ylacombe's picture
ylacombe HF staff
Upload tokenizer.json with huggingface_hub (#2)
4afdde5
raw
history blame contribute delete
No virus
3.49 kB
{
"version": "1.0",
"truncation": null,
"padding": null,
"added_tokens": [
{
"id": 73,
"content": "<unk>",
"single_word": false,
"lstrip": false,
"rstrip": false,
"normalized": false,
"special": true
}
],
"normalizer": {
"type": "Sequence",
"normalizers": [
{
"type": "Lowercase"
},
{
"type": "Replace",
"pattern": {
"Regex": "[^\u0919\u0939\u0906\u092c\u0922\u09162\u0902\u0924\u091f_\u0938\u090a1\u0915\u0914\u0933\u091a\u0913\\-\u0908\u0936\u091c\u0943\u0917\u091d\u0288\u0910\u0926 \u0921\u090f\u0948\u094c97\u0920\u0931\u0935\u094b\u090b04\u092f6\u0903\u0932\u0927\u092a\u091e\u092b\u0909\u0928\u0937\u092e\u091b\u0947\u093f\u0923'\u0905\u0907\u093e\u0941\u0942\u0287\u0934\u0918\u094d\u0930\u092d\u0940\u0925]"
},
"content": ""
},
{
"type": "Strip",
"strip_left": true,
"strip_right": true
},
{
"type": "Replace",
"pattern": {
"Regex": "(?=.)|(?<!^)$"
},
"content": "\u0919"
}
]
},
"pre_tokenizer": {
"type": "Split",
"pattern": {
"Regex": ""
},
"behavior": "Isolated",
"invert": false
},
"post_processor": null,
"decoder": null,
"model": {
"vocab": {
"\u0919": 0,
"\u0939": 1,
"\u0906": 2,
"\u092c": 3,
"\u0922": 4,
"\u0916": 5,
"2": 6,
"\u0902": 7,
"\u0924": 8,
"\u091f": 9,
"_": 10,
"\u0938": 11,
"\u090a": 12,
"1": 13,
"\u0915": 14,
"\u0914": 15,
"\u0933": 16,
"\u091a": 17,
"\u0913": 18,
"-": 19,
"\u0908": 20,
"\u0936": 21,
"\u091c": 22,
"\u0943": 23,
"\u0917": 24,
"\u091d": 25,
"\u0288": 26,
"\u0910": 27,
"\u0926": 28,
" ": 29,
"\u0921": 30,
"\u090f": 31,
"\u0948": 32,
"\u094c": 33,
"9": 34,
"7": 35,
"\u0920": 36,
"\u0931": 37,
"\u0935": 38,
"\u094b": 39,
"\u090b": 40,
"0": 41,
"4": 42,
"\u092f": 43,
"6": 44,
"\u0903": 45,
"\u0932": 46,
"\u0927": 47,
"\u092a": 48,
"\u091e": 49,
"\u092b": 50,
"\u0909": 51,
"\u0928": 52,
"\u0937": 53,
"\u092e": 54,
"\u091b": 55,
"\u0947": 56,
"\u093f": 57,
"\u0923": 58,
"'": 59,
"\u0905": 60,
"\u0907": 61,
"\u093e": 62,
"\u0941": 63,
"\u0942": 64,
"\u0287": 65,
"\u0934": 66,
"\u0918": 67,
"\u094d": 68,
"\u0930": 69,
"\u092d": 70,
"\u0940": 71,
"\u0925": 72,
"<unk>": 73
}
}
}