Commit
·
5666019
1
Parent(s):
bd45749
add tokenizer
Browse files- .gitignore +1 -0
- vocab.json +1 -1
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
checkpoint-*/
|
vocab.json
CHANGED
@@ -1 +1 @@
|
|
1 |
-
{"
|
|
|
1 |
+
{"1": 0, "m": 1, "u": 2, "w": 3, "0": 4, "a": 5, "h": 6, "v": 7, "e": 8, "r": 9, "4": 10, "y": 12, "n": 13, "c": 14, "2": 15, "p": 16, "k": 17, "g": 18, "8": 19, "9": 20, "3": 21, "6": 22, "b": 23, "l": 24, "i": 25, "o": 26, "d": 27, "5": 28, "q": 29, "z": 30, "s": 31, "t": 32, "x": 33, "7": 34, "f": 35, "j": 36, "|": 11, "[UNK]": 37, "[PAD]": 38}
|