Xenova's picture
Xenova HF staff
Upload tokenizer.json with huggingface_hub
d112d51
raw
history blame
No virus
3.14 kB
{
"version": "1.0",
"truncation": null,
"padding": null,
"added_tokens": [
{
"id": 60,
"content": "<unk>",
"single_word": false,
"lstrip": false,
"rstrip": false,
"normalized": false,
"special": true
}
],
"normalizer": {
"type": "Sequence",
"normalizers": [
{
"type": "Lowercase"
},
{
"type": "Replace",
"pattern": {
"Regex": "[^|\u0abe\u0ac7\u0aa4\u0aa8\u0ab0\u0acd\u0aae\u0ac0\u0a95\u0a82\u0acb\u0aaa\u0ab5\u0ac1\u0ab8\u0aaf\u0ab9\u0a9c\u0aa5\u0abf\u0ab6\u0aa3\u0a9b\u0ab2\u0aa6\u0a86\u0a93\u0a85\u0a8f\u0a88\u0aac\u0a97\u0a96\u0aa7\u0aad\u0a9f\u0ab3\u0ac2\u0a9a\u0aa1\u0ab7\u0a89\u0aab\u0a98\u0aa0\u0ac3'\u0a8a\u0a87\u0a9e\u0a9d\u0aa2\u0ac8\u0a83\u0acc\\-\u0a8b\u0a90 ]"
},
"content": ""
},
{
"type": "Strip",
"strip_left": true,
"strip_right": true
},
{
"type": "Replace",
"pattern": {
"Regex": "(?=.)|(?<!^)$"
},
"content": "|"
}
]
},
"pre_tokenizer": {
"type": "Split",
"pattern": {
"Regex": ""
},
"behavior": "Isolated",
"invert": false
},
"post_processor": null,
"decoder": null,
"model": {
"vocab": {
"|": 0,
"\u0abe": 1,
"\u0ac7": 2,
"\u0aa4": 3,
"\u0aa8": 4,
"\u0ab0": 5,
"\u0acd": 6,
"\u0aae": 7,
"\u0ac0": 8,
"\u0a95": 9,
"\u0a82": 10,
"\u0acb": 11,
"\u0aaa": 12,
"\u0ab5": 13,
"\u0ac1": 14,
"\u0ab8": 15,
"\u0aaf": 16,
"\u0ab9": 17,
"\u0a9c": 18,
"\u0aa5": 19,
"\u0abf": 20,
"\u0ab6": 21,
"\u0aa3": 22,
"\u0a9b": 23,
"\u0ab2": 24,
"\u0aa6": 25,
"\u0a86": 26,
"\u0a93": 27,
"\u0a85": 28,
"\u0a8f": 29,
"\u0a88": 30,
"\u0aac": 31,
"\u0a97": 32,
"\u0a96": 33,
"\u0aa7": 34,
"\u0aad": 35,
"\u0a9f": 36,
"\u0ab3": 37,
"\u0ac2": 38,
"\u0a9a": 39,
"\u0aa1": 40,
"\u0ab7": 41,
"\u0a89": 42,
"\u0aab": 43,
"\u0a98": 44,
"\u0aa0": 45,
"\u0ac3": 46,
"'": 47,
"\u0a8a": 48,
"\u0a87": 49,
"\u0a9e": 50,
"\u0a9d": 51,
"\u0aa2": 52,
"\u0ac8": 53,
"\u0a83": 54,
"\u0acc": 55,
"-": 56,
"\u0a8b": 57,
"\u0a90": 58,
" ": 59,
"<unk>": 60
}
}
}