madeinglasgow
commited on
Commit
•
27555c5
1
Parent(s):
dc758dc
Upload tokenizer
Browse files- special_tokens_map.json +1 -1
- tokenizer.json +27 -0
special_tokens_map.json
CHANGED
@@ -13,7 +13,7 @@
|
|
13 |
"rstrip": false,
|
14 |
"single_word": false
|
15 |
},
|
16 |
-
"pad_token": "
|
17 |
"unk_token": {
|
18 |
"content": "<unk>",
|
19 |
"lstrip": false,
|
|
|
13 |
"rstrip": false,
|
14 |
"single_word": false
|
15 |
},
|
16 |
+
"pad_token": "▁***",
|
17 |
"unk_token": {
|
18 |
"content": "<unk>",
|
19 |
"lstrip": false,
|
tokenizer.json
CHANGED
@@ -71,6 +71,12 @@
|
|
71 |
"id": "A",
|
72 |
"type_id": 0
|
73 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
}
|
75 |
],
|
76 |
"pair": [
|
@@ -86,6 +92,12 @@
|
|
86 |
"type_id": 0
|
87 |
}
|
88 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
89 |
{
|
90 |
"SpecialToken": {
|
91 |
"id": "<s>",
|
@@ -97,9 +109,24 @@
|
|
97 |
"id": "B",
|
98 |
"type_id": 1
|
99 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
100 |
}
|
101 |
],
|
102 |
"special_tokens": {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
103 |
"<s>": {
|
104 |
"id": "<s>",
|
105 |
"ids": [
|
|
|
71 |
"id": "A",
|
72 |
"type_id": 0
|
73 |
}
|
74 |
+
},
|
75 |
+
{
|
76 |
+
"SpecialToken": {
|
77 |
+
"id": "</s>",
|
78 |
+
"type_id": 0
|
79 |
+
}
|
80 |
}
|
81 |
],
|
82 |
"pair": [
|
|
|
92 |
"type_id": 0
|
93 |
}
|
94 |
},
|
95 |
+
{
|
96 |
+
"SpecialToken": {
|
97 |
+
"id": "</s>",
|
98 |
+
"type_id": 0
|
99 |
+
}
|
100 |
+
},
|
101 |
{
|
102 |
"SpecialToken": {
|
103 |
"id": "<s>",
|
|
|
109 |
"id": "B",
|
110 |
"type_id": 1
|
111 |
}
|
112 |
+
},
|
113 |
+
{
|
114 |
+
"SpecialToken": {
|
115 |
+
"id": "</s>",
|
116 |
+
"type_id": 1
|
117 |
+
}
|
118 |
}
|
119 |
],
|
120 |
"special_tokens": {
|
121 |
+
"</s>": {
|
122 |
+
"id": "</s>",
|
123 |
+
"ids": [
|
124 |
+
2
|
125 |
+
],
|
126 |
+
"tokens": [
|
127 |
+
"</s>"
|
128 |
+
]
|
129 |
+
},
|
130 |
"<s>": {
|
131 |
"id": "<s>",
|
132 |
"ids": [
|