Update tokenizer_config.json
Browse files- tokenizer_config.json +17 -1
tokenizer_config.json
CHANGED
@@ -1,5 +1,21 @@
|
|
1 |
{
|
2 |
"do_lower_case": true,
|
3 |
"model_type": "bert",
|
4 |
-
"vocab_size": 30522
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
}
|
|
|
1 |
{
|
2 |
"do_lower_case": true,
|
3 |
"model_type": "bert",
|
4 |
+
"vocab_size": 30522,
|
5 |
+
"special_tokens_map": {
|
6 |
+
"unk_token": "[UNK]",
|
7 |
+
"sep_token": "[SEP]",
|
8 |
+
"pad_token": "[PAD]",
|
9 |
+
"cls_token": "[CLS]",
|
10 |
+
"mask_token": "[MASK]",
|
11 |
+
"eos_token": "<|endoftext|>",
|
12 |
+
"bos_token": "<|startoftext|>",
|
13 |
+
"user_token": "<|user|>",
|
14 |
+
"assistant_token": "<|assistant|>"
|
15 |
+
},
|
16 |
+
"padding_side": "right",
|
17 |
+
"truncation_side": "right",
|
18 |
+
"max_length": 1024,
|
19 |
+
"use_fast": true,
|
20 |
+
"tokenizer_class": "BertTokenizer"
|
21 |
}
|