Anna Szymańska
commited on
Commit
·
d198c8f
1
Parent(s):
5bcbf85
Update tokenizer single and pair
Browse files- tokenizer/tokenizer.json +5 -5
tokenizer/tokenizer.json
CHANGED
|
@@ -48,7 +48,7 @@
|
|
| 48 |
"single": [
|
| 49 |
{
|
| 50 |
"SpecialToken": {
|
| 51 |
-
"id": "
|
| 52 |
"type_id": 0
|
| 53 |
}
|
| 54 |
},
|
|
@@ -60,7 +60,7 @@
|
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"SpecialToken": {
|
| 63 |
-
"id": "
|
| 64 |
"type_id": 0
|
| 65 |
}
|
| 66 |
}
|
|
@@ -68,7 +68,7 @@
|
|
| 68 |
"pair": [
|
| 69 |
{
|
| 70 |
"SpecialToken": {
|
| 71 |
-
"id": "
|
| 72 |
"type_id": 0
|
| 73 |
}
|
| 74 |
},
|
|
@@ -80,7 +80,7 @@
|
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"SpecialToken": {
|
| 83 |
-
"id": "
|
| 84 |
"type_id": 0
|
| 85 |
}
|
| 86 |
},
|
|
@@ -92,7 +92,7 @@
|
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"SpecialToken": {
|
| 95 |
-
"id": "
|
| 96 |
"type_id": 1
|
| 97 |
}
|
| 98 |
}
|
|
|
|
| 48 |
"single": [
|
| 49 |
{
|
| 50 |
"SpecialToken": {
|
| 51 |
+
"id": "<|startoftext|>",
|
| 52 |
"type_id": 0
|
| 53 |
}
|
| 54 |
},
|
|
|
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"SpecialToken": {
|
| 63 |
+
"id": "<|endoftext|>",
|
| 64 |
"type_id": 0
|
| 65 |
}
|
| 66 |
}
|
|
|
|
| 68 |
"pair": [
|
| 69 |
{
|
| 70 |
"SpecialToken": {
|
| 71 |
+
"id": "<|startoftext|>",
|
| 72 |
"type_id": 0
|
| 73 |
}
|
| 74 |
},
|
|
|
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"SpecialToken": {
|
| 83 |
+
"id": "<|endoftext|>",
|
| 84 |
"type_id": 0
|
| 85 |
}
|
| 86 |
},
|
|
|
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"SpecialToken": {
|
| 95 |
+
"id": "<|endoftext|>",
|
| 96 |
"type_id": 1
|
| 97 |
}
|
| 98 |
}
|