cbdb's picture
Upload tokenizer files
e02f79f
raw
history blame
468 Bytes
{"<unk>": 0, "<pad>": 1, "<s>": 2, "</s>": 3, "i": 4, "\u014f": 5, "ng": 6, "n": 7, "u": 8, "o": 9, "Y": 10, "K": 11, "Ch": 12, "a": 13, "S": 14, "m": 15, "y": 16, "g": 17, "k": 18, "an": 19, "e": 20, "h": 21, "P": 22, "'": 23, "w": 24, "s": 25, "H": 26, "j": 27, "l": 28, "M": 29, "\u016d": 30, "T": 31, "I": 32, "N": 33, "c": 34, "p": 35, "b": 36, "r": 37, "\u016c": 38, "A": 39, "W": 40, "d": 41, "O": 42, "U": 43, "t": 44, "R": 45, "\u014e": 46, "nan": 47, "D": 0}