Update tokenizer_config.json
Browse files- tokenizer_config.json +3 -8
tokenizer_config.json
CHANGED
|
@@ -113,20 +113,15 @@
|
|
| 113 |
"special": true
|
| 114 |
}
|
| 115 |
},
|
|
|
|
| 116 |
"clean_up_tokenization_spaces": false,
|
| 117 |
"eos_token": "<|endoftext|>",
|
| 118 |
-
"
|
| 119 |
"model_max_length": 131072,
|
| 120 |
-
"
|
| 121 |
-
"pad_token": "<unk>",
|
| 122 |
-
"pad_token_type_id": 0,
|
| 123 |
"padding_side": "left",
|
| 124 |
"sp_model_kwargs": {},
|
| 125 |
-
"stride": 0,
|
| 126 |
-
"legacy": false,
|
| 127 |
"tokenizer_class": "LlamaTokenizer",
|
| 128 |
-
"truncation_side": "right",
|
| 129 |
-
"truncation_strategy": "longest_first",
|
| 130 |
"unk_token": "<unk>",
|
| 131 |
"use_default_system_prompt": false
|
| 132 |
}
|
|
|
|
| 113 |
"special": true
|
| 114 |
}
|
| 115 |
},
|
| 116 |
+
"bos_token": "<s>",
|
| 117 |
"clean_up_tokenization_spaces": false,
|
| 118 |
"eos_token": "<|endoftext|>",
|
| 119 |
+
"legacy": false,
|
| 120 |
"model_max_length": 131072,
|
| 121 |
+
"pad_token": "<|endoftext|>",
|
|
|
|
|
|
|
| 122 |
"padding_side": "left",
|
| 123 |
"sp_model_kwargs": {},
|
|
|
|
|
|
|
| 124 |
"tokenizer_class": "LlamaTokenizer",
|
|
|
|
|
|
|
| 125 |
"unk_token": "<unk>",
|
| 126 |
"use_default_system_prompt": false
|
| 127 |
}
|