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 |
+
{"w": 0, "u": 1, "o": 2, "k": 3, "e": 4, "d": 5, "i": 6, "p": 7, "x": 8, "b": 9, "v": 10, "f": 11, "l": 12, "r": 13, "z": 14, "c": 15, "h": 16, "n": 17, "y": 18, "a": 19, "t": 20, "q": 22, "m": 23, "'": 24, "g": 25, "j": 26, "s": 27, "|": 21, "[UNK]": 28, "[PAD]": 29}
|