End of training
Browse files- tokenizer.json +0 -0
- tokenizer_config.json +44 -0
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"apply_ocr": false,
|
| 4 |
+
"backend": "tokenizers",
|
| 5 |
+
"bos_token": "<s>",
|
| 6 |
+
"cls_token": "<s>",
|
| 7 |
+
"cls_token_box": [
|
| 8 |
+
0,
|
| 9 |
+
0,
|
| 10 |
+
0,
|
| 11 |
+
0
|
| 12 |
+
],
|
| 13 |
+
"eos_token": "</s>",
|
| 14 |
+
"errors": "replace",
|
| 15 |
+
"is_local": false,
|
| 16 |
+
"mask_token": "<mask>",
|
| 17 |
+
"max_length": 512,
|
| 18 |
+
"model_max_length": 512,
|
| 19 |
+
"only_label_first_subword": true,
|
| 20 |
+
"pad_to_multiple_of": null,
|
| 21 |
+
"pad_token": "<pad>",
|
| 22 |
+
"pad_token_box": [
|
| 23 |
+
0,
|
| 24 |
+
0,
|
| 25 |
+
0,
|
| 26 |
+
0
|
| 27 |
+
],
|
| 28 |
+
"pad_token_label": -100,
|
| 29 |
+
"pad_token_type_id": 0,
|
| 30 |
+
"padding_side": "right",
|
| 31 |
+
"sep_token": "</s>",
|
| 32 |
+
"sep_token_box": [
|
| 33 |
+
0,
|
| 34 |
+
0,
|
| 35 |
+
0,
|
| 36 |
+
0
|
| 37 |
+
],
|
| 38 |
+
"stride": 128,
|
| 39 |
+
"tokenizer_class": "TokenizersBackend",
|
| 40 |
+
"trim_offsets": true,
|
| 41 |
+
"truncation_side": "right",
|
| 42 |
+
"truncation_strategy": "longest_first",
|
| 43 |
+
"unk_token": "<unk>"
|
| 44 |
+
}
|