tokenizer.json

{
  "version": "1.0",
  "truncation": null,
  "padding": null,
  "added_tokens": [
    {
      "id": 0,
      "content": "[PAD]",
      "single_word": false,
      "lstrip": false,
      "rstrip": false,
      "normalized": false,
      "special": true
    },
    {
      "id": 1,
      "content": "[UNK]",
      "single_word": false,
      "lstrip": false,
      "rstrip": false,
      "normalized": false,
      "special": true
    }
  ],
  "normalizer": {
    "type": "Lowercase"
  },
  "pre_tokenizer": {
    "type": "Split",
    "pattern": {
      "Regex": ".{4}"
    },
    "behavior": "Isolated",
    "invert": false
  },
  "post_processor": null,
  "decoder": {
    "type": "Fuse"
  },
  "model": {
    "type": "WordLevel",
    "vocab": {
      "[PAD]": 0,
      "[UNK]": 1,
      "aaaa": 2,
      "aaac": 3,
      "aaag": 4,
      "aaat": 5,
      "aaca": 6,
      "aacc": 7,
      "aacg": 8,
      "aact": 9,
      "aaga": 10,
      "aagc": 11,
      "aagg": 12,
      "aagt": 13,
      "aata": 14,
      "aatc": 15,
      "aatg": 16,
      "aatt": 17,
      "acaa": 18,
      "acac": 19,
      "acag": 20,
      "acat": 21,
      "acca": 22,
      "accc": 23,
      "accg": 24,
      "acct": 25,
      "acga": 26,
      "acgc": 27,
      "acgg": 28,
      "acgt": 29,
      "acta": 30,
      "actc": 31,
      "actg": 32,
      "actt": 33,
      "agaa": 34,
      "agac": 35,
      "agag": 36,
      "agat": 37,
      "agca": 38,
      "agcc": 39,
      "agcg": 40,
      "agct": 41,
      "agga": 42,
      "aggc": 43,
      "aggg": 44,
      "aggt": 45,
      "agta": 46,
      "agtc": 47,
      "agtg": 48,
      "agtt": 49,
      "ataa": 50,
      "atac": 51,
      "atag": 52,
      "atat": 53,
      "atca": 54,
      "atcc": 55,
      "atcg": 56,
      "atct": 57,
      "atga": 58,
      "atgc": 59,
      "atgg": 60,
      "atgt": 61,
      "atta": 62,
      "attc": 63,
      "attg": 64,
      "attt": 65,
      "caaa": 66,
      "caac": 67,
      "caag": 68,
      "caat": 69,
      "caca": 70,
      "cacc": 71,
      "cacg": 72,
      "cact": 73,
      "caga": 74,
      "cagc": 75,
      "cagg": 76,
      "cagt": 77,
      "cata": 78,
      "catc": 79,
      "catg": 80,
      "catt": 81,
      "ccaa": 82,
      "ccac": 83,
      "ccag": 84,
      "ccat": 85,
      "ccca": 86,
      "cccc": 87,
      "cccg": 88,
      "ccct": 89,
      "ccga": 90,
      "ccgc": 91,
      "ccgg": 92,
      "ccgt": 93,
      "ccta": 94,
      "cctc": 95,
      "cctg": 96,
      "cctt": 97,
      "cgaa": 98,
      "cgac": 99,
      "cgag": 100,
      "cgat": 101,
      "cgca": 102,
      "cgcc": 103,
      "cgcg": 104,
      "cgct": 105,
      "cgga": 106,
      "cggc": 107,
      "cggg": 108,
      "cggt": 109,
      "cgta": 110,
      "cgtc": 111,
      "cgtg": 112,
      "cgtt": 113,
      "ctaa": 114,
      "ctac": 115,
      "ctag": 116,
      "ctat": 117,
      "ctca": 118,
      "ctcc": 119,
      "ctcg": 120,
      "ctct": 121,
      "ctga": 122,
      "ctgc": 123,
      "ctgg": 124,
      "ctgt": 125,
      "ctta": 126,
      "cttc": 127,
      "cttg": 128,
      "cttt": 129,
      "gaaa": 130,
      "gaac": 131,
      "gaag": 132,
      "gaat": 133,
      "gaca": 134,
      "gacc": 135,
      "gacg": 136,
      "gact": 137,
      "gaga": 138,
      "gagc": 139,
      "gagg": 140,
      "gagt": 141,
      "gata": 142,
      "gatc": 143,
      "gatg": 144,
      "gatt": 145,
      "gcaa": 146,
      "gcac": 147,
      "gcag": 148,
      "gcat": 149,
      "gcca": 150,
      "gccc": 151,
      "gccg": 152,
      "gcct": 153,
      "gcga": 154,
      "gcgc": 155,
      "gcgg": 156,
      "gcgt": 157,
      "gcta": 158,
      "gctc": 159,
      "gctg": 160,
      "gctt": 161,
      "ggaa": 162,
      "ggac": 163,
      "ggag": 164,
      "ggat": 165,
      "ggca": 166,
      "ggcc": 167,
      "ggcg": 168,
      "ggct": 169,
      "ggga": 170,
      "gggc": 171,
      "gggg": 172,
      "gggt": 173,
      "ggta": 174,
      "ggtc": 175,
      "ggtg": 176,
      "ggtt": 177,
      "gtaa": 178,
      "gtac": 179,
      "gtag": 180,
      "gtat": 181,
      "gtca": 182,
      "gtcc": 183,
      "gtcg": 184,
      "gtct": 185,
      "gtga": 186,
      "gtgc": 187,
      "gtgg": 188,
      "gtgt": 189,
      "gtta": 190,
      "gttc": 191,
      "gttg": 192,
      "gttt": 193,
      "taaa": 194,
      "taac": 195,
      "taag": 196,
      "taat": 197,
      "taca": 198,
      "tacc": 199,
      "tacg": 200,
      "tact": 201,
      "taga": 202,
      "tagc": 203,
      "tagg": 204,
      "tagt": 205,
      "tata": 206,
      "tatc": 207,
      "tatg": 208,
      "tatt": 209,
      "tcaa": 210,
      "tcac": 211,
      "tcag": 212,
      "tcat": 213,
      "tcca": 214,
      "tccc": 215,
      "tccg": 216,
      "tcct": 217,
      "tcga": 218,
      "tcgc": 219,
      "tcgg": 220,
      "tcgt": 221,
      "tcta": 222,
      "tctc": 223,
      "tctg": 224,
      "tctt": 225,
      "tgaa": 226,
      "tgac": 227,
      "tgag": 228,
      "tgat": 229,
      "tgca": 230,
      "tgcc": 231,
      "tgcg": 232,
      "tgct": 233,
      "tgga": 234,
      "tggc": 235,
      "tggg": 236,
      "tggt": 237,
      "tgta": 238,
      "tgtc": 239,
      "tgtg": 240,
      "tgtt": 241,
      "ttaa": 242,
      "ttac": 243,
      "ttag": 244,
      "ttat": 245,
      "ttca": 246,
      "ttcc": 247,
      "ttcg": 248,
      "ttct": 249,
      "ttga": 250,
      "ttgc": 251,
      "ttgg": 252,
      "ttgt": 253,
      "ttta": 254,
      "tttc": 255,
      "tttg": 256,
      "tttt": 257
    },
    "unk_token": "[UNK]"
  }
}