bart-base-cantonese / tokenizer_config.json
indiejoseph's picture
Training in progress, step 2000
a34cbfe
raw
history blame
32.3 kB
{
"added_tokens_decoder": {
"0": {
"content": "[PAD]",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"100": {
"content": "[UNK]",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"101": {
"content": "[CLS]",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"102": {
"content": "[SEP]",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"103": {
"content": "[MASK]",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"104": {
"content": "[EOS]",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"51271": {
"content": "㖭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51272": {
"content": "䊦",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51273": {
"content": "㨂",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51274": {
"content": "䁪",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51275": {
"content": "㴝",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51276": {
"content": "䑛",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51277": {
"content": "㽇",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51278": {
"content": "䟕",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51279": {
"content": "㢌",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51280": {
"content": "㗇",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51281": {
"content": "䉪",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51282": {
"content": "䐥",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51283": {
"content": "䘆",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51284": {
"content": "㘉",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51285": {
"content": "䚗",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51286": {
"content": "鰦",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51287": {
"content": "䄛",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51288": {
"content": "㔷",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51289": {
"content": "㗱",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51290": {
"content": "㭾",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51291": {
"content": "㗾",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51292": {
"content": "㧜",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51293": {
"content": "㴫",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51294": {
"content": "䲒",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51295": {
"content": "㣒",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51296": {
"content": "䘀",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51297": {
"content": "㰸",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51298": {
"content": "䕊",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51299": {
"content": "㞓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51300": {
"content": "䚔",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51301": {
"content": "蠾",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51302": {
"content": "䭡",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51303": {
"content": "㵖",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51304": {
"content": "䵯",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51305": {
"content": "䋒",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51306": {
"content": "䐔",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51307": {
"content": "譈",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51308": {
"content": "䛮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51309": {
"content": "䀳",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51310": {
"content": "㬮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51311": {
"content": "䕼",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51312": {
"content": "䱏",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51313": {
"content": "䄮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51314": {
"content": "䠮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51315": {
"content": "㳮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51316": {
"content": "㖥",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51317": {
"content": "籅",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51318": {
"content": "攳",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51319": {
"content": "㹵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51320": {
"content": "䭮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51321": {
"content": "䘳",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51322": {
"content": "䛷",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51323": {
"content": "㾝",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51324": {
"content": "䭽",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51325": {
"content": "㗋",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51326": {
"content": "㩕",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51327": {
"content": "㯰",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51328": {
"content": "㶺",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51329": {
"content": "䞟",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51330": {
"content": "㫗",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51331": {
"content": "㕷",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51332": {
"content": "㧧",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51333": {
"content": "䤄",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51334": {
"content": "㛱",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51335": {
"content": "㰮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51336": {
"content": "㕍",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51337": {
"content": "䶵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51338": {
"content": "㝿",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51339": {
"content": "㱒",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51340": {
"content": "鞝",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51341": {
"content": "㦨",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51342": {
"content": "㨘",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51343": {
"content": "䞾",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51344": {
"content": "䆭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51345": {
"content": "䭃",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51346": {
"content": "㕇",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51347": {
"content": "騇",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51348": {
"content": "䰢",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51349": {
"content": "䦞",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51350": {
"content": "䯧",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51351": {
"content": "㲈",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51352": {
"content": "㸛",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51353": {
"content": "㩻",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51354": {
"content": "䒭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51355": {
"content": "䭋",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51356": {
"content": "䱰",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51357": {
"content": "㶦",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51358": {
"content": "䏺",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51359": {
"content": "㢜",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51360": {
"content": "䩐",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51361": {
"content": "㭛",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51362": {
"content": "䌟",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51363": {
"content": "㮇",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51364": {
"content": "㨾",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51365": {
"content": "㑃",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51366": {
"content": "㱱",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51367": {
"content": "䰣",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51368": {
"content": "䀆",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51369": {
"content": "䊵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51370": {
"content": "㹒",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51371": {
"content": "䁵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51372": {
"content": "䞨",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51373": {
"content": "饇",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51374": {
"content": "㹾",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51375": {
"content": "䣸",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51376": {
"content": "㝫",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51377": {
"content": "㵞",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51378": {
"content": "㵕",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51379": {
"content": "䯡",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51380": {
"content": "䁷",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51381": {
"content": "藀",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51382": {
"content": "㝙",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51383": {
"content": "㭂",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51384": {
"content": "䁓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51385": {
"content": "㲻",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51386": {
"content": "䊯",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51387": {
"content": "㰹",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51388": {
"content": "㥶",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51389": {
"content": "㙒",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51390": {
"content": "㤟",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51391": {
"content": "䦥",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51392": {
"content": "㭬",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51393": {
"content": "㑰",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51394": {
"content": "㤳",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51395": {
"content": "䵎",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51396": {
"content": "䄒",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51397": {
"content": "䰺",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51398": {
"content": "㬢",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51399": {
"content": "䈢",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51400": {
"content": "䏙",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51401": {
"content": "㤓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51402": {
"content": "㔁",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51403": {
"content": "㵻",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51404": {
"content": "㪹",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51405": {
"content": "㓟",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51406": {
"content": "㛻",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51407": {
"content": "䉵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51408": {
"content": "䵜",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51409": {
"content": "䄪",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51410": {
"content": "㭋",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51411": {
"content": "㑭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51412": {
"content": "䀨",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51413": {
"content": "嘺",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51414": {
"content": "㸀",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51415": {
"content": "䈟",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51416": {
"content": "㚲",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51417": {
"content": "㞜",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51418": {
"content": "藬",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51419": {
"content": "䕱",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51420": {
"content": "㗻",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51421": {
"content": "䩢",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51422": {
"content": "齴",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51423": {
"content": "䪐",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51424": {
"content": "㕃",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51425": {
"content": "䰝",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51426": {
"content": "䶡",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51427": {
"content": "䠊",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51428": {
"content": "㓋",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51429": {
"content": "躷",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51430": {
"content": "㓽",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51431": {
"content": "㽂",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51432": {
"content": "䝟",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51433": {
"content": "䧐",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51434": {
"content": "邭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51435": {
"content": "㢸",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51436": {
"content": "䅤",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51437": {
"content": "㿺",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51438": {
"content": "㣜",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51439": {
"content": "䮺",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51440": {
"content": "㗳",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51441": {
"content": "㾓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51442": {
"content": "嚝",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51443": {
"content": "㩙",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51444": {
"content": "䉔",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51445": {
"content": "䂾",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51446": {
"content": "贉",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51447": {
"content": "㾘",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51448": {
"content": "㧵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51449": {
"content": "䥚",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51450": {
"content": "䵦",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51451": {
"content": "䧬",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"51452": {
"content": "䃛",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
}
},
"bos_token": "[CLS]",
"clean_up_tokenization_spaces": true,
"cls_token": "[CLS]",
"do_basic_tokenize": true,
"do_lower_case": false,
"eos_token": "[EOS]",
"mask_token": "[MASK]",
"model_max_length": 1000000000000000019884624838656,
"never_split": null,
"pad_token": "[PAD]",
"sep_token": "[SEP]",
"strip_accents": null,
"tokenize_chinese_chars": true,
"tokenizer_class": "BertTokenizer",
"unk_token": "[UNK]",
"use_fast": true
}