Austin Davis
commited on
Commit
·
be2d11c
1
Parent(s):
7a8bcfa
Update tokenizer_config.json
Browse files- tokenizer_config.json +8 -1
tokenizer_config.json
CHANGED
|
@@ -1,8 +1,15 @@
|
|
| 1 |
{
|
| 2 |
"bos_token": "<s>",
|
| 3 |
"clean_up_tokenization_spaces": true,
|
|
|
|
| 4 |
"model_max_length": 1000000000000000019884624838656,
|
|
|
|
| 5 |
"pad_token": "<pad>",
|
| 6 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
"unk_token": "<pad>"
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"bos_token": "<s>",
|
| 3 |
"clean_up_tokenization_spaces": true,
|
| 4 |
+
"max_length": 512,
|
| 5 |
"model_max_length": 1000000000000000019884624838656,
|
| 6 |
+
"pad_to_multiple_of": null,
|
| 7 |
"pad_token": "<pad>",
|
| 8 |
+
"pad_token_type_id": 0,
|
| 9 |
+
"padding_side": "right",
|
| 10 |
+
"stride": 0,
|
| 11 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 12 |
+
"truncation_side": "right",
|
| 13 |
+
"truncation_strategy": "longest_first",
|
| 14 |
"unk_token": "<pad>"
|
| 15 |
}
|