Upload tokenizer
Browse files- special_tokens_map.json +111 -0
- tokenizer.json +0 -0
- tokenizer_config.json +6 -0
special_tokens_map.json
ADDED
|
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"[sentinel_0]",
|
| 4 |
+
"[sentinel_1]",
|
| 5 |
+
"[sentinel_2]",
|
| 6 |
+
"[sentinel_3]",
|
| 7 |
+
"[sentinel_4]",
|
| 8 |
+
"[sentinel_5]",
|
| 9 |
+
"[sentinel_6]",
|
| 10 |
+
"[sentinel_7]",
|
| 11 |
+
"[sentinel_8]",
|
| 12 |
+
"[sentinel_9]",
|
| 13 |
+
"[sentinel_10]",
|
| 14 |
+
"[sentinel_11]",
|
| 15 |
+
"[sentinel_12]",
|
| 16 |
+
"[sentinel_13]",
|
| 17 |
+
"[sentinel_14]",
|
| 18 |
+
"[sentinel_15]",
|
| 19 |
+
"[sentinel_16]",
|
| 20 |
+
"[sentinel_17]",
|
| 21 |
+
"[sentinel_18]",
|
| 22 |
+
"[sentinel_19]",
|
| 23 |
+
"[sentinel_20]",
|
| 24 |
+
"[sentinel_21]",
|
| 25 |
+
"[sentinel_22]",
|
| 26 |
+
"[sentinel_23]",
|
| 27 |
+
"[sentinel_24]",
|
| 28 |
+
"[sentinel_25]",
|
| 29 |
+
"[sentinel_26]",
|
| 30 |
+
"[sentinel_27]",
|
| 31 |
+
"[sentinel_28]",
|
| 32 |
+
"[sentinel_29]",
|
| 33 |
+
"[sentinel_30]",
|
| 34 |
+
"[sentinel_31]",
|
| 35 |
+
"[sentinel_32]",
|
| 36 |
+
"[sentinel_33]",
|
| 37 |
+
"[sentinel_34]",
|
| 38 |
+
"[sentinel_35]",
|
| 39 |
+
"[sentinel_36]",
|
| 40 |
+
"[sentinel_37]",
|
| 41 |
+
"[sentinel_38]",
|
| 42 |
+
"[sentinel_39]",
|
| 43 |
+
"[sentinel_40]",
|
| 44 |
+
"[sentinel_41]",
|
| 45 |
+
"[sentinel_42]",
|
| 46 |
+
"[sentinel_43]",
|
| 47 |
+
"[sentinel_44]",
|
| 48 |
+
"[sentinel_45]",
|
| 49 |
+
"[sentinel_46]",
|
| 50 |
+
"[sentinel_47]",
|
| 51 |
+
"[sentinel_48]",
|
| 52 |
+
"[sentinel_49]",
|
| 53 |
+
"[sentinel_50]",
|
| 54 |
+
"[sentinel_51]",
|
| 55 |
+
"[sentinel_52]",
|
| 56 |
+
"[sentinel_53]",
|
| 57 |
+
"[sentinel_54]",
|
| 58 |
+
"[sentinel_55]",
|
| 59 |
+
"[sentinel_56]",
|
| 60 |
+
"[sentinel_57]",
|
| 61 |
+
"[sentinel_58]",
|
| 62 |
+
"[sentinel_59]",
|
| 63 |
+
"[sentinel_60]",
|
| 64 |
+
"[sentinel_61]",
|
| 65 |
+
"[sentinel_62]",
|
| 66 |
+
"[sentinel_63]",
|
| 67 |
+
"[sentinel_64]",
|
| 68 |
+
"[sentinel_65]",
|
| 69 |
+
"[sentinel_66]",
|
| 70 |
+
"[sentinel_67]",
|
| 71 |
+
"[sentinel_68]",
|
| 72 |
+
"[sentinel_69]",
|
| 73 |
+
"[sentinel_70]",
|
| 74 |
+
"[sentinel_71]",
|
| 75 |
+
"[sentinel_72]",
|
| 76 |
+
"[sentinel_73]",
|
| 77 |
+
"[sentinel_74]",
|
| 78 |
+
"[sentinel_75]",
|
| 79 |
+
"[sentinel_76]",
|
| 80 |
+
"[sentinel_77]",
|
| 81 |
+
"[sentinel_78]",
|
| 82 |
+
"[sentinel_79]",
|
| 83 |
+
"[sentinel_80]",
|
| 84 |
+
"[sentinel_81]",
|
| 85 |
+
"[sentinel_82]",
|
| 86 |
+
"[sentinel_83]",
|
| 87 |
+
"[sentinel_84]",
|
| 88 |
+
"[sentinel_85]",
|
| 89 |
+
"[sentinel_86]",
|
| 90 |
+
"[sentinel_87]",
|
| 91 |
+
"[sentinel_88]",
|
| 92 |
+
"[sentinel_89]",
|
| 93 |
+
"[sentinel_90]",
|
| 94 |
+
"[sentinel_91]",
|
| 95 |
+
"[sentinel_92]",
|
| 96 |
+
"[sentinel_93]",
|
| 97 |
+
"[sentinel_94]",
|
| 98 |
+
"[sentinel_95]",
|
| 99 |
+
"[sentinel_96]",
|
| 100 |
+
"[sentinel_97]",
|
| 101 |
+
"[sentinel_98]",
|
| 102 |
+
"[sentinel_99]",
|
| 103 |
+
"[R]",
|
| 104 |
+
"[X]",
|
| 105 |
+
"[S]"
|
| 106 |
+
],
|
| 107 |
+
"eos_token": "[EOS]",
|
| 108 |
+
"pad_token": "[PAD]",
|
| 109 |
+
"sep_token": "[SEP]",
|
| 110 |
+
"unk_token": "[UNK]"
|
| 111 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 3 |
+
"name_or_path": "law_tokenizer",
|
| 4 |
+
"special_tokens_map_file": "test_tokenizer/special_tokens_map.json",
|
| 5 |
+
"tokenizer_class": "PreTrainedTokenizerFast"
|
| 6 |
+
}
|