add tokenizer

Browse files

Files changed (1) hide show

vocab.json +1 -1

vocab.json CHANGED Viewed

@@ -1 +1 @@

- {"ت": 0, "ذ": 1, "ق": 2, "ف": 3, "ئ": 4, "ٹ": 5, "ظ": 6, "ڈ": 7, "ک": 8, "ج": 9, "ا": 10, "ص": 11, "ھ": 12, "ہ": 13, "م": 14, "ر": 15, "ح": 16, "چ": 17, "ی": 18, "ء": 19, "گ": 20, "ے": 21, "آ": 22, "و": 23, "ڑ": 24, "ں": 25, "ض": 26, "پ": 27, "ل": 28, " ": 29, "ب": 30, "خ": 31, "س": 32, "ن": 33, "ؤ": 34, "ث": 35, "ط": 36, "ژ": 37, "غ": 38, "ش": 39, "ز": 40, "د": 41, "ع": 42, "[UNK]": 43, "[PAD]": 44}

+ {"ذ": 0, "ز": 1, "ع": 2, "ے": 3, "ش": 4, "ث": 5, "ٹ": 6, "ل": 7, "ب": 8, "ں": 9, "ق": 10, "ف": 11, "ء": 12, "ط": 13, "س": 14, "م": 15, "چ": 16, "ظ": 17, "ت": 18, "ر": 19, "ن": 20, "ھ": 21, "پ": 22, "ج": 23, "ڑ": 24, "ص": 25, "ی": 26, "د": 27, "ژ": 28, "آ": 29, "ح": 30, "ئ": 31, "گ": 32, "ؤ": 33, "ا": 34, "و": 35, "ڈ": 36, "ض": 37, "خ": 38, "غ": 39, "ک": 40, " ": 41, "ہ": 42, "[UNK]": 43, "[PAD]": 44}