Commit
·
1233b5d
1
Parent(s):
97c7c17
Add tokenizer
Browse files- special_tokens_map.json +9 -0
- tokenizer.json +0 -0
- tokenizer_config.json +7 -0
special_tokens_map.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|endoftext|>",
|
| 4 |
+
"<fim-prefix>",
|
| 5 |
+
"<fim-middle>",
|
| 6 |
+
"<fim-suffix>",
|
| 7 |
+
"<fim-pad>"
|
| 8 |
+
]
|
| 9 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"errors": "replace",
|
| 3 |
+
"tokenizer_class": "GPT2TokenizerFast",
|
| 4 |
+
"bos_token": "<|endoftext|>",
|
| 5 |
+
"eos_token": "<|endoftext|>",
|
| 6 |
+
"model_max_length": 2048
|
| 7 |
+
}
|