donut-base-demo2 / tokenizer_config.json
cactusgame's picture
End of training
b7c0220 verified
raw
history blame
29.9 kB
{
"added_tokens_decoder": {
"0": {
"content": "<s>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"1": {
"content": "<pad>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"2": {
"content": "</s>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"3": {
"content": "<unk>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57521": {
"content": "<mask>",
"lstrip": true,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": true
},
"57522": {
"content": "<sep/>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57523": {
"content": "<s_iitcdip>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57524": {
"content": "<s_synthdog>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57525": {
"content": "递",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57526": {
"content": "润",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57527": {
"content": "询",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57528": {
"content": "脉",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57529": {
"content": "峄",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57530": {
"content": "伽",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57531": {
"content": "诸",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57532": {
"content": "膳",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57533": {
"content": "娟",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57534": {
"content": "训",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57535": {
"content": "皕",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57536": {
"content": "埠",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57537": {
"content": "韦",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57538": {
"content": "辐",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57539": {
"content": "昇",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57540": {
"content": "谱",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57541": {
"content": "玺",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57542": {
"content": "励",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57543": {
"content": "铝",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57544": {
"content": "渣",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57545": {
"content": "栋",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57546": {
"content": "昭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57547": {
"content": "卉",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57548": {
"content": "辉",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57549": {
"content": "湫",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57550": {
"content": "昊",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57551": {
"content": "估",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57552": {
"content": "悦",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57553": {
"content": "铭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57554": {
"content": "络",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57555": {
"content": "硕",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57556": {
"content": "绞",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57557": {
"content": "菱",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57558": {
"content": "锐",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57559": {
"content": "冶",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57560": {
"content": "皿",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57561": {
"content": "祜",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57562": {
"content": "铜",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57563": {
"content": "垣",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57564": {
"content": "蕾",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57565": {
"content": ")",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57566": {
"content": "竑",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57567": {
"content": "裕",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57568": {
"content": "磊",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57569": {
"content": "驰",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57570": {
"content": "怡",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57571": {
"content": "臣",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57572": {
"content": "涵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57573": {
"content": "贤",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57574": {
"content": "蕲",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57575": {
"content": "册",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57576": {
"content": "鹰",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57577": {
"content": "碧",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57578": {
"content": "沣",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57579": {
"content": "郓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57580": {
"content": "颢",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57581": {
"content": "伞",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57582": {
"content": "琪",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57583": {
"content": "徽",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57584": {
"content": "芯",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57585": {
"content": "泵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57586": {
"content": "雀",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57587": {
"content": "卓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57588": {
"content": "刃",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57589": {
"content": "愚",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57590": {
"content": "绍",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57591": {
"content": "鑫",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57592": {
"content": "培",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57593": {
"content": "灏",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57594": {
"content": "湘",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57595": {
"content": "洽",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57596": {
"content": "潢",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57597": {
"content": "殖",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57598": {
"content": "际",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57599": {
"content": "钰",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57600": {
"content": "橡",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57601": {
"content": "贷",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57602": {
"content": "腔",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57603": {
"content": "俞",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57604": {
"content": "遣",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57605": {
"content": "曹",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57606": {
"content": "析",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57607": {
"content": "佑",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57608": {
"content": "陂",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57609": {
"content": "缔",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57610": {
"content": "塘",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57611": {
"content": "帽",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57612": {
"content": "恬",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57613": {
"content": "葛",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57614": {
"content": "杉",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57615": {
"content": "勋",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57616": {
"content": "芦",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57617": {
"content": "蝶",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57618": {
"content": "瀚",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57619": {
"content": "衢",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57620": {
"content": "济",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57621": {
"content": "鸣",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57622": {
"content": "朴",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57623": {
"content": "针",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57624": {
"content": "筑",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57625": {
"content": "1",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57626": {
"content": "萍",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57627": {
"content": "巢",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57628": {
"content": "楷",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57629": {
"content": "旭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57630": {
"content": "堃",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57631": {
"content": "苗",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57632": {
"content": "翼",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57633": {
"content": "堤",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57634": {
"content": "渝",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57635": {
"content": "翎",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57636": {
"content": "跃",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57637": {
"content": "皓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57638": {
"content": "构",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57639": {
"content": "鸭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57640": {
"content": "铠",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57641": {
"content": "牡",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57642": {
"content": "蔚",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57643": {
"content": "饲",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57644": {
"content": "抚",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57645": {
"content": "潭",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57646": {
"content": "靖",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57647": {
"content": "鄄",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57648": {
"content": "煌",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57649": {
"content": "削",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57650": {
"content": "鼎",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57651": {
"content": "链",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57652": {
"content": "虹",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57653": {
"content": "粮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57654": {
"content": "铺",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57655": {
"content": "唯",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57656": {
"content": "岑",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57657": {
"content": "峡",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57658": {
"content": "芙",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57659": {
"content": "舟",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57660": {
"content": "铸",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57661": {
"content": "骞",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57662": {
"content": "疗",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57663": {
"content": "(",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57664": {
"content": "澜",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57665": {
"content": "盼",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57666": {
"content": "稻",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57667": {
"content": "弘",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57668": {
"content": "銮",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57669": {
"content": "拓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57670": {
"content": "垒",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57671": {
"content": "坻",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57672": {
"content": "验",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57673": {
"content": "匠",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57674": {
"content": "胶",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57675": {
"content": "滨",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57676": {
"content": "谊",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57677": {
"content": "迅",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57678": {
"content": "闼",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57679": {
"content": "疆",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57680": {
"content": "崴",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57681": {
"content": "粘",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57682": {
"content": "吕",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57683": {
"content": "凤",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57684": {
"content": "焓",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57685": {
"content": "喷",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57686": {
"content": "赣",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57687": {
"content": "啤",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57688": {
"content": "译",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57689": {
"content": "辽",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57690": {
"content": "陵",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
}
},
"additional_special_tokens": [
"<s_iitcdip>",
"<s_synthdog>"
],
"bos_token": "<s>",
"clean_up_tokenization_spaces": true,
"cls_token": "<s>",
"eos_token": "</s>",
"mask_token": "<mask>",
"model_max_length": 1000000000000000019884624838656,
"pad_token": "<pad>",
"processor_class": "DonutProcessor",
"sep_token": "</s>",
"sp_model_kwargs": {},
"tokenizer_class": "XLMRobertaTokenizer",
"unk_token": "<unk>"
}