mms-tts-hin / tokenizer.json
Xenova's picture
Xenova HF staff
Upload tokenizer.json
08f491e
raw
history blame
No virus
3.46 kB
{
"version": "1.0",
"truncation": null,
"padding": null,
"added_tokens": [
{
"id": 72,
"content": "<unk>",
"single_word": false,
"lstrip": false,
"rstrip": false,
"normalized": false,
"special": true
}
],
"normalizer": {
"type": "Sequence",
"normalizers": [
{
"type": "Lowercase"
},
{
"type": "Replace",
"pattern": {
"Regex": "[^\u092b41\\-\u0905\u09078 0\u091b\u0928\u090f\u0910\u093c\u0937\u093f\u0901\u091a2\u0911\u0925\u092d\u0940\u200d\u09453\u091e\u0948\u0941\u0920\u0902\u0949\u0909_\u0908\u0903\u0939\u0927\u0932\u0930\u0938\u092c\u0916\u0923'`\u0935\u0918\u092a\u0917\u0922\u092f\u0947\u094d\u093e\u0906\u0921\u091c\u091d\u0936\u0914\u094b\u0926\u0943\u094c\u090a\u0942\u0913\u091f\u0924\u0915\u092e]"
},
"content": ""
},
{
"type": "Strip",
"strip_left": true,
"strip_right": true
},
{
"type": "Replace",
"pattern": {
"Regex": "(?=.)|(?<!^)$"
},
"content": "\u092b"
}
]
},
"pre_tokenizer": {
"type": "Split",
"pattern": {
"Regex": ""
},
"behavior": "Isolated",
"invert": false
},
"post_processor": null,
"decoder": null,
"model": {
"vocab": {
"\u092b": 0,
"4": 1,
"1": 2,
"-": 3,
"\u0905": 4,
"\u0907": 5,
"8": 6,
" ": 7,
"0": 8,
"\u091b": 9,
"\u0928": 10,
"\u090f": 11,
"\u0910": 12,
"\u093c": 13,
"\u0937": 14,
"\u093f": 15,
"\u0901": 16,
"\u091a": 17,
"2": 18,
"\u0911": 19,
"\u0925": 20,
"\u092d": 21,
"\u0940": 22,
"\u200d": 23,
"\u0945": 24,
"3": 25,
"\u091e": 26,
"\u0948": 27,
"\u0941": 28,
"\u0920": 29,
"\u0902": 30,
"\u0949": 31,
"\u0909": 32,
"_": 33,
"\u0908": 34,
"\u0903": 35,
"\u0939": 36,
"\u0927": 37,
"\u0932": 38,
"\u0930": 39,
"\u0938": 40,
"\u092c": 41,
"\u0916": 42,
"\u0923": 43,
"'": 44,
"`": 45,
"\u0935": 46,
"\u0918": 47,
"\u092a": 48,
"\u0917": 49,
"\u0922": 50,
"\u092f": 51,
"\u0947": 52,
"\u094d": 53,
"\u093e": 54,
"\u0906": 55,
"\u0921": 56,
"\u091c": 57,
"\u091d": 58,
"\u0936": 59,
"\u0914": 60,
"\u094b": 61,
"\u0926": 62,
"\u0943": 63,
"\u094c": 64,
"\u090a": 65,
"\u0942": 66,
"\u0913": 67,
"\u091f": 68,
"\u0924": 69,
"\u0915": 70,
"\u092e": 71,
"<unk>": 72
}
}
}