Upload special_tokens_map.json with huggingface_hub
Browse files- special_tokens_map.json +196 -0
special_tokens_map.json
ADDED
|
@@ -0,0 +1,196 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"<|endoftext|>",
|
| 4 |
+
"<|pad|>",
|
| 5 |
+
"<|unk|>",
|
| 6 |
+
"<|sep|>",
|
| 7 |
+
"<|mask|>",
|
| 8 |
+
"<|cls|>",
|
| 9 |
+
"<|image|>",
|
| 10 |
+
"<|audio|>",
|
| 11 |
+
"<|user|>",
|
| 12 |
+
"<|system|>",
|
| 13 |
+
"<|assistant|>",
|
| 14 |
+
"<|im_start|>",
|
| 15 |
+
"<|im_sep|>",
|
| 16 |
+
"<|im_end|>",
|
| 17 |
+
"<|object_ref_start|>",
|
| 18 |
+
"<|object_ref_end|>",
|
| 19 |
+
"<|ocr_text_start|>",
|
| 20 |
+
"<|ocr_text_end|>",
|
| 21 |
+
"<|point_start|>",
|
| 22 |
+
"<|point_end|>",
|
| 23 |
+
"<|box_start|>",
|
| 24 |
+
"<|box_end|>",
|
| 25 |
+
"<|polygon_start|>",
|
| 26 |
+
"<|polygon_end|>",
|
| 27 |
+
"<|image_start|>",
|
| 28 |
+
"<|image_end|>",
|
| 29 |
+
"<|image_pad|>",
|
| 30 |
+
"<|video_start|>",
|
| 31 |
+
"<|video_end|>",
|
| 32 |
+
"<|video_pad|>",
|
| 33 |
+
"<|fim_prefix|>",
|
| 34 |
+
"<|fim_middle|>",
|
| 35 |
+
"<|fim_suffix|>",
|
| 36 |
+
"<|fim_pad|>",
|
| 37 |
+
"<|resident_reg|>",
|
| 38 |
+
"<|foreigner_reg|>",
|
| 39 |
+
"<|business_reg|>",
|
| 40 |
+
"<|credit_card|>",
|
| 41 |
+
"<|passport|>",
|
| 42 |
+
"<|driver_license|>",
|
| 43 |
+
"<|telephone|>",
|
| 44 |
+
"<|health_insurance|>",
|
| 45 |
+
"<|bank_account|>",
|
| 46 |
+
"<|extra_id_0|>",
|
| 47 |
+
"<|extra_id_1|>",
|
| 48 |
+
"<|extra_id_2|>",
|
| 49 |
+
"<|extra_id_3|>",
|
| 50 |
+
"<|extra_id_4|>",
|
| 51 |
+
"<|extra_id_5|>",
|
| 52 |
+
"<|extra_id_6|>",
|
| 53 |
+
"<|extra_id_7|>",
|
| 54 |
+
"<|extra_id_8|>",
|
| 55 |
+
"<|extra_id_9|>",
|
| 56 |
+
"<|extra_id_10|>",
|
| 57 |
+
"<|extra_id_11|>",
|
| 58 |
+
"<|extra_id_12|>",
|
| 59 |
+
"<|extra_id_13|>",
|
| 60 |
+
"<|extra_id_14|>",
|
| 61 |
+
"<|extra_id_15|>",
|
| 62 |
+
"<|extra_id_16|>",
|
| 63 |
+
"<|extra_id_17|>",
|
| 64 |
+
"<|extra_id_18|>",
|
| 65 |
+
"<|extra_id_19|>",
|
| 66 |
+
"<|extra_id_20|>",
|
| 67 |
+
"<|extra_id_21|>",
|
| 68 |
+
"<|extra_id_22|>",
|
| 69 |
+
"<|extra_id_23|>",
|
| 70 |
+
"<|extra_id_24|>",
|
| 71 |
+
"<|extra_id_25|>",
|
| 72 |
+
"<|extra_id_26|>",
|
| 73 |
+
"<|extra_id_27|>",
|
| 74 |
+
"<|extra_id_28|>",
|
| 75 |
+
"<|extra_id_29|>",
|
| 76 |
+
"<|extra_id_30|>",
|
| 77 |
+
"<|extra_id_31|>",
|
| 78 |
+
"<|extra_id_32|>",
|
| 79 |
+
"<|extra_id_33|>",
|
| 80 |
+
"<|extra_id_34|>",
|
| 81 |
+
"<|extra_id_35|>",
|
| 82 |
+
"<|extra_id_36|>",
|
| 83 |
+
"<|extra_id_37|>",
|
| 84 |
+
"<|extra_id_38|>",
|
| 85 |
+
"<|extra_id_39|>",
|
| 86 |
+
"<|extra_id_40|>",
|
| 87 |
+
"<|extra_id_41|>",
|
| 88 |
+
"<|extra_id_42|>",
|
| 89 |
+
"<|extra_id_43|>",
|
| 90 |
+
"<|extra_id_44|>",
|
| 91 |
+
"<|extra_id_45|>",
|
| 92 |
+
"<|extra_id_46|>",
|
| 93 |
+
"<|extra_id_47|>",
|
| 94 |
+
"<|extra_id_48|>",
|
| 95 |
+
"<|extra_id_49|>",
|
| 96 |
+
"<|extra_id_50|>",
|
| 97 |
+
"<|extra_id_51|>",
|
| 98 |
+
"<|extra_id_52|>",
|
| 99 |
+
"<|extra_id_53|>",
|
| 100 |
+
"<|extra_id_54|>",
|
| 101 |
+
"<|extra_id_55|>",
|
| 102 |
+
"<|extra_id_56|>",
|
| 103 |
+
"<|extra_id_57|>",
|
| 104 |
+
"<|extra_id_58|>",
|
| 105 |
+
"<|extra_id_59|>",
|
| 106 |
+
"<|extra_id_60|>",
|
| 107 |
+
"<|extra_id_61|>",
|
| 108 |
+
"<|extra_id_62|>",
|
| 109 |
+
"<|extra_id_63|>",
|
| 110 |
+
"<|extra_id_64|>",
|
| 111 |
+
"<|extra_id_65|>",
|
| 112 |
+
"<|extra_id_66|>",
|
| 113 |
+
"<|extra_id_67|>",
|
| 114 |
+
"<|extra_id_68|>",
|
| 115 |
+
"<|extra_id_69|>",
|
| 116 |
+
"<|extra_id_70|>",
|
| 117 |
+
"<|extra_id_71|>",
|
| 118 |
+
"<|extra_id_72|>",
|
| 119 |
+
"<|extra_id_73|>",
|
| 120 |
+
"<|extra_id_74|>",
|
| 121 |
+
"<|extra_id_75|>",
|
| 122 |
+
"<|extra_id_76|>",
|
| 123 |
+
"<|extra_id_77|>",
|
| 124 |
+
"<|extra_id_78|>",
|
| 125 |
+
"<|extra_id_79|>",
|
| 126 |
+
"<|extra_id_80|>",
|
| 127 |
+
"<|extra_id_81|>",
|
| 128 |
+
"<|extra_id_82|>",
|
| 129 |
+
"<|extra_id_83|>",
|
| 130 |
+
"<|extra_id_84|>",
|
| 131 |
+
"<|extra_id_85|>",
|
| 132 |
+
"<|extra_id_86|>",
|
| 133 |
+
"<|extra_id_87|>",
|
| 134 |
+
"<|extra_id_88|>",
|
| 135 |
+
"<|extra_id_89|>",
|
| 136 |
+
"<|extra_id_90|>",
|
| 137 |
+
"<|extra_id_91|>",
|
| 138 |
+
"<|extra_id_92|>",
|
| 139 |
+
"<|extra_id_93|>",
|
| 140 |
+
"<|extra_id_94|>",
|
| 141 |
+
"<|extra_id_95|>",
|
| 142 |
+
"<|extra_id_96|>",
|
| 143 |
+
"<|extra_id_97|>",
|
| 144 |
+
"<|extra_id_98|>",
|
| 145 |
+
"<|extra_id_99|>"
|
| 146 |
+
],
|
| 147 |
+
"bos_token": {
|
| 148 |
+
"content": "<|endoftext|>",
|
| 149 |
+
"lstrip": false,
|
| 150 |
+
"normalized": false,
|
| 151 |
+
"rstrip": false,
|
| 152 |
+
"single_word": false
|
| 153 |
+
},
|
| 154 |
+
"cls_token": {
|
| 155 |
+
"content": "<|cls|>",
|
| 156 |
+
"lstrip": false,
|
| 157 |
+
"normalized": false,
|
| 158 |
+
"rstrip": false,
|
| 159 |
+
"single_word": false
|
| 160 |
+
},
|
| 161 |
+
"eos_token": {
|
| 162 |
+
"content": "<|im_end|>",
|
| 163 |
+
"lstrip": false,
|
| 164 |
+
"normalized": false,
|
| 165 |
+
"rstrip": false,
|
| 166 |
+
"single_word": false
|
| 167 |
+
},
|
| 168 |
+
"mask_token": {
|
| 169 |
+
"content": "<|mask|>",
|
| 170 |
+
"lstrip": false,
|
| 171 |
+
"normalized": false,
|
| 172 |
+
"rstrip": false,
|
| 173 |
+
"single_word": false
|
| 174 |
+
},
|
| 175 |
+
"pad_token": {
|
| 176 |
+
"content": "<|pad|>",
|
| 177 |
+
"lstrip": false,
|
| 178 |
+
"normalized": false,
|
| 179 |
+
"rstrip": false,
|
| 180 |
+
"single_word": false
|
| 181 |
+
},
|
| 182 |
+
"sep_token": {
|
| 183 |
+
"content": "<|sep|>",
|
| 184 |
+
"lstrip": false,
|
| 185 |
+
"normalized": false,
|
| 186 |
+
"rstrip": false,
|
| 187 |
+
"single_word": false
|
| 188 |
+
},
|
| 189 |
+
"unk_token": {
|
| 190 |
+
"content": "<|unk|>",
|
| 191 |
+
"lstrip": false,
|
| 192 |
+
"normalized": false,
|
| 193 |
+
"rstrip": false,
|
| 194 |
+
"single_word": false
|
| 195 |
+
}
|
| 196 |
+
}
|