Upload 4 files
Browse files- added_tokens.json +5 -4
- special_tokens_map.json +1 -0
- tokenizer_config.json +12 -3
added_tokens.json
CHANGED
@@ -99,8 +99,9 @@
|
|
99 |
"<extra_id_98>": 32001,
|
100 |
"<extra_id_99>": 32000,
|
101 |
"<extra_id_9>": 32090,
|
102 |
-
"
|
103 |
-
"
|
104 |
-
"
|
105 |
-
"
|
|
|
106 |
}
|
|
|
99 |
"<extra_id_98>": 32001,
|
100 |
"<extra_id_99>": 32000,
|
101 |
"<extra_id_9>": 32090,
|
102 |
+
"<s>": 32100,
|
103 |
+
"article:": 32101,
|
104 |
+
"category:": 32103,
|
105 |
+
"classifiction:": 32102,
|
106 |
+
"generation:": 32104
|
107 |
}
|
special_tokens_map.json
CHANGED
@@ -29,6 +29,7 @@
|
|
29 |
"single_word": false
|
30 |
}
|
31 |
],
|
|
|
32 |
"eos_token": {
|
33 |
"content": "</s>",
|
34 |
"lstrip": false,
|
|
|
29 |
"single_word": false
|
30 |
}
|
31 |
],
|
32 |
+
"bos_token": "<s>",
|
33 |
"eos_token": {
|
34 |
"content": "</s>",
|
35 |
"lstrip": false,
|
tokenizer_config.json
CHANGED
@@ -825,7 +825,7 @@
|
|
825 |
"special": true
|
826 |
},
|
827 |
"32100": {
|
828 |
-
"content": "
|
829 |
"lstrip": false,
|
830 |
"normalized": false,
|
831 |
"rstrip": false,
|
@@ -833,7 +833,7 @@
|
|
833 |
"special": true
|
834 |
},
|
835 |
"32101": {
|
836 |
-
"content": "
|
837 |
"lstrip": false,
|
838 |
"normalized": false,
|
839 |
"rstrip": false,
|
@@ -841,7 +841,7 @@
|
|
841 |
"special": true
|
842 |
},
|
843 |
"32102": {
|
844 |
-
"content": "
|
845 |
"lstrip": false,
|
846 |
"normalized": false,
|
847 |
"rstrip": false,
|
@@ -849,6 +849,14 @@
|
|
849 |
"special": true
|
850 |
},
|
851 |
"32103": {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
852 |
"content": "generation:",
|
853 |
"lstrip": false,
|
854 |
"normalized": false,
|
@@ -863,6 +871,7 @@
|
|
863 |
"category:",
|
864 |
"generation:"
|
865 |
],
|
|
|
866 |
"clean_up_tokenization_spaces": true,
|
867 |
"eos_token": "</s>",
|
868 |
"extra_ids": 100,
|
|
|
825 |
"special": true
|
826 |
},
|
827 |
"32100": {
|
828 |
+
"content": "<s>",
|
829 |
"lstrip": false,
|
830 |
"normalized": false,
|
831 |
"rstrip": false,
|
|
|
833 |
"special": true
|
834 |
},
|
835 |
"32101": {
|
836 |
+
"content": "article:",
|
837 |
"lstrip": false,
|
838 |
"normalized": false,
|
839 |
"rstrip": false,
|
|
|
841 |
"special": true
|
842 |
},
|
843 |
"32102": {
|
844 |
+
"content": "classifiction:",
|
845 |
"lstrip": false,
|
846 |
"normalized": false,
|
847 |
"rstrip": false,
|
|
|
849 |
"special": true
|
850 |
},
|
851 |
"32103": {
|
852 |
+
"content": "category:",
|
853 |
+
"lstrip": false,
|
854 |
+
"normalized": false,
|
855 |
+
"rstrip": false,
|
856 |
+
"single_word": false,
|
857 |
+
"special": true
|
858 |
+
},
|
859 |
+
"32104": {
|
860 |
"content": "generation:",
|
861 |
"lstrip": false,
|
862 |
"normalized": false,
|
|
|
871 |
"category:",
|
872 |
"generation:"
|
873 |
],
|
874 |
+
"bos_token": "<s>",
|
875 |
"clean_up_tokenization_spaces": true,
|
876 |
"eos_token": "</s>",
|
877 |
"extra_ids": 100,
|