Upload tokenizer
4cfb4cd verified | { |
| "__PAD__": 0, |
| "__UNK__": 1, |
| "|": 2, |
| "ँ": 3, |
| "ं": 4, |
| "ः": 5, |
| "अ": 6, |
| "आ": 7, |
| "इ": 8, |
| "ई": 9, |
| "उ": 10, |
| "ऊ": 11, |
| "ऋ": 12, |
| "ए": 13, |
| "ऐ": 14, |
| "ओ": 15, |
| "औ": 16, |
| "क": 17, |
| "ख": 18, |
| "ग": 19, |
| "घ": 20, |
| "ङ": 21, |
| "च": 22, |
| "छ": 23, |
| "ज": 24, |
| "झ": 25, |
| "ञ": 26, |
| "ट": 27, |
| "ठ": 28, |
| "ड": 29, |
| "ढ": 30, |
| "ण": 31, |
| "त": 32, |
| "थ": 33, |
| "द": 34, |
| "ध": 35, |
| "न": 36, |
| "प": 37, |
| "फ": 38, |
| "ब": 39, |
| "भ": 40, |
| "म": 41, |
| "य": 42, |
| "र": 43, |
| "ऱ": 44, |
| "ल": 45, |
| "व": 46, |
| "श": 47, |
| "ष": 48, |
| "स": 49, |
| "ह": 50, |
| "़": 51, |
| "ा": 52, |
| "ि": 53, |
| "ी": 54, |
| "ु": 55, |
| "ू": 56, |
| "ृ": 57, |
| "े": 58, |
| "ै": 59, |
| "ॉ": 60, |
| "ो": 61, |
| "ौ": 62, |
| "्": 63, |
| "०": 64, |
| "१": 65, |
| "२": 66, |
| "३": 67, |
| "४": 68, |
| "५": 69, |
| "६": 70, |
| "७": 71, |
| "८": 72, |
| "९": 73 |
| } |
|
|