add: tokenizer config
Browse files- merges.txt +0 -0
- special_tokens.json +5 -0
- vocab.json +0 -0
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
special_tokens.json
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<|startoftext|>": 50256,
|
| 3 |
+
"<|endoftext|>": 50257,
|
| 4 |
+
"<|mdm_mask|>": 50258,
|
| 5 |
+
}
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|