update tokenizer
Browse files- tokenizer_config.json +4 -3
tokenizer_config.json
CHANGED
|
@@ -1,9 +1,10 @@
|
|
| 1 |
{
|
| 2 |
"add_prefix_space": false,
|
| 3 |
-
"
|
| 4 |
"tokenizer_class": "GPT2Tokenizer",
|
| 5 |
-
"
|
|
|
|
| 6 |
"eos_token": "<|endoftext|>",
|
| 7 |
"model_max_length": 1000000000000000019884624838656,
|
| 8 |
-
"
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"add_prefix_space": false,
|
| 3 |
+
"add_bos_token": false,
|
| 4 |
"tokenizer_class": "GPT2Tokenizer",
|
| 5 |
+
"unk_token": "<|endoftext|>",
|
| 6 |
+
"bos_token": "<|endoftext|>",
|
| 7 |
"eos_token": "<|endoftext|>",
|
| 8 |
"model_max_length": 1000000000000000019884624838656,
|
| 9 |
+
"clean_up_tokenization_spaces": true
|
| 10 |
}
|