alvynabranches
commited on
Commit
•
a2ae85c
1
Parent(s):
51b5092
Upload tokenizer
Browse files- special_tokens_map.json +10 -0
- tokenizer_config.json +22 -0
special_tokens_map.json
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"eos_token": {
|
3 |
+
"content": "<|endoftext|>",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": true,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"pad_token": "<|endoftext|>"
|
10 |
+
}
|
tokenizer_config.json
ADDED
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_eos_token": false,
|
3 |
+
"add_special_tokens": true,
|
4 |
+
"auto_map": {
|
5 |
+
"AutoTokenizer": [
|
6 |
+
"Salesforce/xgen-7b-8k-base--tokenization_xgen.XgenTokenizer",
|
7 |
+
null
|
8 |
+
]
|
9 |
+
},
|
10 |
+
"clean_up_tokenization_spaces": true,
|
11 |
+
"eos_token": {
|
12 |
+
"__type": "AddedToken",
|
13 |
+
"content": "<|endoftext|>",
|
14 |
+
"lstrip": false,
|
15 |
+
"normalized": true,
|
16 |
+
"rstrip": false,
|
17 |
+
"single_word": false
|
18 |
+
},
|
19 |
+
"model_max_length": 8192,
|
20 |
+
"pad_token": null,
|
21 |
+
"tokenizer_class": "XgenTokenizer"
|
22 |
+
}
|