57:24.626525Z","iopub.status.idle":"2026-08-14T12:57:24.633081Z","shell.execute_reply.started":"2026-08-14T12:57:24.626496Z","shell.execute_reply":"2026-08-14T12:57:24.631925Z"}},"outputs":[{"name":"stdout","text":"缓存文件已删除\n","output_type":"stream"}],"execution_count":262},{"cell_type":"code","source":"train_df, target_columns = make_train_dataset(train_seq_df, train_label_df_pd, max_label=Config.max_label)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:57:28.751981Z","iopub.execute_input":"2026-08-14T12:57:28.752362Z","iopub.status.idle":"2026-08-14T12:57:28.841015Z","shell.execute_reply.started":"2026-08-14T12:57:28.75234Z","shell.execute_reply":"2026-08-14T12:57:28.839626Z"}},"outputs":[],"execution_count":263},{"cell_type":"markdown","source":"## CV Split","metadata":{}},{"cell_type":"code","source":"def add_fold_idx(config, train_df, target_columns):\n fold = KFold(n_splits=config.n_fold, shuffle=True, random_state=config.seed)\n train_df[\"fold\"] = -1\n for i_fold, (train_index, val_index) in enumerate(fold.split(train_df, train_df[target_columns])):\n train_df.iloc[val_index, train_df.columns.get_loc(\"fold\")] = int(i_fold)\n train_df[\"fold\"] = train_df[\"fold\"].astype(int)\n return train_df\n\n# 调用放在函数外面，仅此一处改动\ntrain_df, target_columns = make_train_dataset(train_seq_df, train_label_df_pd, max_label=Config.max_label)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:04:51.854152Z","iopub.execute_input":"2026-08-14T13:04:51.854464Z","iopub.status.idle":"2026-08-14T13:04:51.944476Z","shell.execute_reply.started":"2026-08-14T13:04:51.854445Z","shell.execute_reply":"2026-08-14T13:04:51.943619Z"}},"outputs":[],"execution_count":282},{"cell_type":"code","source":"train_df = add_fold_idx(config=config, train_df=train_df, target_columns=target_columns)\ntrain_df.head()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:04:39.189324Z","iopub.execute_input":"2026-08-14T13:04:39.18964Z","iopub.status.idle":"2026-08-14T13:04:39.203501Z","shell.execute_reply.started":"2026-08-14T13:04:39.189619Z","shell.execute_reply":"2026-08-14T13:04:39.202721Z"}},"outputs":[{"execution_count":280,"output_type":"execute_result","data":{"text/plain":" EntryID seq fold\n0 A0A009IHW8 (M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ... 0\n1 A0A009IHW8 (M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ... 2\n2 A0A009IHW8 (M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ... 3\n3 A0A009IHW8 (M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ... 1\n4 A0A009IHW8 (M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ... 1","text/html":"
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
EntryIDseqfold
0A0A009IHW8(M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ...0
1A0A009IHW8(M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ...2
2A0A009IHW8(M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ...3
3A0A009IHW8(M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ...1
4A0A009IHW8(M, S, L, E, Q, K, K, G, A, D, I, I, S, K, I, ...1
\n
"},"metadata":{}}],"execution_count":280},{"cell_type":"code","source":"train_df[\"fold\"].value_counts()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:57:43.296172Z","iopub.execute_input":"2026-08-14T12:57:43.296541Z","iopub.status.idle":"2026-08-14T12:57:43.303516Z","shell.execute_reply.started":"2026-08-14T12:57:43.296516Z","shell.execute_reply":"2026-08-14T12:57:43.302667Z"}},"outputs":[{"execution_count":266,"output_type":"execute_result","data":{"text/plain":"0 14\n1 14\n2 14\n3 14\n4 13\nName: fold, dtype: int64"},"metadata":{}}],"execution_count":266},{"cell_type":"markdown","source":"## Dataset","metadata":{}},{"cell_type":"code","source":"import torch\nimport numpy as np\nfrom torch.utils.data import Dataset\n\ndef find_max_length(tokenizer, texts: list[str]):\n \"\"\"\n [memo]\n train : facebook/esm2_t6_8M_UR50D : 35377\n \"\"\"\n lengths = [\n len(tokenizer(text, add_special_tokens=False)[\"input_ids\"])\n for text in tqdm(texts, total=len(texts))\n ]\n return lengths\n\n\ndef prepare_input(config, tokenizer, text):\n inputs = tokenizer.encode_plus(\n text,\n return_tensors=None,\n add_special_tokens=True,\n max_length=config.max_len,\n padding=\"max_length\",\n truncation=True,\n )\n for k, v in inputs.items():\n inputs[k] = torch.tensor(v, dtype=torch.long)\n return inputs\n\n\nclass TrainDataset(Dataset):\n def __init__(self, config, tokenizer, texts: list[str], targets: list | np.ndarray):\n self.config = config\n self.texts = texts\n self.targets = targets\n self.tokenizer = tokenizer\n\n def __len__(self):\n return len(self.texts)\n\n def __getitem__(self, item):\n inputs = prepare_input(config=self.config, tokenizer=self.tokenizer, text=self.texts[item])\n targets = torch.tensor(self.targets[item], dtype=torch.float16) # label\n return {\n \"input_ids\": inputs[\"input_ids\"],\n \"attention_mask\": inputs[\"attention_mask\"],\n \"targets\": targets,\n }\n\n\nclass TestDataset(Dataset):\n def __init__(self, config, tokenizer, texts: list[str]):\n self.config = config\n self.texts = texts\n self.tokenizer = tokenizer\n\n def __len__(self):\n return len(self.texts)\n\n def __getitem__(self, item):\n inputs = prepare_input(config=self.config, tokenizer=self.tokenizer, text=self.texts[item])\n return inputs\n\n\ndef collate(inputs):\n mask_len = int(inputs[\"attention_mask\"].sum(axis=1).max())\n for k, v in inputs.items():\n if k != \"targets\":\n inputs[k] = inputs[k][:, :mask_len]\n return inputs","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:06:01.096734Z","iopub.execute_input":"2026-08-14T13:06:01.097068Z","iopub.status.idle":"2026-08-14T13:06:01.106908Z","shell.execute_reply.started":"2026-08-14T13:06:01.097048Z","shell.execute_reply":"2026-08-14T13:06:01.105898Z"}},"outputs":[],"execution_count":287},{"cell_type":"markdown","source":"## Model","metadata":{}},{"cell_type":"code","source":"import torch\nimport torch.nn as nn\ndef get_model_config(config):\n model_config = AutoConfig.from_pretrained(config.model, output_hidden_states=True)\n model_config.hidden_dropout = 0.0\n model_config.hidden_dropout_prob = 0.0\n model_config.attention_dropout = 0.0\n model_config.attention_probs_dropout_prob = 0.0\n return model_config\n\n\nclass CustomModel(nn.Module):\n def __init__(self, config, num_label, pretrained=False):\n super().__init__()\n self.model_config = get_model_config(config=config)\n\n if pretrained:\n self.model = AutoModel.from_pretrained(config.model, config=self.model_config)\n else:\n self.model = AutoModel.from_config(self.model_config)\n\n if config.gradient_checkpointing:\n self.model.gradient_checkpointing_enable()\n\n self.fc = nn.Linear(self.model_config.hidden_size, num_label)\n self._init_weights(self.fc)\n\n def _init_weights(self, module):\n if isinstance(module, nn.Linear):\n module.weight.data.normal_(mean=0.0, std=self.model_config.initializer_range)\n if module.bias is not None:\n module.bias.data.zero_()\n elif isinstance(module, nn.Embedding):\n module.weight.data.normal_(mean=0.0, std=self.model_config.initializer_range)\n if module.padding_idx is not None:\n module.weight.data[module.padding_idx].zero_()\n elif isinstance(module, nn.LayerNorm):\n module.bias.data.zero_()\n module.weight.data.fill_(1.0)\n\n def feature(self, input_ids, attention_mask):\n outputs = self.model(input_ids=input_ids, attention_mask=attention_mask)\n feature = outputs[0][:, 0, :]\n return feature\n\n def forward(self, batch):\n feature = self.feature(batch[\"input_ids\"], batch[\"attention_mask\"])\n output = self.fc(feature)\n return output","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:57:53.172605Z","iopub.execute_input":"2026-08-14T12:57:53.172955Z","iopub.status.idle":"2026-08-14T12:57:53.183324Z","shell.execute_reply.started":"2026-08-14T12:57:53.172932Z","shell.execute_reply":"2026-08-14T12:57:53.182275Z"}},"outputs":[],"execution_count":268},{"cell_type":"markdown","source":"## Funcs","metadata":{}},{"cell_type":"code","source":"def train_fn(\n config,\n model,\n dataloader,\n criterion,\n optimizer,\n scheduler,\n device,\n wandb_logger,\n _custom_step,\n):\n model.train()\n scaler = torch.cuda.amp.GradScaler()\n losses = []\n\n tbar = tqdm(enumerate(dataloader), total=len(dataloader))\n for step, batch in tbar:\n batch = collate(batch)\n for k, v in batch.items():\n batch[k] = v.to(device)\n targets = batch[\"targets\"]\n batch_size = targets.size(0)\n\n with torch.cuda.amp.autocast():\n batch_outputs = model(batch)\n loss = criterion(batch_outputs, targets)\n\n if config.gradient_accumulation_steps > 1:\n loss = loss / config.gradient_accumulation_steps\n\n scaler.scale(loss).backward()\n if config.clip_grad_norm is not None:\n torch.nn.utils.clip_grad_norm_(model.parameters(), config.clip_grad_norm)\n\n if (step + 1) % config.gradient_accumulation_steps == 0:\n scaler.step(optimizer)\n scaler.update()\n optimizer.zero_grad()\n _custom_step += 1\n if config.batch_scheduler:\n scheduler.step()\n\n wandb_logger.log(\n {\"train_loss\": loss, \"lr\": scheduler.get_lr()[0], \"train_step\": _custom_step}\n )\n\n losses.append(float(loss))\n tbar.set_description(f\"loss: {np.mean(losses):.4f} lr: {scheduler.get_lr()[0]:.6f}\")\n\n loss = np.mean(losses)\n return loss, _custom_step\n\n\ndef valid_fn(\n config,\n model,\n dataloader,\n criterion,\n device,\n wandb_logger,\n _custom_step,\n):\n model.eval()\n outputs, targets = [], []\n losses = []\n\n tbar = tqdm(enumerate(dataloader), total=len(dataloader))\n for step, batch in tbar:\n batch = collate(batch)\n targets.append(batch[\"targets\"])\n\n for k, v in batch.items():\n batch[k] = v.to(device)\n\n batch_size = batch[\"targets\"].size(0)\n with torch.no_grad():\n batch_outputs = model(batch)\n loss = criterion(batch_outputs, batch[\"targets\"])\n\n if config.gradient_accumulation_steps > 1:\n loss = loss / config.gradient_accumulation_steps\n\n batch_outputs = torch.sigmoid(batch_outputs) # to proba\n batch_outputs = batch_outputs.to(\"cpu\").numpy()\n outputs.append(batch_outputs)\n\n wandb_logger.log({\"valid_loss\": loss, \"valid_step\": _custom_step})\n _custom_step += 1\n losses.append(float(loss))\n\n tbar.set_description(f\"loss: {np.mean(losses):.4f}\")\n\n outputs = np.concatenate(outputs)\n targets = np.concatenate(targets)\n\n loss = np.mean(losses)\n return (loss, outputs, targets, _custom_step)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T12:57:57.889173Z","iopub.execute_input":"2026-08-14T12:57:57.889478Z","iopub.status.idle":"2026-08-14T12:57:57.900895Z","shell.execute_reply.started":"2026-08-14T12:57:57.889458Z","shell.execute_reply":"2026-08-14T12:57:57.899711Z"}},"outputs":[],"execution_count":269},{"cell_type":"code","source":"def get_optimizer(optimizer_config, model):\n if optimizer_config[\"optimizer_name\"] == \"AdamW\":\n param_optimizer = list(model.named_parameters())\n no_decay = [\"bias\", \"LayerNorm.bias\", \"LayerNorm.weight\"]\n optimizer_parameters = [\n {\n \"params\": [\n p\n for n, p in model.model.named_parameters()\n if not any(nd in n for nd in no_decay)\n ],\n \"lr\": optimizer_config[\"encoder_lr\"],\n \"weight_decay\": optimizer_config[\"weight_decay\"],\n },\n {\n \"params\": [\n p for n, p in model.model.named_parameters() if any(nd in n for nd in no_decay)\n ],\n \"lr\": optimizer_config[\"encoder_lr\"],\n \"weight_decay\": 0.0,\n },\n {\n \"params\": [p for n, p in model.named_parameters() if \"model\" not in n],\n \"lr\": optimizer_config[\"decoder_lr\"],\n \"weight_decay\": 0.0,\n },\n ]\n optimizer = AdamW(\n optimizer_parameters,\n lr=optimizer_config[\"lr\"],\n betas=optimizer_config[\"beta\"],\n eps=optimizer_config[\"eps\"],\n )\n return optimizer\n else:\n raise NotImplementedError\n\n\ndef get_scheduler(scheduler_config, optimizer, num_train_steps):\n if scheduler_config[\"scheduler_name\"] == \"linear\":\n scheduler = get_linear_schedule_with_warmup(\n optimizer,\n num_warmup_steps=int(scheduler_config[\"num_warmup_steps_rate\"] * num_train_steps),\n num_training_steps=num_train_steps,\n )\n return scheduler\n\n elif scheduler_config[\"scheduler_name\"] == \"cosine\":\n scheduler = get_cosine_schedule_with_warmup(\n optimizer,\n num_warmup_steps=int(scheduler_config[\"num_warmup_steps_rate\"] * num_train_steps),\n num_training_steps=num_train_steps,\n num_cycles=scheduler_config[\"num_cycles\"],\n )\n return scheduler\n\n elif scheduler_config[\"scheduler_name\"] == \"cosine_restarts\":\n \"\"\"\n example:\n first_cycle_steps_ratio = 0.25,\n cycle_mult = 1.0,\n max_lr = 2e-5,\n min_lr = 1e-7,\n warmup_steps=100,\n gamma=0.8)\n \"\"\"\n scheduler = CosineAnnealingWarmupRestarts(\n optimizer,\n first_cycle_steps=int(num_train_steps * scheduler_config[\"first_cycle_steps_ratio\"]),\n cycle_mult=scheduler_config[\"cycle_mult\"],\n max_lr=scheduler_config[\"max_lr\"],\n min_lr=scheduler_config[\"min_lr\"],\n warmup_steps=scheduler_config[\"warmup_steps\"],\n gamma=scheduler_config[\"gamma\"],\n )\n return scheduler\n\n else:\n raise NotImplementedError\n\n\ndef train_loop(config, name, train_df, valid_df, target_columns, tokenizer, device):\n LOGGER.info(f\"========== {name} training ==========\")\n\n # set wandb logger\n wandb.init(\n project=config.competition,\n name=name,\n group=f\"{config.name}\",\n job_type=\"train\",\n anonymous=None,\n reinit=True,\n )\n\n # dataset, dataloader\n train_dataset = TrainDataset(\n config=config,\n texts=train_df[\"seq\"].tolist(),\n targets=train_df[target_columns].to_numpy(),\n tokenizer=tokenizer,\n )\n valid_dataset = TrainDataset(\n config=config,\n texts=valid_df[\"seq\"].tolist(),\n targets=valid_df[target_columns].to_numpy(),\n tokenizer=tokenizer,\n )\n train_dataloader = DataLoader(\n train_dataset,\n batch_size=config.train_batch_size,\n shuffle=True,\n num_workers=config.num_workers,\n pin_memory=True,\n drop_last=True,\n )\n valid_dataloader = DataLoader(\n valid_dataset,\n batch_size=config.valid_batch_size,\n shuffle=False,\n num_workers=config.num_workers,\n pin_memory=True,\n drop_last=False,\n )\n\n # set model & optimizer\n model = CustomModel(config=config, num_label=len(target_columns), pretrained=True)\n model.to(device)\n optimizer = get_optimizer(optimizer_config=config.optimizer, model=model)\n\n # set scheduler\n num_train_steps = int(\n len(train_dataloader) * config.max_epochs // config.gradient_accumulation_steps\n )\n scheduler = get_scheduler(\n scheduler_config=config.scheduler, optimizer=optimizer, num_train_steps=num_train_steps\n )\n\n # loop\n criterion = nn.BCEWithLogitsLoss()\n score_fn = MultilabelF1Score(num_labels=len(target_columns)) # TODO : use custom f1\n\n best_score = -np.inf\n tr_step, val_step = 0, 0\n for epoch in range(Config.max_epochs):\n # training\n loss, tr_step = train_fn(\n config=config,\n model=model,\n dataloader=train_dataloader,\n criterion=criterion,\n optimizer=optimizer,\n scheduler=scheduler,\n device=device,\n wandb_logger=wandb,\n _custom_step=tr_step,\n )\n\n # validation\n val_loss, val_outputs, val_targets, val_step = valid_fn(\n config=config,\n model=model,\n dataloader=valid_dataloader,\n criterion=criterion,\n device=device,\n wandb_logger=wandb,\n _custom_step=val_step,\n )\n\n # calc score (the higher, the better)\n eval_score = score_fn(torch.tensor(val_outputs).float(), torch.tensor(val_targets).float())\n logs = {\n \"Epoch\": epoch,\n \"eval_score\": eval_score,\n \"train_loss_epoch\": loss.item(),\n \"valid_loss_epoch\": val_loss.item(),\n }\n wandb.log(logs)\n LOGGER.info(logs)\n\n if best_score < eval_score:\n best_score = eval_score\n LOGGER.info(f\"epoch {epoch} - best score: {best_score:.4f} model\")\n\n torch.save(model.state_dict(), EXP_MODELS / f\"{name}.pth\") # save model weight\n joblib.dump(val_outputs, EXP_PREDS / f\"{name}.pkl\") # save outputs\n\n if not config.batch_scheduler:\n scheduler.step()\n\n torch.cuda.empty_cache()\n gc.collect()\n wandb.finish(quiet=True)\n\n # save best predictions with id\n best_val_outputs = joblib.load(EXP_PREDS / f\"{name}.pkl\")\n outputs = {\n \"id\": valid_df.index.tolist(),\n \"target_names\": target_columns,\n \"predictions\": best_val_outputs,\n }\n joblib.dump(outputs, EXP_PREDS / f\"{name}_best.pkl\")\n\n return outputs","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:06:15.334969Z","iopub.execute_input":"2026-08-14T13:06:15.335385Z","iopub.status.idle":"2026-08-14T13:06:15.354223Z","shell.execute_reply.started":"2026-08-14T13:06:15.335357Z","shell.execute_reply":"2026-08-14T13:06:15.352736Z"}},"outputs":[],"execution_count":289},{"cell_type":"code","source":"def inference_fn(test_dataloader, model, device):\n preds, targets_masks = [], []\n model.eval()\n model.to(device)\n\n tbar = tqdm(test_dataloader, total=len(test_dataloader))\n for batch in tbar:\n for k, v in batch.items():\n batch[k] = v.to(device)\n\n with torch.no_grad():\n outputs = model(batch)\n outputs = torch.sigmoid(outputs)\n\n outputs = outputs.cpu().detach().numpy()\n preds.append(outputs)\n\n return np.concatenate(preds)\n\n\ndef get_predictions(config, test_df, tokenizer, model_path, target_columns, device):\n test_dataset = TestDataset(config=config, tokenizer=tokenizer, texts=test_df[\"seq\"])\n test_dataloader = DataLoader(\n test_dataset,\n batch_size=config.valid_batch_size,\n shuffle=False,\n collate_fn=DataCollatorWithPadding(tokenizer=tokenizer, padding=\"longest\"),\n num_workers=config.num_workers,\n pin_memory=True,\n drop_last=False,\n )\n\n # get model\n model = CustomModel(config=config, num_label=len(target_columns), pretrained=False)\n state = torch.load(model_path)\n model.load_state_dict(state)\n predictions = inference_fn(test_dataloader, model, device)\n\n del model, state, test_dataloader\n gc.collect()\n torch.cuda.empty_cache()\n\n outputs = {\n \"id\": test_df.index.tolist(),\n \"target_names\": target_columns,\n \"predictions\": predictions,\n }\n return outputs","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:06:21.731678Z","iopub.execute_input":"2026-08-14T13:06:21.732767Z","iopub.status.idle":"2026-08-14T13:06:21.740996Z","shell.execute_reply.started":"2026-08-14T13:06:21.732733Z","shell.execute_reply":"2026-08-14T13:06:21.740064Z"}},"outputs":[],"execution_count":290},{"cell_type":"markdown","source":"## Training","metadata":{}},{"cell_type":"code","source":"import os\nos.environ[\"HF_ENDPOINT\"] = \"https://hf-mirror.com\"\nfrom transformers import AutoTokenizer\n# 这里引号内的模型名要和config.model的值一模一样\ntokenizer = AutoTokenizer.from_pretrained(\"nvidia/esm2_t33_650M_UR50D\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:28:21.093048Z","iopub.execute_input":"2026-08-14T13:28:21.093359Z","iopub.status.idle":"2026-08-14T13:28:53.593005Z","shell.execute_reply.started":"2026-08-14T13:28:21.093338Z","shell.execute_reply":"2026-08-14T13:28:53.591136Z"}},"outputs":[{"traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mgaierror\u001b[0m Traceback (most recent call last)","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/urllib3/connection.py:174\u001b[0m, in \u001b[0;36mHTTPConnection._new_conn\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 173\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m--> 174\u001b[0m conn \u001b[38;5;241m=\u001b[39m \u001b[43mconnection\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcreate_connection\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 175\u001b[0m \u001b[43m \u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_dns_host\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mport\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtimeout\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mextra_kw\u001b[49m\n\u001b[1;32m 176\u001b[0m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 178\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m SocketTimeout:\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/urllib3/util/connection.py:72\u001b[0m, in \u001b[0;36mcreate_connection\u001b[0;34m(address, timeout, source_address, socket_options)\u001b[0m\n\u001b[1;32m 68\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m six\u001b[38;5;241m.\u001b[39mraise_from(\n\u001b[1;32m 69\u001b[0m LocationParseError(\u001b[38;5;124mu\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;132;01m%s\u001b[39;00m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m, label empty or too long\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;241m%\u001b[39m host), \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m 70\u001b[0m )\n\u001b[0;32m---> 72\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m res \u001b[38;5;129;01min\u001b[39;00m \u001b[43msocket\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mgetaddrinfo\u001b[49m\u001b[43m(\u001b[49m\u001b[43mhost\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mport\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mfamily\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msocket\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mSOCK_STREAM\u001b[49m\u001b[43m)\u001b[49m:\n\u001b[1;32m 73\u001b[0m af, socktype, proto, canonname, sa \u001b[38;5;241m=\u001b[39m res\n","File \u001b[0;32m/opt/conda/lib/python3.10/socket.py:955\u001b[0m, in \u001b[0;36mgetaddrinfo\u001b[0;34m(host, port, family, type, proto, flags)\u001b[0m\n\u001b[1;32m 954\u001b[0m addrlist \u001b[38;5;241m=\u001b[39m []\n\u001b[0;32m--> 955\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m res \u001b[38;5;129;01min\u001b[39;00m \u001b[43m_socket\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mgetaddrinfo\u001b[49m\u001b[43m(\u001b[49m\u001b[43mhost\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mport\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mfamily\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mtype\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mproto\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mflags\u001b[49m\u001b[43m)\u001b[49m:\n\u001b[1;32m 956\u001b[0m af, socktype, proto, canonname, sa \u001b[38;5;241m=\u001b[39m res\n","\u001b[0;31mgaierror\u001b[0m: [Errno -3] Temporary failure in name resolution","\nDuring handling of the above exception, another exception occurred:\n","\u001b[0;31mKeyboardInterrupt\u001b[0m Traceback (most recent call last)","Cell \u001b[0;32mIn[309], line 5\u001b[0m\n\u001b[1;32m 3\u001b[0m \u001b[38;5;28;01mfrom\u001b[39;00m \u001b[38;5;21;01mtransformers\u001b[39;00m \u001b[38;5;28;01mimport\u001b[39;00m AutoTokenizer\n\u001b[1;32m 4\u001b[0m \u001b[38;5;66;03m# 这里引号内的模型名要和config.model的值一模一样\u001b[39;00m\n\u001b[0;32m----> 5\u001b[0m tokenizer \u001b[38;5;241m=\u001b[39m \u001b[43mAutoTokenizer\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mfrom_pretrained\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mnvidia/esm2_t33_650M_UR50D\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m)\u001b[49m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/transformers/models/auto/tokenization_auto.py:643\u001b[0m, in \u001b[0;36mAutoTokenizer.from_pretrained\u001b[0;34m(cls, pretrained_model_name_or_path, *inputs, **kwargs)\u001b[0m\n\u001b[1;32m 640\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m tokenizer_class\u001b[38;5;241m.\u001b[39mfrom_pretrained(pretrained_model_name_or_path, \u001b[38;5;241m*\u001b[39minputs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n\u001b[1;32m 642\u001b[0m \u001b[38;5;66;03m# Next, let's try to use the tokenizer_config file to get the tokenizer class.\u001b[39;00m\n\u001b[0;32m--> 643\u001b[0m tokenizer_config \u001b[38;5;241m=\u001b[39m \u001b[43mget_tokenizer_config\u001b[49m\u001b[43m(\u001b[49m\u001b[43mpretrained_model_name_or_path\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 644\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m_commit_hash\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;129;01min\u001b[39;00m tokenizer_config:\n\u001b[1;32m 645\u001b[0m kwargs[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m_commit_hash\u001b[39m\u001b[38;5;124m\"\u001b[39m] \u001b[38;5;241m=\u001b[39m tokenizer_config[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m_commit_hash\u001b[39m\u001b[38;5;124m\"\u001b[39m]\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/transformers/models/auto/tokenization_auto.py:487\u001b[0m, in \u001b[0;36mget_tokenizer_config\u001b[0;34m(pretrained_model_name_or_path, cache_dir, force_download, resume_download, proxies, use_auth_token, revision, local_files_only, subfolder, **kwargs)\u001b[0m\n\u001b[1;32m 425\u001b[0m \u001b[38;5;250m\u001b[39m\u001b[38;5;124;03m\"\"\"\u001b[39;00m\n\u001b[1;32m 426\u001b[0m \u001b[38;5;124;03mLoads the tokenizer configuration from a pretrained model tokenizer configuration.\u001b[39;00m\n\u001b[1;32m 427\u001b[0m \n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 484\u001b[0m \u001b[38;5;124;03mtokenizer_config = get_tokenizer_config(\"tokenizer-test\")\u001b[39;00m\n\u001b[1;32m 485\u001b[0m \u001b[38;5;124;03m```\"\"\"\u001b[39;00m\n\u001b[1;32m 486\u001b[0m commit_hash \u001b[38;5;241m=\u001b[39m kwargs\u001b[38;5;241m.\u001b[39mget(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m_commit_hash\u001b[39m\u001b[38;5;124m\"\u001b[39m, \u001b[38;5;28;01mNone\u001b[39;00m)\n\u001b[0;32m--> 487\u001b[0m resolved_config_file \u001b[38;5;241m=\u001b[39m \u001b[43mcached_file\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 488\u001b[0m \u001b[43m \u001b[49m\u001b[43mpretrained_model_name_or_path\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 489\u001b[0m \u001b[43m \u001b[49m\u001b[43mTOKENIZER_CONFIG_FILE\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 490\u001b[0m \u001b[43m \u001b[49m\u001b[43mcache_dir\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mcache_dir\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 491\u001b[0m \u001b[43m \u001b[49m\u001b[43mforce_download\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mforce_download\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 492\u001b[0m \u001b[43m \u001b[49m\u001b[43mresume_download\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mresume_download\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 493\u001b[0m \u001b[43m \u001b[49m\u001b[43mproxies\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mproxies\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 494\u001b[0m \u001b[43m \u001b[49m\u001b[43muse_auth_token\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43muse_auth_token\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 495\u001b[0m \u001b[43m \u001b[49m\u001b[43mrevision\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mrevision\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 496\u001b[0m \u001b[43m \u001b[49m\u001b[43mlocal_files_only\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mlocal_files_only\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 497\u001b[0m \u001b[43m \u001b[49m\u001b[43msubfolder\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43msubfolder\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 498\u001b[0m \u001b[43m \u001b[49m\u001b[43m_raise_exceptions_for_missing_entries\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 499\u001b[0m \u001b[43m \u001b[49m\u001b[43m_raise_exceptions_for_connection_errors\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 500\u001b[0m \u001b[43m \u001b[49m\u001b[43m_commit_hash\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mcommit_hash\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 501\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 502\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m resolved_config_file \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[1;32m 503\u001b[0m logger\u001b[38;5;241m.\u001b[39minfo(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mCould not locate the tokenizer configuration file, will try to use the model config instead.\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/transformers/utils/hub.py:417\u001b[0m, in \u001b[0;36mcached_file\u001b[0;34m(path_or_repo_id, filename, cache_dir, force_download, resume_download, proxies, use_auth_token, revision, local_files_only, subfolder, repo_type, user_agent, _raise_exceptions_for_missing_entries, _raise_exceptions_for_connection_errors, _commit_hash)\u001b[0m\n\u001b[1;32m 414\u001b[0m user_agent \u001b[38;5;241m=\u001b[39m http_user_agent(user_agent)\n\u001b[1;32m 415\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m 416\u001b[0m \u001b[38;5;66;03m# Load from URL or cache if already cached\u001b[39;00m\n\u001b[0;32m--> 417\u001b[0m resolved_file \u001b[38;5;241m=\u001b[39m \u001b[43mhf_hub_download\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 418\u001b[0m \u001b[43m \u001b[49m\u001b[43mpath_or_repo_id\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 419\u001b[0m \u001b[43m \u001b[49m\u001b[43mfilename\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 420\u001b[0m \u001b[43m \u001b[49m\u001b[43msubfolder\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mNone\u001b[39;49;00m\u001b[43m \u001b[49m\u001b[38;5;28;43;01mif\u001b[39;49;00m\u001b[43m \u001b[49m\u001b[38;5;28;43mlen\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43msubfolder\u001b[49m\u001b[43m)\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m==\u001b[39;49m\u001b[43m \u001b[49m\u001b[38;5;241;43m0\u001b[39;49m\u001b[43m \u001b[49m\u001b[38;5;28;43;01melse\u001b[39;49;00m\u001b[43m \u001b[49m\u001b[43msubfolder\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 421\u001b[0m \u001b[43m \u001b[49m\u001b[43mrepo_type\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mrepo_type\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 422\u001b[0m \u001b[43m \u001b[49m\u001b[43mrevision\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mrevision\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 423\u001b[0m \u001b[43m \u001b[49m\u001b[43mcache_dir\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mcache_dir\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 424\u001b[0m \u001b[43m \u001b[49m\u001b[43muser_agent\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43muser_agent\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 425\u001b[0m \u001b[43m \u001b[49m\u001b[43mforce_download\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mforce_download\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 426\u001b[0m \u001b[43m \u001b[49m\u001b[43mproxies\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mproxies\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 427\u001b[0m \u001b[43m \u001b[49m\u001b[43mresume_download\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mresume_download\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 428\u001b[0m \u001b[43m \u001b[49m\u001b[43muse_auth_token\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43muse_auth_token\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 429\u001b[0m \u001b[43m \u001b[49m\u001b[43mlocal_files_only\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mlocal_files_only\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 430\u001b[0m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 432\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m RepositoryNotFoundError:\n\u001b[1;32m 433\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mEnvironmentError\u001b[39;00m(\n\u001b[1;32m 434\u001b[0m \u001b[38;5;124mf\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;132;01m{\u001b[39;00mpath_or_repo_id\u001b[38;5;132;01m}\u001b[39;00m\u001b[38;5;124m is not a local folder and is not a valid model identifier \u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m 435\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mlisted on \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mhttps://huggingface.co/models\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;130;01m\\n\u001b[39;00m\u001b[38;5;124mIf this is a private repository, make sure to \u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m 436\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mpass a token having permission to this repo with `use_auth_token` or log in with \u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m 437\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124m`huggingface-cli login` and pass `use_auth_token=True`.\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m 438\u001b[0m )\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py:120\u001b[0m, in \u001b[0;36mvalidate_hf_hub_args.._inner_fn\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m 117\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m check_use_auth_token:\n\u001b[1;32m 118\u001b[0m kwargs \u001b[38;5;241m=\u001b[39m smoothly_deprecate_use_auth_token(fn_name\u001b[38;5;241m=\u001b[39mfn\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, has_token\u001b[38;5;241m=\u001b[39mhas_token, kwargs\u001b[38;5;241m=\u001b[39mkwargs)\n\u001b[0;32m--> 120\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfn\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/huggingface_hub/file_download.py:1195\u001b[0m, in \u001b[0;36mhf_hub_download\u001b[0;34m(repo_id, filename, subfolder, repo_type, revision, library_name, library_version, cache_dir, local_dir, local_dir_use_symlinks, user_agent, force_download, force_filename, proxies, etag_timeout, resume_download, token, local_files_only, legacy_cache_layout)\u001b[0m\n\u001b[1;32m 1193\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[1;32m 1194\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m-> 1195\u001b[0m metadata \u001b[38;5;241m=\u001b[39m \u001b[43mget_hf_file_metadata\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 1196\u001b[0m \u001b[43m \u001b[49m\u001b[43murl\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43murl\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1197\u001b[0m \u001b[43m \u001b[49m\u001b[43mtoken\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtoken\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1198\u001b[0m \u001b[43m \u001b[49m\u001b[43mproxies\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mproxies\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1199\u001b[0m \u001b[43m \u001b[49m\u001b[43mtimeout\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43metag_timeout\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1200\u001b[0m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1201\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m EntryNotFoundError \u001b[38;5;28;01mas\u001b[39;00m http_error:\n\u001b[1;32m 1202\u001b[0m \u001b[38;5;66;03m# Cache the non-existence of the file and raise\u001b[39;00m\n\u001b[1;32m 1203\u001b[0m commit_hash \u001b[38;5;241m=\u001b[39m http_error\u001b[38;5;241m.\u001b[39mresponse\u001b[38;5;241m.\u001b[39mheaders\u001b[38;5;241m.\u001b[39mget(HUGGINGFACE_HEADER_X_REPO_COMMIT)\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py:120\u001b[0m, in \u001b[0;36mvalidate_hf_hub_args.._inner_fn\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m 117\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m check_use_auth_token:\n\u001b[1;32m 118\u001b[0m kwargs \u001b[38;5;241m=\u001b[39m smoothly_deprecate_use_auth_token(fn_name\u001b[38;5;241m=\u001b[39mfn\u001b[38;5;241m.\u001b[39m\u001b[38;5;18m__name__\u001b[39m, has_token\u001b[38;5;241m=\u001b[39mhas_token, kwargs\u001b[38;5;241m=\u001b[39mkwargs)\n\u001b[0;32m--> 120\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfn\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/huggingface_hub/file_download.py:1532\u001b[0m, in \u001b[0;36mget_hf_file_metadata\u001b[0;34m(url, token, proxies, timeout)\u001b[0m\n\u001b[1;32m 1529\u001b[0m headers[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mAccept-Encoding\u001b[39m\u001b[38;5;124m\"\u001b[39m] \u001b[38;5;241m=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124midentity\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;66;03m# prevent any compression => we want to know the real size of the file\u001b[39;00m\n\u001b[1;32m 1531\u001b[0m \u001b[38;5;66;03m# Retrieve metadata\u001b[39;00m\n\u001b[0;32m-> 1532\u001b[0m r \u001b[38;5;241m=\u001b[39m \u001b[43m_request_wrapper\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 1533\u001b[0m \u001b[43m \u001b[49m\u001b[43mmethod\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mHEAD\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1534\u001b[0m \u001b[43m \u001b[49m\u001b[43murl\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43murl\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1535\u001b[0m \u001b[43m \u001b[49m\u001b[43mheaders\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mheaders\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1536\u001b[0m \u001b[43m \u001b[49m\u001b[43mallow_redirects\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 1537\u001b[0m \u001b[43m \u001b[49m\u001b[43mfollow_relative_redirects\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 1538\u001b[0m \u001b[43m \u001b[49m\u001b[43mproxies\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mproxies\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1539\u001b[0m \u001b[43m \u001b[49m\u001b[43mtimeout\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtimeout\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 1540\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1541\u001b[0m hf_raise_for_status(r)\n\u001b[1;32m 1543\u001b[0m \u001b[38;5;66;03m# Return\u001b[39;00m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/huggingface_hub/file_download.py:407\u001b[0m, in \u001b[0;36m_request_wrapper\u001b[0;34m(method, url, max_retries, base_wait_time, max_wait_time, timeout, follow_relative_redirects, **params)\u001b[0m\n\u001b[1;32m 405\u001b[0m \u001b[38;5;66;03m# 2. Force relative redirection\u001b[39;00m\n\u001b[1;32m 406\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m follow_relative_redirects:\n\u001b[0;32m--> 407\u001b[0m response \u001b[38;5;241m=\u001b[39m \u001b[43m_request_wrapper\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 408\u001b[0m \u001b[43m \u001b[49m\u001b[43mmethod\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmethod\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 409\u001b[0m \u001b[43m \u001b[49m\u001b[43murl\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43murl\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 410\u001b[0m \u001b[43m \u001b[49m\u001b[43mmax_retries\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmax_retries\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 411\u001b[0m \u001b[43m \u001b[49m\u001b[43mbase_wait_time\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mbase_wait_time\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 412\u001b[0m \u001b[43m \u001b[49m\u001b[43mmax_wait_time\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmax_wait_time\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 413\u001b[0m \u001b[43m \u001b[49m\u001b[43mtimeout\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtimeout\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 414\u001b[0m \u001b[43m \u001b[49m\u001b[43mfollow_relative_redirects\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 415\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mparams\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 416\u001b[0m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 418\u001b[0m \u001b[38;5;66;03m# If redirection, we redirect only relative paths.\u001b[39;00m\n\u001b[1;32m 419\u001b[0m \u001b[38;5;66;03m# This is useful in case of a renamed repository.\u001b[39;00m\n\u001b[1;32m 420\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;241m300\u001b[39m \u001b[38;5;241m<\u001b[39m\u001b[38;5;241m=\u001b[39m response\u001b[38;5;241m.\u001b[39mstatus_code \u001b[38;5;241m<\u001b[39m\u001b[38;5;241m=\u001b[39m \u001b[38;5;241m399\u001b[39m:\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/huggingface_hub/file_download.py:442\u001b[0m, in \u001b[0;36m_request_wrapper\u001b[0;34m(method, url, max_retries, base_wait_time, max_wait_time, timeout, follow_relative_redirects, **params)\u001b[0m\n\u001b[1;32m 439\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m response\n\u001b[1;32m 441\u001b[0m \u001b[38;5;66;03m# 3. Exponential backoff\u001b[39;00m\n\u001b[0;32m--> 442\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mhttp_backoff\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 443\u001b[0m \u001b[43m \u001b[49m\u001b[43mmethod\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmethod\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 444\u001b[0m \u001b[43m \u001b[49m\u001b[43murl\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43murl\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 445\u001b[0m \u001b[43m \u001b[49m\u001b[43mmax_retries\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmax_retries\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 446\u001b[0m \u001b[43m \u001b[49m\u001b[43mbase_wait_time\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mbase_wait_time\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 447\u001b[0m \u001b[43m \u001b[49m\u001b[43mmax_wait_time\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmax_wait_time\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 448\u001b[0m \u001b[43m \u001b[49m\u001b[43mretry_on_exceptions\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mConnectTimeout\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mProxyError\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 449\u001b[0m \u001b[43m \u001b[49m\u001b[43mretry_on_status_codes\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 450\u001b[0m \u001b[43m \u001b[49m\u001b[43mtimeout\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtimeout\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 451\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mparams\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 452\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/huggingface_hub/utils/_http.py:212\u001b[0m, in \u001b[0;36mhttp_backoff\u001b[0;34m(method, url, max_retries, base_wait_time, max_wait_time, retry_on_exceptions, retry_on_status_codes, **kwargs)\u001b[0m\n\u001b[1;32m 209\u001b[0m kwargs[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mdata\u001b[39m\u001b[38;5;124m\"\u001b[39m]\u001b[38;5;241m.\u001b[39mseek(io_obj_initial_pos)\n\u001b[1;32m 211\u001b[0m \u001b[38;5;66;03m# Perform request and return if status_code is not in the retry list.\u001b[39;00m\n\u001b[0;32m--> 212\u001b[0m response \u001b[38;5;241m=\u001b[39m \u001b[43msession\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mrequest\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmethod\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmethod\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43murl\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43murl\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 213\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m response\u001b[38;5;241m.\u001b[39mstatus_code \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;129;01min\u001b[39;00m retry_on_status_codes:\n\u001b[1;32m 214\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m response\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/requests/sessions.py:587\u001b[0m, in \u001b[0;36mSession.request\u001b[0;34m(self, method, url, params, data, headers, cookies, files, auth, timeout, allow_redirects, proxies, hooks, stream, verify, cert, json)\u001b[0m\n\u001b[1;32m 582\u001b[0m send_kwargs \u001b[38;5;241m=\u001b[39m {\n\u001b[1;32m 583\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mtimeout\u001b[39m\u001b[38;5;124m\"\u001b[39m: timeout,\n\u001b[1;32m 584\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mallow_redirects\u001b[39m\u001b[38;5;124m\"\u001b[39m: allow_redirects,\n\u001b[1;32m 585\u001b[0m }\n\u001b[1;32m 586\u001b[0m send_kwargs\u001b[38;5;241m.\u001b[39mupdate(settings)\n\u001b[0;32m--> 587\u001b[0m resp \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43msend\u001b[49m\u001b[43m(\u001b[49m\u001b[43mprep\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43msend_kwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 589\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m resp\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/requests/sessions.py:701\u001b[0m, in \u001b[0;36mSession.send\u001b[0;34m(self, request, **kwargs)\u001b[0m\n\u001b[1;32m 698\u001b[0m start \u001b[38;5;241m=\u001b[39m preferred_clock()\n\u001b[1;32m 700\u001b[0m \u001b[38;5;66;03m# Send the request\u001b[39;00m\n\u001b[0;32m--> 701\u001b[0m r \u001b[38;5;241m=\u001b[39m \u001b[43madapter\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43msend\u001b[49m\u001b[43m(\u001b[49m\u001b[43mrequest\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 703\u001b[0m \u001b[38;5;66;03m# Total elapsed time of the request (approximately)\u001b[39;00m\n\u001b[1;32m 704\u001b[0m elapsed \u001b[38;5;241m=\u001b[39m preferred_clock() \u001b[38;5;241m-\u001b[39m start\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/requests/adapters.py:487\u001b[0m, in \u001b[0;36mHTTPAdapter.send\u001b[0;34m(self, request, stream, timeout, verify, cert, proxies)\u001b[0m\n\u001b[1;32m 484\u001b[0m timeout \u001b[38;5;241m=\u001b[39m TimeoutSauce(connect\u001b[38;5;241m=\u001b[39mtimeout, read\u001b[38;5;241m=\u001b[39mtimeout)\n\u001b[1;32m 486\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m--> 487\u001b[0m resp \u001b[38;5;241m=\u001b[39m \u001b[43mconn\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43murlopen\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 488\u001b[0m \u001b[43m \u001b[49m\u001b[43mmethod\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mrequest\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mmethod\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 489\u001b[0m \u001b[43m \u001b[49m\u001b[43murl\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43murl\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 490\u001b[0m \u001b[43m \u001b[49m\u001b[43mbody\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mrequest\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbody\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 491\u001b[0m \u001b[43m \u001b[49m\u001b[43mheaders\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mrequest\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mheaders\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 492\u001b[0m \u001b[43m \u001b[49m\u001b[43mredirect\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 493\u001b[0m \u001b[43m \u001b[49m\u001b[43massert_same_host\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 494\u001b[0m \u001b[43m \u001b[49m\u001b[43mpreload_content\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 495\u001b[0m \u001b[43m \u001b[49m\u001b[43mdecode_content\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mFalse\u001b[39;49;00m\u001b[43m,\u001b[49m\n\u001b[1;32m 496\u001b[0m \u001b[43m \u001b[49m\u001b[43mretries\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mmax_retries\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 497\u001b[0m \u001b[43m \u001b[49m\u001b[43mtimeout\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtimeout\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 498\u001b[0m \u001b[43m \u001b[49m\u001b[43mchunked\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mchunked\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 499\u001b[0m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 501\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m (ProtocolError, \u001b[38;5;167;01mOSError\u001b[39;00m) \u001b[38;5;28;01mas\u001b[39;00m err:\n\u001b[1;32m 502\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mConnectionError\u001b[39;00m(err, request\u001b[38;5;241m=\u001b[39mrequest)\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/urllib3/connectionpool.py:703\u001b[0m, in \u001b[0;36mHTTPConnectionPool.urlopen\u001b[0;34m(self, method, url, body, headers, retries, redirect, assert_same_host, timeout, pool_timeout, release_conn, chunked, body_pos, **response_kw)\u001b[0m\n\u001b[1;32m 700\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_prepare_proxy(conn)\n\u001b[1;32m 702\u001b[0m \u001b[38;5;66;03m# Make the request on the httplib connection object.\u001b[39;00m\n\u001b[0;32m--> 703\u001b[0m httplib_response \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_make_request\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 704\u001b[0m \u001b[43m \u001b[49m\u001b[43mconn\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 705\u001b[0m \u001b[43m \u001b[49m\u001b[43mmethod\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 706\u001b[0m \u001b[43m \u001b[49m\u001b[43murl\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 707\u001b[0m \u001b[43m \u001b[49m\u001b[43mtimeout\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtimeout_obj\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 708\u001b[0m \u001b[43m \u001b[49m\u001b[43mbody\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mbody\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 709\u001b[0m \u001b[43m \u001b[49m\u001b[43mheaders\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mheaders\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 710\u001b[0m \u001b[43m \u001b[49m\u001b[43mchunked\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mchunked\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 711\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 713\u001b[0m \u001b[38;5;66;03m# If we're going to release the connection in ``finally:``, then\u001b[39;00m\n\u001b[1;32m 714\u001b[0m \u001b[38;5;66;03m# the response doesn't need to know about the connection. Otherwise\u001b[39;00m\n\u001b[1;32m 715\u001b[0m \u001b[38;5;66;03m# it will also try to release it and we'll have a double-release\u001b[39;00m\n\u001b[1;32m 716\u001b[0m \u001b[38;5;66;03m# mess.\u001b[39;00m\n\u001b[1;32m 717\u001b[0m response_conn \u001b[38;5;241m=\u001b[39m conn \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m release_conn \u001b[38;5;28;01melse\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/urllib3/connectionpool.py:386\u001b[0m, in \u001b[0;36mHTTPConnectionPool._make_request\u001b[0;34m(self, conn, method, url, timeout, chunked, **httplib_request_kw)\u001b[0m\n\u001b[1;32m 384\u001b[0m \u001b[38;5;66;03m# Trigger any extra validation we need to do.\u001b[39;00m\n\u001b[1;32m 385\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m--> 386\u001b[0m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_validate_conn\u001b[49m\u001b[43m(\u001b[49m\u001b[43mconn\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 387\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m (SocketTimeout, BaseSSLError) \u001b[38;5;28;01mas\u001b[39;00m e:\n\u001b[1;32m 388\u001b[0m \u001b[38;5;66;03m# Py2 raises this as a BaseSSLError, Py3 raises it as socket timeout.\u001b[39;00m\n\u001b[1;32m 389\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_raise_timeout(err\u001b[38;5;241m=\u001b[39me, url\u001b[38;5;241m=\u001b[39murl, timeout_value\u001b[38;5;241m=\u001b[39mconn\u001b[38;5;241m.\u001b[39mtimeout)\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/urllib3/connectionpool.py:1042\u001b[0m, in \u001b[0;36mHTTPSConnectionPool._validate_conn\u001b[0;34m(self, conn)\u001b[0m\n\u001b[1;32m 1040\u001b[0m \u001b[38;5;66;03m# Force connect early to allow us to validate the connection.\u001b[39;00m\n\u001b[1;32m 1041\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28mgetattr\u001b[39m(conn, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124msock\u001b[39m\u001b[38;5;124m\"\u001b[39m, \u001b[38;5;28;01mNone\u001b[39;00m): \u001b[38;5;66;03m# AppEngine might not have `.sock`\u001b[39;00m\n\u001b[0;32m-> 1042\u001b[0m \u001b[43mconn\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mconnect\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1044\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m conn\u001b[38;5;241m.\u001b[39mis_verified:\n\u001b[1;32m 1045\u001b[0m warnings\u001b[38;5;241m.\u001b[39mwarn(\n\u001b[1;32m 1046\u001b[0m (\n\u001b[1;32m 1047\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mUnverified HTTPS request is being made to host \u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;132;01m%s\u001b[39;00m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124m. \u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 1052\u001b[0m InsecureRequestWarning,\n\u001b[1;32m 1053\u001b[0m )\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/urllib3/connection.py:363\u001b[0m, in \u001b[0;36mHTTPSConnection.connect\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 361\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mconnect\u001b[39m(\u001b[38;5;28mself\u001b[39m):\n\u001b[1;32m 362\u001b[0m \u001b[38;5;66;03m# Add certificate verification\u001b[39;00m\n\u001b[0;32m--> 363\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39msock \u001b[38;5;241m=\u001b[39m conn \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_new_conn\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 364\u001b[0m hostname \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mhost\n\u001b[1;32m 365\u001b[0m tls_in_tls \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mFalse\u001b[39;00m\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/urllib3/connection.py:174\u001b[0m, in \u001b[0;36mHTTPConnection._new_conn\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 171\u001b[0m extra_kw[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124msocket_options\u001b[39m\u001b[38;5;124m\"\u001b[39m] \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39msocket_options\n\u001b[1;32m 173\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m--> 174\u001b[0m conn \u001b[38;5;241m=\u001b[39m \u001b[43mconnection\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcreate_connection\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 175\u001b[0m \u001b[43m \u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_dns_host\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mport\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtimeout\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mextra_kw\u001b[49m\n\u001b[1;32m 176\u001b[0m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 178\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m SocketTimeout:\n\u001b[1;32m 179\u001b[0m \u001b[38;5;28;01mraise\u001b[39;00m ConnectTimeoutError(\n\u001b[1;32m 180\u001b[0m \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m 181\u001b[0m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mConnection to \u001b[39m\u001b[38;5;132;01m%s\u001b[39;00m\u001b[38;5;124m timed out. (connect timeout=\u001b[39m\u001b[38;5;132;01m%s\u001b[39;00m\u001b[38;5;124m)\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m 182\u001b[0m \u001b[38;5;241m%\u001b[39m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mhost, \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mtimeout),\n\u001b[1;32m 183\u001b[0m )\n","\u001b[0;31mKeyboardInterrupt\u001b[0m: "],"ename":"KeyboardInterrupt","evalue":"","output_type":"error"}],"execution_count":309},{"cell_type":"code","source":"train_df, target_columns = make_train_dataset(train_seq_df, train_label_df, max_label=Config.max_label)\ntrain_df = add_fold_idx(config=config, train_df=train_df, target_columns=target_columns)\n\nif config.train:\n for i_fold in range(config.n_fold):\n if i_fold not in config.trn_fold:\n continue\n\n train_df_fold = train_df[train_df[\"fold\"] != i_fold]\n valid_df_fold = train_df[train_df[\"fold\"] == i_fold]\n\n train_loop(\n config=config,\n name=f\"fold_{i_fold}\",\n train_df=train_df_fold,\n valid_df=valid_df_fold,\n target_columns=target_columns,\n device=DEVICE,\n )\n ","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:27:31.760536Z","iopub.execute_input":"2026-08-14T13:27:31.760962Z","iopub.status.idle":"2026-08-14T13:27:31.890003Z","shell.execute_reply.started":"2026-08-14T13:27:31.760937Z","shell.execute_reply":"2026-08-14T13:27:31.888584Z"}},"outputs":[{"traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mTypeError\u001b[0m Traceback (most recent call last)","Cell \u001b[0;32mIn[308], line 12\u001b[0m\n\u001b[1;32m 9\u001b[0m train_df_fold \u001b[38;5;241m=\u001b[39m train_df[train_df[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mfold\u001b[39m\u001b[38;5;124m\"\u001b[39m] \u001b[38;5;241m!=\u001b[39m i_fold]\n\u001b[1;32m 10\u001b[0m valid_df_fold \u001b[38;5;241m=\u001b[39m train_df[train_df[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mfold\u001b[39m\u001b[38;5;124m\"\u001b[39m] \u001b[38;5;241m==\u001b[39m i_fold]\n\u001b[0;32m---> 12\u001b[0m \u001b[43mtrain_loop\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 13\u001b[0m \u001b[43m \u001b[49m\u001b[43mconfig\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mconfig\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 14\u001b[0m \u001b[43m \u001b[49m\u001b[43mname\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43mf\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mfold_\u001b[39;49m\u001b[38;5;132;43;01m{\u001b[39;49;00m\u001b[43mi_fold\u001b[49m\u001b[38;5;132;43;01m}\u001b[39;49;00m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m 15\u001b[0m \u001b[43m \u001b[49m\u001b[43mtrain_df\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtrain_df_fold\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 16\u001b[0m \u001b[43m \u001b[49m\u001b[43mvalid_df\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mvalid_df_fold\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 17\u001b[0m \u001b[43m \u001b[49m\u001b[43mtarget_columns\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtarget_columns\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 18\u001b[0m \u001b[43m \u001b[49m\u001b[43mdevice\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mDEVICE\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m 19\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n","\u001b[0;31mTypeError\u001b[0m: train_loop() missing 1 required positional argument: 'tokenizer'"],"ename":"TypeError","evalue":"train_loop() missing 1 required positional argument: 'tokenizer'","output_type":"error"}],"execution_count":308},{"cell_type":"markdown","source":"## Evaluation","metadata":{}},{"cell_type":"code","source":"def make_submission_formatted_outputs_df(outputs: dict) -> pd.DataFrame:\n data = {\n \"id\": np.repeat(outputs[\"id\"], len(outputs[\"target_names\"])),\n \"terms\": np.tile(outputs[\"target_names\"], len(outputs[\"id\"])),\n \"preds\": outputs[\"predictions\"].ravel(),\n }\n submission_df = pd.DataFrame(data)\n return submission_df\n\n\ndef make_crosstab_formatted_outputs_df(outputs: dict) -> pd.DataFrame:\n output_df = pd.DataFrame(\n outputs[\"predictions\"],\n columns=outputs[\"target_names\"],\n index=pd.Index(outputs[\"id\"], name=\"id\"),\n )\n return output_df\n\nif config.evaluation:\n pred_dfs, gt_dfs = [], []\n for i_fold in range(config.n_fold):\n if i_fold not in Config.trn_fold:\n continue\n\n # get data\n outputs = joblib.load(EXP_PREDS / f\"fold_{i_fold}_best.pkl\")\n pred_df = make_submission_formatted_outputs_df(outputs)\n gt_df = train_terms_df[train_terms_df[\"EntryID\"].isin(pred_df[\"id\"])].reset_index(drop=True)\n\n # calc score\n score, score_df = get_score(gt_df=gt_df, pred_df=pred_df, th_step=0.1)\n\n # save result\n LOGGER.info(f\"fold {i_fold} : score = {score:.4f}\")\n score_df.to_csv(EXP_REPORTS / f\"score_df_fold_{i_fold}.csv\", index=False)\n\n # stack\n pred_dfs.append(pred_df)\n gt_dfs.append(gt_df)\n\n all_pred_df = pd.concat(pred_dfs, axis=0).reset_index(drop=True)\n all_gt_df = pd.concat(gt_dfs, axis=0).reset_index(drop=True)\n all_score, all_score_df = get_score(gt_df=all_gt_df, pred_df=all_pred_df, th_step=0.1)\n LOGGER.info(f\"all : score = {all_score:.4f}\")\n\n joblib.dump(all_pred_df, EXP_PREDS / \"all_pred_df.pkl\")\n joblib.dump(all_gt_df, EXP_PREDS / \"all_gt_df.pkl\")\n joblib.dump(all_score_df, EXP_PREDS / \"all_score_df.pkl\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:13:28.087971Z","iopub.execute_input":"2026-08-14T13:13:28.088342Z","iopub.status.idle":"2026-08-14T13:13:28.14886Z","shell.execute_reply.started":"2026-08-14T13:13:28.088317Z","shell.execute_reply":"2026-08-14T13:13:28.147415Z"}},"outputs":[{"traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mFileNotFoundError\u001b[0m Traceback (most recent call last)","Cell \u001b[0;32mIn[299], line 26\u001b[0m\n\u001b[1;32m 23\u001b[0m \u001b[38;5;28;01mcontinue\u001b[39;00m\n\u001b[1;32m 25\u001b[0m \u001b[38;5;66;03m# get data\u001b[39;00m\n\u001b[0;32m---> 26\u001b[0m outputs \u001b[38;5;241m=\u001b[39m \u001b[43mjoblib\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mload\u001b[49m\u001b[43m(\u001b[49m\u001b[43mEXP_PREDS\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m/\u001b[39;49m\u001b[43m \u001b[49m\u001b[38;5;124;43mf\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mfold_\u001b[39;49m\u001b[38;5;132;43;01m{\u001b[39;49;00m\u001b[43mi_fold\u001b[49m\u001b[38;5;132;43;01m}\u001b[39;49;00m\u001b[38;5;124;43m_best.pkl\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m)\u001b[49m\n\u001b[1;32m 27\u001b[0m pred_df \u001b[38;5;241m=\u001b[39m make_submission_formatted_outputs_df(outputs)\n\u001b[1;32m 28\u001b[0m gt_df \u001b[38;5;241m=\u001b[39m train_terms_df[train_terms_df[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mEntryID\u001b[39m\u001b[38;5;124m\"\u001b[39m]\u001b[38;5;241m.\u001b[39misin(pred_df[\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mid\u001b[39m\u001b[38;5;124m\"\u001b[39m])]\u001b[38;5;241m.\u001b[39mreset_index(drop\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n","File \u001b[0;32m/opt/conda/lib/python3.10/site-packages/joblib/numpy_pickle.py:650\u001b[0m, in \u001b[0;36mload\u001b[0;34m(filename, mmap_mode)\u001b[0m\n\u001b[1;32m 648\u001b[0m obj \u001b[38;5;241m=\u001b[39m _unpickle(fobj)\n\u001b[1;32m 649\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m--> 650\u001b[0m \u001b[38;5;28;01mwith\u001b[39;00m \u001b[38;5;28;43mopen\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mfilename\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[38;5;124;43mrb\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[43m)\u001b[49m \u001b[38;5;28;01mas\u001b[39;00m f:\n\u001b[1;32m 651\u001b[0m \u001b[38;5;28;01mwith\u001b[39;00m _read_fileobject(f, filename, mmap_mode) \u001b[38;5;28;01mas\u001b[39;00m fobj:\n\u001b[1;32m 652\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(fobj, \u001b[38;5;28mstr\u001b[39m):\n\u001b[1;32m 653\u001b[0m \u001b[38;5;66;03m# if the returned file object is a string, this means we\u001b[39;00m\n\u001b[1;32m 654\u001b[0m \u001b[38;5;66;03m# try to load a pickle file generated with an version of\u001b[39;00m\n\u001b[1;32m 655\u001b[0m \u001b[38;5;66;03m# Joblib so we load it with joblib compatibility function.\u001b[39;00m\n","\u001b[0;31mFileNotFoundError\u001b[0m: [Errno 2] No such file or directory: '/kaggle/working/exp_preds/fold_0_best.pkl'"],"ename":"FileNotFoundError","evalue":"[Errno 2] No such file or directory: '/kaggle/working/exp_preds/fold_0_best.pkl'","output_type":"error"}],"execution_count":299},{"cell_type":"markdown","source":"## Inference","metadata":{}},{"cell_type":"code","source":"if Config.inference:\n test_pred_df = pd.DataFrame()\n predictions = 0\n n_model = 0\n for i_fold in range(Config.n_fold):\n if i_fold not in Config.trn_fold:\n continue\n\n outputs = get_predictions(\n config=Config, \n test_df=test_df, \n tokenizer=tokenizer, \n model_path=EXP_MODELS / f\"fold_{i_fold}.pth\", \n target_columns=target_columns, \n device=DEVICE)\n predictions += outputs[\"predictions\"]\n n_model += 1\n\n outputs[\"predictions\"] = predictions / n_model # fold ave, same id & targets\n test_pred_df = make_submission_formatted_outputs_df(outputs)\n test_pred_df.reset_index(drop=True, inplace=True)\n joblib.dump(test_pred_df, EXP_PREDS / \"test_pred_df.pkl\")\n test_pred_df.to_csv(SUBMISSIONS / f\"{Config.name}.tsv\", header=False, index=False, sep=\"\\t\")\n","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:06:38.820525Z","iopub.execute_input":"2026-08-14T13:06:38.821289Z","iopub.status.idle":"2026-08-14T13:06:38.858672Z","shell.execute_reply.started":"2026-08-14T13:06:38.82126Z","shell.execute_reply":"2026-08-14T13:06:38.856921Z"}},"outputs":[{"traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mAttributeError\u001b[0m Traceback (most recent call last)","Cell \u001b[0;32mIn[294], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[43mConfig\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43minference\u001b[49m:\n\u001b[1;32m 2\u001b[0m test_pred_df \u001b[38;5;241m=\u001b[39m pd\u001b[38;5;241m.\u001b[39mDataFrame()\n\u001b[1;32m 3\u001b[0m predictions \u001b[38;5;241m=\u001b[39m \u001b[38;5;241m0\u001b[39m\n","\u001b[0;31mAttributeError\u001b[0m: type object 'Config' has no attribute 'inference'"],"ename":"AttributeError","evalue":"type object 'Config' has no attribute 'inference'","output_type":"error"}],"execution_count":294},{"cell_type":"markdown","source":"## Submission","metadata":{}},{"cell_type":"code","source":"if Config.submission:\n submission_df = joblib.load(EXP_PREDS / \"test_pred_df.pkl\")\n upload_submission_file_to_kaggle(dataset_name=f\"CAFA5-{Config.name}\", submission_df=submission_df)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2026-08-14T13:06:43.479651Z","iopub.execute_input":"2026-08-14T13:06:43.479996Z","iopub.status.idle":"2026-08-14T13:06:43.51961Z","shell.execute_reply.started":"2026-08-14T13:06:43.479975Z","shell.execute_reply":"2026-08-14T13:06:43.518642Z"}},"outputs":[{"traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mAttributeError\u001b[0m Traceback (most recent call last)","Cell \u001b[0;32mIn[295], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[43mConfig\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43msubmission\u001b[49m:\n\u001b[1;32m 2\u001b[0m submission_df \u001b[38;5;241m=\u001b[39m joblib\u001b[38;5;241m.\u001b[39mload(EXP_PREDS \u001b[38;5;241m/\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mtest_pred_df.pkl\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m 3\u001b[0m upload_submission_file_to_kaggle(dataset_name\u001b[38;5;241m=\u001b[39m\u001b[38;5;124mf\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mCAFA5-\u001b[39m\u001b[38;5;132;01m{\u001b[39;00mConfig\u001b[38;5;241m.\u001b[39mname\u001b[38;5;132;01m}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m, submission_df\u001b[38;5;241m=\u001b[39msubmission_df)\n","\u001b[0;31mAttributeError\u001b[0m: type object 'Config' has no attribute 'submission'"],"ename":"AttributeError","evalue":"type object 'Config' has no attribute 'submission'","output_type":"error"}],"execution_count":295}]}