Update tokenizer files
Browse files- tokenizer.json +1 -10
- tokenizer_config.json +2 -9
tokenizer.json
CHANGED
|
@@ -254,15 +254,6 @@
|
|
| 254 |
"rstrip": false,
|
| 255 |
"normalized": true,
|
| 256 |
"special": false
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"id": 50280,
|
| 260 |
-
"content": "<padding>",
|
| 261 |
-
"single_word": false,
|
| 262 |
-
"lstrip": false,
|
| 263 |
-
"rstrip": false,
|
| 264 |
-
"normalized": false,
|
| 265 |
-
"special": true
|
| 266 |
}
|
| 267 |
],
|
| 268 |
"normalizer": {
|
|
@@ -250610,4 +250601,4 @@
|
|
| 250610 |
]
|
| 250611 |
]
|
| 250612 |
}
|
| 250613 |
-
}
|
|
|
|
| 254 |
"rstrip": false,
|
| 255 |
"normalized": true,
|
| 256 |
"special": false
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 257 |
}
|
| 258 |
],
|
| 259 |
"normalizer": {
|
|
|
|
| 250601 |
]
|
| 250602 |
]
|
| 250603 |
}
|
| 250604 |
+
}
|
tokenizer_config.json
CHANGED
|
@@ -226,14 +226,6 @@
|
|
| 226 |
"rstrip": false,
|
| 227 |
"single_word": false,
|
| 228 |
"special": false
|
| 229 |
-
},
|
| 230 |
-
"50280": {
|
| 231 |
-
"content": "<|padding|>",
|
| 232 |
-
"lstrip": false,
|
| 233 |
-
"normalized": false,
|
| 234 |
-
"rstrip": false,
|
| 235 |
-
"single_word": false,
|
| 236 |
-
"special": true
|
| 237 |
}
|
| 238 |
},
|
| 239 |
"bos_token": "<|endoftext|>",
|
|
@@ -242,7 +234,8 @@
|
|
| 242 |
"eos_token": "<|endoftext|>",
|
| 243 |
"extra_special_tokens": {},
|
| 244 |
"model_max_length": 1000000000000000019884624838656,
|
| 245 |
-
"pad_token": "
|
| 246 |
"tokenizer_class": "GPTNeoXTokenizer",
|
| 247 |
"unk_token": "<|endoftext|>"
|
| 248 |
}
|
|
|
|
|
|
| 226 |
"rstrip": false,
|
| 227 |
"single_word": false,
|
| 228 |
"special": false
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 229 |
}
|
| 230 |
},
|
| 231 |
"bos_token": "<|endoftext|>",
|
|
|
|
| 234 |
"eos_token": "<|endoftext|>",
|
| 235 |
"extra_special_tokens": {},
|
| 236 |
"model_max_length": 1000000000000000019884624838656,
|
| 237 |
+
"pad_token": "<|padding|>",
|
| 238 |
"tokenizer_class": "GPTNeoXTokenizer",
|
| 239 |
"unk_token": "<|endoftext|>"
|
| 240 |
}
|
| 241 |
+
|