tokenizer.json

{
  "vocab_size": 32000,
  "max_length": 128,
  "char_to_id": {
    "p": 4,
    "e": 5,
    "t": 6,
    "r": 7,
    " ": 8,
    "a": 9,
    "n": 10,
    "d": 11,
    "w": 12,
    "f": 13,
    "u": 14,
    "c": 15,
    "i": 16,
    "j": 17,
    "q": 18,
    "l": 19,
    "y": 20,
    "o": 21,
    "b": 22,
    "s": 23,
    "k": 24,
    "v": 25,
    "m": 26,
    "h": 27,
    "g": 28,
    "(": 29,
    ")": 30,
    "\u00e9": 31,
    "z": 32,
    ":": 33,
    "x": 34,
    "'": 35,
    "?": 36,
    "-": 37,
    "!": 38,
    "3": 39,
    ",": 40,
    "/": 41,
    ".": 42,
    "#": 43,
    "9": 44,
    "4": 45,
    "2": 46,
    "6": 47,
    "@": 48,
    "\u2013": 49,
    "1": 50,
    "0": 51,
    "\u00a0": 52,
    "\u00a3": 53,
    "8": 54,
    "7": 55,
    "+": 56,
    "5": 57,
    "&": 58,
    "|": 59,
    "\u2019": 60,
    "\ud83d\udc8c": 61,
    "<": 62,
    "\ud83d\udcde": 63,
    "%": 64,
    "\ud83d\udce3": 65,
    "\u2014": 66,
    "*": 67,
    "_": 68,
    "[": 69,
    "]": 70,
    "\u2122": 71,
    "\u26a1": 72,
    "\ufe0f": 73,
    "$": 74,
    "\u2600": 75,
    "\ud83d\udd25": 76,
    "\ud83d\udd0c": 77,
    "\ud83c\udf4e": 78,
    "\ud83d\ude80": 79,
    "\ud83e\udd14": 80,
    "\ud83d\udc4b": 81,
    "\ud83d\udc40": 82,
    "\ud83d\udea8": 83,
    "\u23f0": 84,
    "\ud83d\udfe2": 85,
    "\ud83c\udfe0": 86,
    "\ud83e\uddf3": 87,
    "\ud83d\udd52": 88,
    "\ud83d\udd17": 89,
    "\ud83d\ude97": 90,
    "\"": 91,
    "\ud83c\udf81": 92,
    "\ud83d\udcf8": 93,
    "\u203c": 94,
    "=": 95,
    "\u23e9": 96,
    "\ud83c\udfba": 97,
    "\ud83c\udfac": 98,
    "\ud83d\udc64": 99,
    ">": 100,
    "\u26bd": 101,
    "\ud83d\ude26": 102,
    "\ud83c\udf82": 103,
    "~": 104,
    "\ud83d\udd06": 105,
    "\ud835\udce1": 106,
    "\ud835\udcee": 107,
    "\ud835\udcf0": 108,
    "\ud835\udcf2": 109,
    "\ud835\udcf6": 110,
    "\ud835\udcd0": 111,
    "\ud835\udce4": 112,
    "\ud835\udcda": 113,
    "\ud83d\udcfa": 114,
    "\ud83c\udf31": 115,
    "\u201c": 116,
    "\u201d": 117,
    "\u2606": 118,
    "\ud83d\uddd3": 119,
    "\u23f3": 120,
    "\ud83d\ude2d": 121,
    "`": 122,
    ";": 123,
    "\u2744": 124,
    "\ud83e\udd16": 125,
    "\ud83d\udcc4": 126,
    "\ud83d\udca1": 127,
    "\ud83c\udf89": 128,
    "\ud83d\ude0a": 129,
    "\ud83c\udfc6": 130,
    "\u231b": 131,
    "\ud83e\udd2f": 132,
    "\u2728": 133,
    "\u2022": 134,
    "\ud83d\udcac": 135,
    "\u2026": 136
  },
  "next_id": 137
}