RichardErkhov
commited on
Commit
•
bfdd60f
1
Parent(s):
7029003
uploaded model
Browse files- tokenizer_config.json +38 -0
tokenizer_config.json
ADDED
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_bos_token": true,
|
3 |
+
"add_eos_token": false,
|
4 |
+
"added_tokens_decoder": {
|
5 |
+
"0": {
|
6 |
+
"content": "<unk>",
|
7 |
+
"lstrip": false,
|
8 |
+
"normalized": false,
|
9 |
+
"rstrip": false,
|
10 |
+
"single_word": false,
|
11 |
+
"special": true
|
12 |
+
},
|
13 |
+
"71013": {
|
14 |
+
"content": "|ENDOFTEXT|",
|
15 |
+
"lstrip": false,
|
16 |
+
"normalized": false,
|
17 |
+
"rstrip": false,
|
18 |
+
"single_word": false,
|
19 |
+
"special": true
|
20 |
+
}
|
21 |
+
},
|
22 |
+
"additional_special_tokens": [
|
23 |
+
"<unk>",
|
24 |
+
"|ENDOFTEXT|"
|
25 |
+
],
|
26 |
+
"bos_token": "|ENDOFTEXT|",
|
27 |
+
"clean_up_tokenization_spaces": false,
|
28 |
+
"eos_token": "|ENDOFTEXT|",
|
29 |
+
"legacy": true,
|
30 |
+
"model_max_length": 1000000000000000019884624838656,
|
31 |
+
"pad_token": null,
|
32 |
+
"processor_class": "FuyuProcessor",
|
33 |
+
"sp_model_kwargs": {},
|
34 |
+
"spaces_between_special_tokens": false,
|
35 |
+
"tokenizer_class": "LlamaTokenizer",
|
36 |
+
"unk_token": "<unk>",
|
37 |
+
"use_default_system_prompt": true
|
38 |
+
}
|