{ "version": "1.0", "truncation": null, "padding": null, "added_tokens": [ { "id": 0, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 1, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 2, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 121, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 122, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 123, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 124, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 125, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 126, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 127, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 128, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 129, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 130, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 131, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 132, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 133, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 134, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 135, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 136, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 137, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 138, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 139, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 140, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 141, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 142, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 143, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 144, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 145, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 146, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 147, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 148, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 149, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 150, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 151, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 152, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 153, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 154, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 155, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 156, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 157, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 158, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 159, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 160, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 161, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 162, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 163, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 164, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 165, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 166, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 167, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 168, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 169, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 170, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 171, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 172, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 173, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 174, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 175, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 176, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 177, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 178, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 179, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 180, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 181, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 182, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 183, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 184, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 185, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 186, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 187, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 188, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 189, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 190, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 191, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 192, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 193, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 194, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 195, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 196, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 197, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 198, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 199, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 200, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 201, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 202, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 203, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 204, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 205, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 206, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 207, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 208, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 209, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 210, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 211, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 212, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 213, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 214, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 215, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 216, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 217, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 218, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 219, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 220, "content": "", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 221, "content": ".", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 222, "content": "P", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 223, "content": ">", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 224, "content": "<", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 225, "content": "Pd", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "special": false }, { "id": 226, "content": "REACTANT:", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 227, "content": "PRODUCT:", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true }, { "id": 228, "content": "REAGENT:", "single_word": false, "lstrip": false, "rstrip": false, "normalized": false, "special": true } ], "normalizer": { "type": "Sequence", "normalizers": [ { "type": "Nmt" }, { "type": "NFKC" }, { "type": "Replace", "pattern": { "Regex": " {2,}" }, "content": " " } ] }, "pre_tokenizer": { "type": "Sequence", "pretokenizers": [ { "type": "Metaspace", "replacement": "▁", "add_prefix_space": true }, { "type": "Digits", "individual_digits": true }, { "type": "Punctuation", "behavior": "Isolated" } ] }, "post_processor": { "type": "TemplateProcessing", "single": [ { "Sequence": { "id": "A", "type_id": 0 } }, { "SpecialToken": { "id": "", "type_id": 0 } } ], "pair": [ { "Sequence": { "id": "A", "type_id": 0 } }, { "Sequence": { "id": "B", "type_id": 1 } } ], "special_tokens": { "": { "id": "", "ids": [ 1 ], "tokens": [ "" ] } } }, "decoder": { "type": "Metaspace", "replacement": "▁", "add_prefix_space": true }, "model": { "type": "Unigram", "unk_id": 2, "vocab": [ [ "", 0.0 ], [ "", 0.0 ], [ "", 0.0 ], [ "C", -1.934062026530869 ], [ "c", -2.005135123981896 ], [ ")", -2.3305416341016603 ], [ "(", -2.3305416341016603 ], [ "1", -2.659489340254079 ], [ "2", -3.0147314517826524 ], [ "O", -3.1705660462226426 ], [ "]", -3.306692943444802 ], [ "[", -3.306692943444802 ], [ "N", -3.3379525475363145 ], [ "@", -3.375613737475505 ], [ "H", -3.3960003181228675 ], [ "=", -3.4342358584336026 ], [ "n", -3.9996436484078535 ], [ "3", -4.06624150723795 ], [ "+", -4.576587821405814 ], [ "ccc", -4.581329068949029 ], [ "cccc", -4.794168477161054 ], [ "F", -4.844444469349959 ], [ "▁C", -4.930731418706828 ], [ "-", -5.128151794820781 ], [ "CC", -5.188054847421637 ], [ "▁", -5.243375183264384 ], [ "▁CC", -5.43665410318811 ], [ "cc", -5.5072446210190025 ], [ "S", -5.607990514129369 ], [ "CCC", -5.610845771534942 ], [ "4", -5.6728058623804705 ], [ "CCN", -5.984229431727735 ], [ "▁CO", -6.136725098692515 ], [ "s", -6.163972117410355 ], [ "l", -6.244242813422005 ], [ "nc", -6.260369976115332 ], [ "Cl", -6.29041030199948 ], [ "NC", -6.301741325285667 ], [ "#", -6.475695381808137 ], [ "CCCC", -6.630889749196982 ], [ "Br", -6.708882955827546 ], [ "o", -6.8821690990922 ], [ "no", -7.085301372799254 ], [ "nn", -7.118314757276033 ], [ "CN", -7.148023077707199 ], [ "▁O", -7.184905330255651 ], [ "CCO", -7.223262672570243 ], [ "▁CN", -7.340957119808042 ], [ "ccs", -7.358439269448935 ], [ "▁CCC", -7.384181818093239 ], [ "OC", -7.401588967543117 ], [ "/", -7.412130091887384 ], [ "▁COC", -7.453342081307827 ], [ "nnc", -7.47808245293119 ], [ "cccn", -7.526117466256242 ], [ "▁CCOC", -7.581460400684973 ], [ "sc", -7.587476370908391 ], [ "cco", -7.601812309424911 ], [ "CCOCC", -7.70962259415791 ], [ "CCCN", -7.716050546596486 ], [ "oc", -7.76950331715161 ], [ "ccnc", -7.885377075776418 ], [ "▁CCO", -7.958354042052764 ], [ "cs", -7.963618836898841 ], [ "▁CCN", -7.969745690077621 ], [ "NCC", -7.982831223650715 ], [ "CNC", -8.015812540478317 ], [ "CCCO", -8.02283744281797 ], [ "nccn", -8.023781204384559 ], [ "CCCCC", -8.113410349713947 ], [ "csc", -8.123340128859317 ], [ "nnn", -8.16923209155111 ], [ "CO", -8.197076591417558 ], [ "cnc", -8.24948897316689 ], [ "OCC", -8.337244843539168 ], [ "ncc", -8.35350988501787 ], [ "CCNC", -8.374584543880331 ], [ "cnn", -8.379764985534148 ], [ "▁CCCC", -8.408946579060395 ], [ "ccn", -8.441316367356151 ], [ "▁CCCN", -8.455940453267049 ], [ "ncn", -8.509281787409483 ], [ "CCOC", -8.545791064619086 ], [ "5", -8.595231126796552 ], [ "OCCO", -8.620568633537287 ], [ "SCC", -8.675945846432537 ], [ "\\", -8.738088269693655 ], [ "▁CS", -8.745760398405787 ], [ "cn", -8.81287462848524 ], [ "CS", -8.852680543942768 ], [ "COC", -8.901532120821837 ], [ "▁CCCO", -8.971729419021525 ], [ "▁COCC", -8.986602855501445 ], [ "▁F", -9.08172188898926 ], [ "▁CCNC", -9.213239929783953 ], [ "nccc", -9.267617565078302 ], [ "▁CCS", -9.284123840101111 ], [ "▁N", -9.46465900903722 ], [ "▁CCCCO", -9.549659454320029 ], [ "NCCC", -9.592808545839064 ], [ "NCCN", -9.616093114468232 ], [ "CCS", -9.68568843530331 ], [ "cncc", -9.884757958758378 ], [ "CCCS", -10.07490132384844 ], [ "▁NC", -10.174180411834037 ], [ "▁COCCN", -10.182867019690027 ], [ "▁COCCO", -10.210382702506402 ], [ "▁COCCC", -10.390960567106395 ], [ "OCO", -10.445584736697652 ], [ "COCC", -10.469925833149428 ], [ "▁CCOCCN", -10.487938027145828 ], [ "SC", -10.868655865451124 ], [ "co", -11.175046369147724 ], [ "I", -11.188088269693656 ], [ "on", -11.199838029113598 ], [ "CCCCN", -11.273835881601958 ], [ "▁CCCCN", -11.694782342260607 ], [ "CCOCCN", -11.98760417613926 ], [ "r", -12.313357692926305 ], [ "B", -12.313457692926304 ], [ "CCCCO", -12.313457692926304 ] ] } }