Commit
•
c4b7d74
1
Parent(s):
37b7cce
Upload tokenizer
Browse files- added_tokens.json +24 -0
- tokenizer.model +3 -0
added_tokens.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<pad>": 32018,
|
| 3 |
+
"<|Assistant|>": 32020,
|
| 4 |
+
"<|EOT|>": 32021,
|
| 5 |
+
"<|User|>": 32019,
|
| 6 |
+
"<｜begin▁of▁sentence｜>": 32013,
|
| 7 |
+
"<｜end▁of▁sentence｜>": 32014,
|
| 8 |
+
"<｜fim▁begin｜>": 32016,
|
| 9 |
+
"<｜fim▁end｜>": 32017,
|
| 10 |
+
"<｜fim▁hole｜>": 32015,
|
| 11 |
+
"À": 32004,
|
| 12 |
+
"Á": 32002,
|
| 13 |
+
"õ": 32000,
|
| 14 |
+
"ö": 32011,
|
| 15 |
+
"÷": 32001,
|
| 16 |
+
"ø": 32006,
|
| 17 |
+
"ù": 32010,
|
| 18 |
+
"ú": 32007,
|
| 19 |
+
"û": 32012,
|
| 20 |
+
"ü": 32009,
|
| 21 |
+
"ý": 32003,
|
| 22 |
+
"þ": 32008,
|
| 23 |
+
"ÿ": 32005
|
| 24 |
+
}
|
tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45ccb9c8b6b561889acea59191d66986d314e7cbd6a78abc6e49b139ca91c1e6
|
| 3 |
+
size 500058
|