diff --git "a/tokenizer.json" "b/tokenizer.json" new file mode 100644--- /dev/null +++ "b/tokenizer.json" @@ -0,0 +1,506994 @@ +{ + "version": "1.0", + "truncation": null, + "padding": null, + "added_tokens": [ + { + "id": 0, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": true + }, + { + "id": 1, + "content": "<|startoftext|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": true + }, + { + "id": 2, + "content": "<|endoftext|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": true + }, + { + "id": 3, + "content": "<|Human|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 4, + "content": "<|Assistant|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 5, + "content": "<|System|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 6, + "content": "<|im_start|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 7, + "content": "<|im_end|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": true + }, + { + "id": 8, + "content": "<|im_sep|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 9, + "content": "<|reserved003|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 10, + "content": "<|reserved004|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 11, + "content": "<|reserved005|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 12, + "content": "<|reserved006|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 13, + "content": "<|reserved007|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 14, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 15, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 16, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 17, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 18, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 19, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 20, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 21, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 22, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 23, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 24, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 25, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 26, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 27, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 28, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 29, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 30, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 31, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 32, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 33, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 34, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 35, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 36, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 37, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 38, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 39, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 40, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 41, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 42, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 43, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 44, + "content": "
", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 45, + "content": "
", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 46, + "content": "
", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 47, + "content": "
", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 48, + "content": "
", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 49, + "content": "
", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 50, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 51, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 52, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 53, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 54, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 55, + "content": "

", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 56, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 57, + "content": "
", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 58, + "content": "
", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 59, + "content": "
  • ", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 60, + "content": "
  • ", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 61, + "content": "
  • ", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 62, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 63, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 64, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 65, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 66, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 67, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 68, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 69, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 70, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 71, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 72, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 73, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 74, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 75, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 76, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 77, + "content": "0", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 78, + "content": "1", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 79, + "content": "2", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 80, + "content": "3", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 81, + "content": "4", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 82, + "content": "5", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 83, + "content": "6", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 84, + "content": "7", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 85, + "content": "8", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 86, + "content": "9", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 87, + "content": "0", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 88, + "content": "1", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 89, + "content": "2", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 90, + "content": "3", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 91, + "content": "4", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 92, + "content": "5", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 93, + "content": "6", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 94, + "content": "7", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 95, + "content": "8", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 96, + "content": "9", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 97, + "content": ",", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 98, + "content": ".", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 99, + "content": "!", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 100, + "content": "?", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 101, + "content": ",", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 102, + "content": "。", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 103, + "content": "!", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 104, + "content": "?", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 105, + "content": "、", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 106, + "content": ":", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 107, + "content": "¥", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 108, + "content": "《", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 109, + "content": "》", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 110, + "content": "【", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 111, + "content": "】", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 112, + "content": "『", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 113, + "content": "』", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 114, + "content": "```", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 115, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 117, + "content": "---", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 118, + "content": "", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 119, + "content": "\t\t\t\t\t\t\t\t", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 120, + "content": "\t\t\t\t\t\t\t", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 121, + "content": "\t\t\t\t\t\t", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 122, + "content": "\t\t\t\t\t", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 123, + "content": "\t\t\t\t", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 124, + "content": "\t\t\t", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 125, + "content": "\t\t", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 126, + "content": "\t", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 127, + "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 128, + "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 129, + "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 130, + "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 131, + "content": "▁▁▁▁▁▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 132, + "content": "▁▁▁▁▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 133, + "content": "▁▁▁▁▁▁▁▁▁���", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 134, + "content": "▁▁▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 135, + "content": "▁▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 136, + "content": "▁▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 137, + "content": "▁▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 138, + "content": "▁▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 139, + "content": "▁▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 140, + "content": "▁▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 141, + "content": "▁▁", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 142, + "content": "\b", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 143, + "content": "\r", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 144, + "content": "\n", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 145, + "content": "<|unused000|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 146, + "content": "<|unused001|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 147, + "content": "<|unused002|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 148, + "content": "<|unused003|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 149, + "content": "<|unused004|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 150, + "content": "<|unused005|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 151, + "content": "<|unused006|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 152, + "content": "<|unused007|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 153, + "content": "<|unused008|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 154, + "content": "<|unused009|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 155, + "content": "<|unused010|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 156, + "content": "<|unused011|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 157, + "content": "<|unused012|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 158, + "content": "<|unused013|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 159, + "content": "<|unused014|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 160, + "content": "<|unused015|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 161, + "content": "<|unused016|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 162, + "content": "<|unused017|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 163, + "content": "<|unused018|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 164, + "content": "<|unused019|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 165, + "content": "<|unused020|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 166, + "content": "<|unused021|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 167, + "content": "<|unused022|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 168, + "content": "<|unused023|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 169, + "content": "<|unused024|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 170, + "content": "<|unused025|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 171, + "content": "<|unused026|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 172, + "content": "<|unused027|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 173, + "content": "<|unused028|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 174, + "content": "<|unused029|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 175, + "content": "<|unused030|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 176, + "content": "<|unused031|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 177, + "content": "<|unused032|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 178, + "content": "<|unused033|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 179, + "content": "<|unused034|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 180, + "content": "<|unused035|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 181, + "content": "<|unused036|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 182, + "content": "<|unused037|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 183, + "content": "<|unused038|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 184, + "content": "<|unused039|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 185, + "content": "<|unused040|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 186, + "content": "<|unused041|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 187, + "content": "<|unused042|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 188, + "content": "<|unused043|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 189, + "content": "<|unused044|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 190, + "content": "<|unused045|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 191, + "content": "<|unused046|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 192, + "content": "<|unused047|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 193, + "content": "<|unused048|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 194, + "content": "<|unused049|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 195, + "content": "<|unused050|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 196, + "content": "<|unused051|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 197, + "content": "<|unused052|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 198, + "content": "<|unused053|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 199, + "content": "<|unused054|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 200, + "content": "<|unused055|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 201, + "content": "<|unused056|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 202, + "content": "<|unused057|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 203, + "content": "<|unused058|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 204, + "content": "<|unused059|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 205, + "content": "<|unused060|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 206, + "content": "<|unused061|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 207, + "content": "<|unused062|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 208, + "content": "<|unused063|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 209, + "content": "<|unused064|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 210, + "content": "<|unused065|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 211, + "content": "<|unused066|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 212, + "content": "<|unused067|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 213, + "content": "<|unused068|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 214, + "content": "<|unused069|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 215, + "content": "<|unused070|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 216, + "content": "<|unused071|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 217, + "content": "<|unused072|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 218, + "content": "<|unused073|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 219, + "content": "<|unused074|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 220, + "content": "<|unused075|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 221, + "content": "<|unused076|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 222, + "content": "<|unused077|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 223, + "content": "<|unused078|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 224, + "content": "<|unused079|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 225, + "content": "<|unused080|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 226, + "content": "<|unused081|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 227, + "content": "<|unused082|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 228, + "content": "<|unused083|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 229, + "content": "<|unused084|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 230, + "content": "<|unused085|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 231, + "content": "<|unused086|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 232, + "content": "<|unused087|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 233, + "content": "<|unused088|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 234, + "content": "<|unused089|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 235, + "content": "<|unused090|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 236, + "content": "<|unused091|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 237, + "content": "<|unused092|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 238, + "content": "<|unused093|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 239, + "content": "<|unused094|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 240, + "content": "<|unused095|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 241, + "content": "<|unused096|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 242, + "content": "<|unused097|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 243, + "content": "<|unused098|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 244, + "content": "<|unused099|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 245, + "content": "<|unused100|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 246, + "content": "<|unused101|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 247, + "content": "<|unused102|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 248, + "content": "<|unused103|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 249, + "content": "<|unused104|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 250, + "content": "<|unused105|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 251, + "content": "<|unused106|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 252, + "content": "<|unused107|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 253, + "content": "<|unused108|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 254, + "content": "<|unused109|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 255, + "content": "<|unused110|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 256, + "content": "<|unused111|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 257, + "content": "<|unused112|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 258, + "content": "<|unused113|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 259, + "content": "<|unused114|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 260, + "content": "<|unused115|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 261, + "content": "<|unused116|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 262, + "content": "<|unused117|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 263, + "content": "<|unused118|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 264, + "content": "<|unused119|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 265, + "content": "<|unused120|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 266, + "content": "<|unused121|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 267, + "content": "<|unused122|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 268, + "content": "<|unused123|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 269, + "content": "<|unused124|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 270, + "content": "<|unused125|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 271, + "content": "<|unused126|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 272, + "content": "<|unused127|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 273, + "content": "<|unused128|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 274, + "content": "<|unused129|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 275, + "content": "<|unused130|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 276, + "content": "<|unused131|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 277, + "content": "<|unused132|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 278, + "content": "<|unused133|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 279, + "content": "<|unused134|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 280, + "content": "<|unused135|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 281, + "content": "<|unused136|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 282, + "content": "<|unused137|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 283, + "content": "<|unused138|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 284, + "content": "<|unused139|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 285, + "content": "<|unused140|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 286, + "content": "<|unused141|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 287, + "content": "<|unused142|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 288, + "content": "<|unused143|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 289, + "content": "<|unused144|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 290, + "content": "<|unused145|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 291, + "content": "<|unused146|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 292, + "content": "<|unused147|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 293, + "content": "<|unused148|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 294, + "content": "<|unused149|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 295, + "content": "<|unused150|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 296, + "content": "<|unused151|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 297, + "content": "<|unused152|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 298, + "content": "<|unused153|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 299, + "content": "<|unused154|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 300, + "content": "<|unused155|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 301, + "content": "<|unused156|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 302, + "content": "<|unused157|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 303, + "content": "<|unused158|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 304, + "content": "<|unused159|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 64000, + "content": "<|reserved000|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 64001, + "content": "<|reserved001|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 64002, + "content": "<|reserved002|>", + "single_word": false, + "lstrip": false, + "rstrip": false, + "normalized": false, + "special": false + }, + { + "id": 64003, + "content": "