Upload tokenizer
Browse files- tokenizer_config.json +10 -11
tokenizer_config.json
CHANGED
|
@@ -4,7 +4,7 @@
|
|
| 4 |
"add_prefix_space": null,
|
| 5 |
"added_tokens_decoder": {
|
| 6 |
"128000": {
|
| 7 |
-
"content": "
|
| 8 |
"lstrip": false,
|
| 9 |
"normalized": false,
|
| 10 |
"rstrip": false,
|
|
@@ -12,7 +12,7 @@
|
|
| 12 |
"special": true
|
| 13 |
},
|
| 14 |
"128001": {
|
| 15 |
-
"content": "
|
| 16 |
"lstrip": false,
|
| 17 |
"normalized": false,
|
| 18 |
"rstrip": false,
|
|
@@ -92,7 +92,7 @@
|
|
| 92 |
"special": true
|
| 93 |
},
|
| 94 |
"128011": {
|
| 95 |
-
"content": "
|
| 96 |
"lstrip": false,
|
| 97 |
"normalized": false,
|
| 98 |
"rstrip": false,
|
|
@@ -100,7 +100,7 @@
|
|
| 100 |
"special": false
|
| 101 |
},
|
| 102 |
"128012": {
|
| 103 |
-
"content": "
|
| 104 |
"lstrip": false,
|
| 105 |
"normalized": false,
|
| 106 |
"rstrip": false,
|
|
@@ -124,7 +124,7 @@
|
|
| 124 |
"special": false
|
| 125 |
},
|
| 126 |
"128015": {
|
| 127 |
-
"content": "
|
| 128 |
"lstrip": false,
|
| 129 |
"normalized": false,
|
| 130 |
"rstrip": false,
|
|
@@ -2052,17 +2052,16 @@
|
|
| 2052 |
"special": true
|
| 2053 |
}
|
| 2054 |
},
|
| 2055 |
-
"bos_token": "
|
| 2056 |
"clean_up_tokenization_spaces": false,
|
| 2057 |
-
"eos_token": "
|
| 2058 |
"extra_special_tokens": {},
|
| 2059 |
"legacy": true,
|
| 2060 |
-
"model_max_length":
|
| 2061 |
"pad_token": "<|finetune_right_pad_id|>",
|
| 2062 |
"padding_side": "left",
|
| 2063 |
"sp_model_kwargs": {},
|
| 2064 |
"tokenizer_class": "LlamaTokenizerFast",
|
| 2065 |
"unk_token": null,
|
| 2066 |
-
"use_default_system_prompt": false
|
| 2067 |
-
|
| 2068 |
-
}
|
|
|
|
| 4 |
"add_prefix_space": null,
|
| 5 |
"added_tokens_decoder": {
|
| 6 |
"128000": {
|
| 7 |
+
"content": "<|begin▁of▁sentence|>",
|
| 8 |
"lstrip": false,
|
| 9 |
"normalized": false,
|
| 10 |
"rstrip": false,
|
|
|
|
| 12 |
"special": true
|
| 13 |
},
|
| 14 |
"128001": {
|
| 15 |
+
"content": "<|end▁of▁sentence|>",
|
| 16 |
"lstrip": false,
|
| 17 |
"normalized": false,
|
| 18 |
"rstrip": false,
|
|
|
|
| 92 |
"special": true
|
| 93 |
},
|
| 94 |
"128011": {
|
| 95 |
+
"content": "<|User|>",
|
| 96 |
"lstrip": false,
|
| 97 |
"normalized": false,
|
| 98 |
"rstrip": false,
|
|
|
|
| 100 |
"special": false
|
| 101 |
},
|
| 102 |
"128012": {
|
| 103 |
+
"content": "<|Assistant|>",
|
| 104 |
"lstrip": false,
|
| 105 |
"normalized": false,
|
| 106 |
"rstrip": false,
|
|
|
|
| 124 |
"special": false
|
| 125 |
},
|
| 126 |
"128015": {
|
| 127 |
+
"content": "<|▁pad▁|>",
|
| 128 |
"lstrip": false,
|
| 129 |
"normalized": false,
|
| 130 |
"rstrip": false,
|
|
|
|
| 2052 |
"special": true
|
| 2053 |
}
|
| 2054 |
},
|
| 2055 |
+
"bos_token": "<|begin▁of▁sentence|>",
|
| 2056 |
"clean_up_tokenization_spaces": false,
|
| 2057 |
+
"eos_token": "<|end▁of▁sentence|>",
|
| 2058 |
"extra_special_tokens": {},
|
| 2059 |
"legacy": true,
|
| 2060 |
+
"model_max_length": 131072,
|
| 2061 |
"pad_token": "<|finetune_right_pad_id|>",
|
| 2062 |
"padding_side": "left",
|
| 2063 |
"sp_model_kwargs": {},
|
| 2064 |
"tokenizer_class": "LlamaTokenizerFast",
|
| 2065 |
"unk_token": null,
|
| 2066 |
+
"use_default_system_prompt": false
|
| 2067 |
+
}
|
|
|