Upload folder using huggingface_hub
Browse files- tokenizer.json +1 -93
tokenizer.json
CHANGED
|
@@ -159,98 +159,6 @@
|
|
| 159 |
}
|
| 160 |
]
|
| 161 |
},
|
| 162 |
-
"post_processor": {
|
| 163 |
-
"type": "Sequence",
|
| 164 |
-
"processors": [
|
| 165 |
-
{
|
| 166 |
-
"type": "ByteLevel",
|
| 167 |
-
"add_prefix_space": true,
|
| 168 |
-
"trim_offsets": false,
|
| 169 |
-
"use_regex": true
|
| 170 |
-
},
|
| 171 |
-
{
|
| 172 |
-
"type": "TemplateProcessing",
|
| 173 |
-
"single": [
|
| 174 |
-
{
|
| 175 |
-
"SpecialToken": {
|
| 176 |
-
"id": "<|begin_of_text|>",
|
| 177 |
-
"type_id": 0
|
| 178 |
-
}
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"Sequence": {
|
| 182 |
-
"id": "A",
|
| 183 |
-
"type_id": 0
|
| 184 |
-
}
|
| 185 |
-
}
|
| 186 |
-
],
|
| 187 |
-
"pair": [
|
| 188 |
-
{
|
| 189 |
-
"SpecialToken": {
|
| 190 |
-
"id": "<|begin_of_text|>",
|
| 191 |
-
"type_id": 0
|
| 192 |
-
}
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"Sequence": {
|
| 196 |
-
"id": "A",
|
| 197 |
-
"type_id": 0
|
| 198 |
-
}
|
| 199 |
-
},
|
| 200 |
-
{
|
| 201 |
-
"SpecialToken": {
|
| 202 |
-
"id": "<|begin_of_text|>",
|
| 203 |
-
"type_id": 0
|
| 204 |
-
}
|
| 205 |
-
},
|
| 206 |
-
{
|
| 207 |
-
"Sequence": {
|
| 208 |
-
"id": "B",
|
| 209 |
-
"type_id": 0
|
| 210 |
-
}
|
| 211 |
-
}
|
| 212 |
-
],
|
| 213 |
-
"special_tokens": {
|
| 214 |
-
"<|begin_of_text|>": {
|
| 215 |
-
"id": "<|begin_of_text|>",
|
| 216 |
-
"ids": [
|
| 217 |
-
1
|
| 218 |
-
],
|
| 219 |
-
"tokens": [
|
| 220 |
-
"<|begin_of_text|>"
|
| 221 |
-
]
|
| 222 |
-
},
|
| 223 |
-
"<|end_of_text|>": {
|
| 224 |
-
"id": "<|end_of_text|>",
|
| 225 |
-
"ids": [
|
| 226 |
-
2
|
| 227 |
-
],
|
| 228 |
-
"tokens": [
|
| 229 |
-
"<|end_of_text|>"
|
| 230 |
-
]
|
| 231 |
-
},
|
| 232 |
-
"[PAD]": {
|
| 233 |
-
"id": "[PAD]",
|
| 234 |
-
"ids": [
|
| 235 |
-
3
|
| 236 |
-
],
|
| 237 |
-
"tokens": [
|
| 238 |
-
"[PAD]"
|
| 239 |
-
]
|
| 240 |
-
},
|
| 241 |
-
"[UNK]": {
|
| 242 |
-
"id": "[UNK]",
|
| 243 |
-
"ids": [
|
| 244 |
-
0
|
| 245 |
-
],
|
| 246 |
-
"tokens": [
|
| 247 |
-
"[UNK]"
|
| 248 |
-
]
|
| 249 |
-
}
|
| 250 |
-
}
|
| 251 |
-
}
|
| 252 |
-
]
|
| 253 |
-
},
|
| 254 |
"decoder": {
|
| 255 |
"type": "ByteLevel",
|
| 256 |
"add_prefix_space": true,
|
|
@@ -131064,4 +130972,4 @@
|
|
| 131064 |
"ĠÏĮ ÏĦαν"
|
| 131065 |
]
|
| 131066 |
}
|
| 131067 |
-
}
|
|
|
|
| 159 |
}
|
| 160 |
]
|
| 161 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 162 |
"decoder": {
|
| 163 |
"type": "ByteLevel",
|
| 164 |
"add_prefix_space": true,
|
|
|
|
| 130972 |
"ĠÏĮ ÏĦαν"
|
| 130973 |
]
|
| 130974 |
}
|
| 130975 |
+
}
|