Commit
·
4e933fb
1
Parent(s):
e2861b9
Upload 6 files
Browse files- config.json +736 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +7 -0
- tokenizer.json +0 -0
- tokenizer_config.json +15 -0
- vocab.txt +0 -0
config.json
ADDED
|
@@ -0,0 +1,736 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"Avg. Length": 99.17900753442453,
|
| 3 |
+
"_name_or_path": "../models/envbert/",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"BertCRF"
|
| 6 |
+
],
|
| 7 |
+
"attention_probs_dropout_prob": 0.1,
|
| 8 |
+
"classifier_dropout": null,
|
| 9 |
+
"contact": "[email protected]",
|
| 10 |
+
"counter": {
|
| 11 |
+
"a": 1366,
|
| 12 |
+
"act": 1337,
|
| 13 |
+
"animal": 285,
|
| 14 |
+
"b": 2359,
|
| 15 |
+
"c": 180,
|
| 16 |
+
"code": 300,
|
| 17 |
+
"com": 201,
|
| 18 |
+
"conj": 6814,
|
| 19 |
+
"d": 3921,
|
| 20 |
+
"desease": 650,
|
| 21 |
+
"doc": 940,
|
| 22 |
+
"env": 3012,
|
| 23 |
+
"event": 169,
|
| 24 |
+
"f": 2095,
|
| 25 |
+
"gov": 729,
|
| 26 |
+
"group": 533,
|
| 27 |
+
"hy": 562,
|
| 28 |
+
"ins": 555,
|
| 29 |
+
"loc": 1566,
|
| 30 |
+
"m": 2608,
|
| 31 |
+
"means": 969,
|
| 32 |
+
"med": 1641,
|
| 33 |
+
"meet": 199,
|
| 34 |
+
"microbe": 296,
|
| 35 |
+
"n": 4565,
|
| 36 |
+
"ord": 727,
|
| 37 |
+
"org": 109,
|
| 38 |
+
"p": 3171,
|
| 39 |
+
"per": 223,
|
| 40 |
+
"phe": 671,
|
| 41 |
+
"plant": 370,
|
| 42 |
+
"pol": 3945,
|
| 43 |
+
"policy": 626,
|
| 44 |
+
"pro": 466,
|
| 45 |
+
"q": 1851,
|
| 46 |
+
"r": 1742,
|
| 47 |
+
"time": 960,
|
| 48 |
+
"u": 5949,
|
| 49 |
+
"v": 12564,
|
| 50 |
+
"vn": 2884,
|
| 51 |
+
"w": 10260,
|
| 52 |
+
"xc": 129
|
| 53 |
+
},
|
| 54 |
+
"directionality": "bidi",
|
| 55 |
+
"entities": [
|
| 56 |
+
"doc",
|
| 57 |
+
"d",
|
| 58 |
+
"v",
|
| 59 |
+
"meet",
|
| 60 |
+
"w",
|
| 61 |
+
"conj",
|
| 62 |
+
"group",
|
| 63 |
+
"env",
|
| 64 |
+
"time",
|
| 65 |
+
"u",
|
| 66 |
+
"gov",
|
| 67 |
+
"act",
|
| 68 |
+
"a",
|
| 69 |
+
"n",
|
| 70 |
+
"pro",
|
| 71 |
+
"loc",
|
| 72 |
+
"med",
|
| 73 |
+
"vn",
|
| 74 |
+
"event",
|
| 75 |
+
"per",
|
| 76 |
+
"c",
|
| 77 |
+
"m",
|
| 78 |
+
"phe",
|
| 79 |
+
"pol",
|
| 80 |
+
"f",
|
| 81 |
+
"p",
|
| 82 |
+
"com",
|
| 83 |
+
"policy",
|
| 84 |
+
"ord",
|
| 85 |
+
"q",
|
| 86 |
+
"r",
|
| 87 |
+
"desease",
|
| 88 |
+
"b",
|
| 89 |
+
"ins",
|
| 90 |
+
"means",
|
| 91 |
+
"org",
|
| 92 |
+
"microbe",
|
| 93 |
+
"hy",
|
| 94 |
+
"plant",
|
| 95 |
+
"xc",
|
| 96 |
+
"code",
|
| 97 |
+
"animal"
|
| 98 |
+
],
|
| 99 |
+
"entity2id": {
|
| 100 |
+
"a": 12,
|
| 101 |
+
"act": 11,
|
| 102 |
+
"animal": 41,
|
| 103 |
+
"b": 32,
|
| 104 |
+
"c": 20,
|
| 105 |
+
"code": 40,
|
| 106 |
+
"com": 26,
|
| 107 |
+
"conj": 5,
|
| 108 |
+
"d": 1,
|
| 109 |
+
"desease": 31,
|
| 110 |
+
"doc": 0,
|
| 111 |
+
"env": 7,
|
| 112 |
+
"event": 18,
|
| 113 |
+
"f": 24,
|
| 114 |
+
"gov": 10,
|
| 115 |
+
"group": 6,
|
| 116 |
+
"hy": 37,
|
| 117 |
+
"ins": 33,
|
| 118 |
+
"loc": 15,
|
| 119 |
+
"m": 21,
|
| 120 |
+
"means": 34,
|
| 121 |
+
"med": 16,
|
| 122 |
+
"meet": 3,
|
| 123 |
+
"microbe": 36,
|
| 124 |
+
"n": 13,
|
| 125 |
+
"ord": 28,
|
| 126 |
+
"org": 35,
|
| 127 |
+
"p": 25,
|
| 128 |
+
"per": 19,
|
| 129 |
+
"phe": 22,
|
| 130 |
+
"plant": 38,
|
| 131 |
+
"pol": 23,
|
| 132 |
+
"policy": 27,
|
| 133 |
+
"pro": 14,
|
| 134 |
+
"q": 29,
|
| 135 |
+
"r": 30,
|
| 136 |
+
"time": 8,
|
| 137 |
+
"u": 9,
|
| 138 |
+
"v": 2,
|
| 139 |
+
"vn": 17,
|
| 140 |
+
"w": 4,
|
| 141 |
+
"xc": 39
|
| 142 |
+
},
|
| 143 |
+
"hidden_act": "gelu",
|
| 144 |
+
"hidden_dropout_prob": 0.1,
|
| 145 |
+
"hidden_size": 768,
|
| 146 |
+
"id2entity": {
|
| 147 |
+
"0": "doc",
|
| 148 |
+
"1": "d",
|
| 149 |
+
"2": "v",
|
| 150 |
+
"3": "meet",
|
| 151 |
+
"4": "w",
|
| 152 |
+
"5": "conj",
|
| 153 |
+
"6": "group",
|
| 154 |
+
"7": "env",
|
| 155 |
+
"8": "time",
|
| 156 |
+
"9": "u",
|
| 157 |
+
"10": "gov",
|
| 158 |
+
"11": "act",
|
| 159 |
+
"12": "a",
|
| 160 |
+
"13": "n",
|
| 161 |
+
"14": "pro",
|
| 162 |
+
"15": "loc",
|
| 163 |
+
"16": "med",
|
| 164 |
+
"17": "vn",
|
| 165 |
+
"18": "event",
|
| 166 |
+
"19": "per",
|
| 167 |
+
"20": "c",
|
| 168 |
+
"21": "m",
|
| 169 |
+
"22": "phe",
|
| 170 |
+
"23": "pol",
|
| 171 |
+
"24": "f",
|
| 172 |
+
"25": "p",
|
| 173 |
+
"26": "com",
|
| 174 |
+
"27": "policy",
|
| 175 |
+
"28": "ord",
|
| 176 |
+
"29": "q",
|
| 177 |
+
"30": "r",
|
| 178 |
+
"31": "desease",
|
| 179 |
+
"32": "b",
|
| 180 |
+
"33": "ins",
|
| 181 |
+
"34": "means",
|
| 182 |
+
"35": "org",
|
| 183 |
+
"36": "microbe",
|
| 184 |
+
"37": "hy",
|
| 185 |
+
"38": "plant",
|
| 186 |
+
"39": "xc",
|
| 187 |
+
"40": "code",
|
| 188 |
+
"41": "animal"
|
| 189 |
+
},
|
| 190 |
+
"id2label": {
|
| 191 |
+
"0": "O",
|
| 192 |
+
"1": "B-doc",
|
| 193 |
+
"2": "I-doc",
|
| 194 |
+
"3": "E-doc",
|
| 195 |
+
"4": "S-doc",
|
| 196 |
+
"5": "B-d",
|
| 197 |
+
"6": "I-d",
|
| 198 |
+
"7": "E-d",
|
| 199 |
+
"8": "S-d",
|
| 200 |
+
"9": "B-v",
|
| 201 |
+
"10": "I-v",
|
| 202 |
+
"11": "E-v",
|
| 203 |
+
"12": "S-v",
|
| 204 |
+
"13": "B-meet",
|
| 205 |
+
"14": "I-meet",
|
| 206 |
+
"15": "E-meet",
|
| 207 |
+
"16": "S-meet",
|
| 208 |
+
"17": "B-w",
|
| 209 |
+
"18": "I-w",
|
| 210 |
+
"19": "E-w",
|
| 211 |
+
"20": "S-w",
|
| 212 |
+
"21": "B-conj",
|
| 213 |
+
"22": "I-conj",
|
| 214 |
+
"23": "E-conj",
|
| 215 |
+
"24": "S-conj",
|
| 216 |
+
"25": "B-group",
|
| 217 |
+
"26": "I-group",
|
| 218 |
+
"27": "E-group",
|
| 219 |
+
"28": "S-group",
|
| 220 |
+
"29": "B-env",
|
| 221 |
+
"30": "I-env",
|
| 222 |
+
"31": "E-env",
|
| 223 |
+
"32": "S-env",
|
| 224 |
+
"33": "B-time",
|
| 225 |
+
"34": "I-time",
|
| 226 |
+
"35": "E-time",
|
| 227 |
+
"36": "S-time",
|
| 228 |
+
"37": "B-u",
|
| 229 |
+
"38": "I-u",
|
| 230 |
+
"39": "E-u",
|
| 231 |
+
"40": "S-u",
|
| 232 |
+
"41": "B-gov",
|
| 233 |
+
"42": "I-gov",
|
| 234 |
+
"43": "E-gov",
|
| 235 |
+
"44": "S-gov",
|
| 236 |
+
"45": "B-act",
|
| 237 |
+
"46": "I-act",
|
| 238 |
+
"47": "E-act",
|
| 239 |
+
"48": "S-act",
|
| 240 |
+
"49": "B-a",
|
| 241 |
+
"50": "I-a",
|
| 242 |
+
"51": "E-a",
|
| 243 |
+
"52": "S-a",
|
| 244 |
+
"53": "B-n",
|
| 245 |
+
"54": "I-n",
|
| 246 |
+
"55": "E-n",
|
| 247 |
+
"56": "S-n",
|
| 248 |
+
"57": "B-pro",
|
| 249 |
+
"58": "I-pro",
|
| 250 |
+
"59": "E-pro",
|
| 251 |
+
"60": "S-pro",
|
| 252 |
+
"61": "B-loc",
|
| 253 |
+
"62": "I-loc",
|
| 254 |
+
"63": "E-loc",
|
| 255 |
+
"64": "S-loc",
|
| 256 |
+
"65": "B-med",
|
| 257 |
+
"66": "I-med",
|
| 258 |
+
"67": "E-med",
|
| 259 |
+
"68": "S-med",
|
| 260 |
+
"69": "B-vn",
|
| 261 |
+
"70": "I-vn",
|
| 262 |
+
"71": "E-vn",
|
| 263 |
+
"72": "S-vn",
|
| 264 |
+
"73": "B-event",
|
| 265 |
+
"74": "I-event",
|
| 266 |
+
"75": "E-event",
|
| 267 |
+
"76": "S-event",
|
| 268 |
+
"77": "B-per",
|
| 269 |
+
"78": "I-per",
|
| 270 |
+
"79": "E-per",
|
| 271 |
+
"80": "S-per",
|
| 272 |
+
"81": "B-c",
|
| 273 |
+
"82": "I-c",
|
| 274 |
+
"83": "E-c",
|
| 275 |
+
"84": "S-c",
|
| 276 |
+
"85": "B-m",
|
| 277 |
+
"86": "I-m",
|
| 278 |
+
"87": "E-m",
|
| 279 |
+
"88": "S-m",
|
| 280 |
+
"89": "B-phe",
|
| 281 |
+
"90": "I-phe",
|
| 282 |
+
"91": "E-phe",
|
| 283 |
+
"92": "S-phe",
|
| 284 |
+
"93": "B-pol",
|
| 285 |
+
"94": "I-pol",
|
| 286 |
+
"95": "E-pol",
|
| 287 |
+
"96": "S-pol",
|
| 288 |
+
"97": "B-f",
|
| 289 |
+
"98": "I-f",
|
| 290 |
+
"99": "E-f",
|
| 291 |
+
"100": "S-f",
|
| 292 |
+
"101": "B-p",
|
| 293 |
+
"102": "I-p",
|
| 294 |
+
"103": "E-p",
|
| 295 |
+
"104": "S-p",
|
| 296 |
+
"105": "B-com",
|
| 297 |
+
"106": "I-com",
|
| 298 |
+
"107": "E-com",
|
| 299 |
+
"108": "S-com",
|
| 300 |
+
"109": "B-policy",
|
| 301 |
+
"110": "I-policy",
|
| 302 |
+
"111": "E-policy",
|
| 303 |
+
"112": "S-policy",
|
| 304 |
+
"113": "B-ord",
|
| 305 |
+
"114": "I-ord",
|
| 306 |
+
"115": "E-ord",
|
| 307 |
+
"116": "S-ord",
|
| 308 |
+
"117": "B-q",
|
| 309 |
+
"118": "I-q",
|
| 310 |
+
"119": "E-q",
|
| 311 |
+
"120": "S-q",
|
| 312 |
+
"121": "B-r",
|
| 313 |
+
"122": "I-r",
|
| 314 |
+
"123": "E-r",
|
| 315 |
+
"124": "S-r",
|
| 316 |
+
"125": "B-desease",
|
| 317 |
+
"126": "I-desease",
|
| 318 |
+
"127": "E-desease",
|
| 319 |
+
"128": "S-desease",
|
| 320 |
+
"129": "B-b",
|
| 321 |
+
"130": "I-b",
|
| 322 |
+
"131": "E-b",
|
| 323 |
+
"132": "S-b",
|
| 324 |
+
"133": "B-ins",
|
| 325 |
+
"134": "I-ins",
|
| 326 |
+
"135": "E-ins",
|
| 327 |
+
"136": "S-ins",
|
| 328 |
+
"137": "B-means",
|
| 329 |
+
"138": "I-means",
|
| 330 |
+
"139": "E-means",
|
| 331 |
+
"140": "S-means",
|
| 332 |
+
"141": "B-org",
|
| 333 |
+
"142": "I-org",
|
| 334 |
+
"143": "E-org",
|
| 335 |
+
"144": "S-org",
|
| 336 |
+
"145": "B-microbe",
|
| 337 |
+
"146": "I-microbe",
|
| 338 |
+
"147": "E-microbe",
|
| 339 |
+
"148": "S-microbe",
|
| 340 |
+
"149": "B-hy",
|
| 341 |
+
"150": "I-hy",
|
| 342 |
+
"151": "E-hy",
|
| 343 |
+
"152": "S-hy",
|
| 344 |
+
"153": "B-plant",
|
| 345 |
+
"154": "I-plant",
|
| 346 |
+
"155": "E-plant",
|
| 347 |
+
"156": "S-plant",
|
| 348 |
+
"157": "B-xc",
|
| 349 |
+
"158": "I-xc",
|
| 350 |
+
"159": "E-xc",
|
| 351 |
+
"160": "S-xc",
|
| 352 |
+
"161": "B-code",
|
| 353 |
+
"162": "I-code",
|
| 354 |
+
"163": "E-code",
|
| 355 |
+
"164": "S-code",
|
| 356 |
+
"165": "B-animal",
|
| 357 |
+
"166": "I-animal",
|
| 358 |
+
"167": "E-animal",
|
| 359 |
+
"168": "S-animal"
|
| 360 |
+
},
|
| 361 |
+
"initializer_range": 0.02,
|
| 362 |
+
"intermediate_size": 3072,
|
| 363 |
+
"jieba": false,
|
| 364 |
+
"key_metric": "macro_f1",
|
| 365 |
+
"label2id": {
|
| 366 |
+
"B-a": 49,
|
| 367 |
+
"B-act": 45,
|
| 368 |
+
"B-animal": 165,
|
| 369 |
+
"B-b": 129,
|
| 370 |
+
"B-c": 81,
|
| 371 |
+
"B-code": 161,
|
| 372 |
+
"B-com": 105,
|
| 373 |
+
"B-conj": 21,
|
| 374 |
+
"B-d": 5,
|
| 375 |
+
"B-desease": 125,
|
| 376 |
+
"B-doc": 1,
|
| 377 |
+
"B-env": 29,
|
| 378 |
+
"B-event": 73,
|
| 379 |
+
"B-f": 97,
|
| 380 |
+
"B-gov": 41,
|
| 381 |
+
"B-group": 25,
|
| 382 |
+
"B-hy": 149,
|
| 383 |
+
"B-ins": 133,
|
| 384 |
+
"B-loc": 61,
|
| 385 |
+
"B-m": 85,
|
| 386 |
+
"B-means": 137,
|
| 387 |
+
"B-med": 65,
|
| 388 |
+
"B-meet": 13,
|
| 389 |
+
"B-microbe": 145,
|
| 390 |
+
"B-n": 53,
|
| 391 |
+
"B-ord": 113,
|
| 392 |
+
"B-org": 141,
|
| 393 |
+
"B-p": 101,
|
| 394 |
+
"B-per": 77,
|
| 395 |
+
"B-phe": 89,
|
| 396 |
+
"B-plant": 153,
|
| 397 |
+
"B-pol": 93,
|
| 398 |
+
"B-policy": 109,
|
| 399 |
+
"B-pro": 57,
|
| 400 |
+
"B-q": 117,
|
| 401 |
+
"B-r": 121,
|
| 402 |
+
"B-time": 33,
|
| 403 |
+
"B-u": 37,
|
| 404 |
+
"B-v": 9,
|
| 405 |
+
"B-vn": 69,
|
| 406 |
+
"B-w": 17,
|
| 407 |
+
"B-xc": 157,
|
| 408 |
+
"E-a": 51,
|
| 409 |
+
"E-act": 47,
|
| 410 |
+
"E-animal": 167,
|
| 411 |
+
"E-b": 131,
|
| 412 |
+
"E-c": 83,
|
| 413 |
+
"E-code": 163,
|
| 414 |
+
"E-com": 107,
|
| 415 |
+
"E-conj": 23,
|
| 416 |
+
"E-d": 7,
|
| 417 |
+
"E-desease": 127,
|
| 418 |
+
"E-doc": 3,
|
| 419 |
+
"E-env": 31,
|
| 420 |
+
"E-event": 75,
|
| 421 |
+
"E-f": 99,
|
| 422 |
+
"E-gov": 43,
|
| 423 |
+
"E-group": 27,
|
| 424 |
+
"E-hy": 151,
|
| 425 |
+
"E-ins": 135,
|
| 426 |
+
"E-loc": 63,
|
| 427 |
+
"E-m": 87,
|
| 428 |
+
"E-means": 139,
|
| 429 |
+
"E-med": 67,
|
| 430 |
+
"E-meet": 15,
|
| 431 |
+
"E-microbe": 147,
|
| 432 |
+
"E-n": 55,
|
| 433 |
+
"E-ord": 115,
|
| 434 |
+
"E-org": 143,
|
| 435 |
+
"E-p": 103,
|
| 436 |
+
"E-per": 79,
|
| 437 |
+
"E-phe": 91,
|
| 438 |
+
"E-plant": 155,
|
| 439 |
+
"E-pol": 95,
|
| 440 |
+
"E-policy": 111,
|
| 441 |
+
"E-pro": 59,
|
| 442 |
+
"E-q": 119,
|
| 443 |
+
"E-r": 123,
|
| 444 |
+
"E-time": 35,
|
| 445 |
+
"E-u": 39,
|
| 446 |
+
"E-v": 11,
|
| 447 |
+
"E-vn": 71,
|
| 448 |
+
"E-w": 19,
|
| 449 |
+
"E-xc": 159,
|
| 450 |
+
"I-a": 50,
|
| 451 |
+
"I-act": 46,
|
| 452 |
+
"I-animal": 166,
|
| 453 |
+
"I-b": 130,
|
| 454 |
+
"I-c": 82,
|
| 455 |
+
"I-code": 162,
|
| 456 |
+
"I-com": 106,
|
| 457 |
+
"I-conj": 22,
|
| 458 |
+
"I-d": 6,
|
| 459 |
+
"I-desease": 126,
|
| 460 |
+
"I-doc": 2,
|
| 461 |
+
"I-env": 30,
|
| 462 |
+
"I-event": 74,
|
| 463 |
+
"I-f": 98,
|
| 464 |
+
"I-gov": 42,
|
| 465 |
+
"I-group": 26,
|
| 466 |
+
"I-hy": 150,
|
| 467 |
+
"I-ins": 134,
|
| 468 |
+
"I-loc": 62,
|
| 469 |
+
"I-m": 86,
|
| 470 |
+
"I-means": 138,
|
| 471 |
+
"I-med": 66,
|
| 472 |
+
"I-meet": 14,
|
| 473 |
+
"I-microbe": 146,
|
| 474 |
+
"I-n": 54,
|
| 475 |
+
"I-ord": 114,
|
| 476 |
+
"I-org": 142,
|
| 477 |
+
"I-p": 102,
|
| 478 |
+
"I-per": 78,
|
| 479 |
+
"I-phe": 90,
|
| 480 |
+
"I-plant": 154,
|
| 481 |
+
"I-pol": 94,
|
| 482 |
+
"I-policy": 110,
|
| 483 |
+
"I-pro": 58,
|
| 484 |
+
"I-q": 118,
|
| 485 |
+
"I-r": 122,
|
| 486 |
+
"I-time": 34,
|
| 487 |
+
"I-u": 38,
|
| 488 |
+
"I-v": 10,
|
| 489 |
+
"I-vn": 70,
|
| 490 |
+
"I-w": 18,
|
| 491 |
+
"I-xc": 158,
|
| 492 |
+
"O": 0,
|
| 493 |
+
"S-a": 52,
|
| 494 |
+
"S-act": 48,
|
| 495 |
+
"S-animal": 168,
|
| 496 |
+
"S-b": 132,
|
| 497 |
+
"S-c": 84,
|
| 498 |
+
"S-code": 164,
|
| 499 |
+
"S-com": 108,
|
| 500 |
+
"S-conj": 24,
|
| 501 |
+
"S-d": 8,
|
| 502 |
+
"S-desease": 128,
|
| 503 |
+
"S-doc": 4,
|
| 504 |
+
"S-env": 32,
|
| 505 |
+
"S-event": 76,
|
| 506 |
+
"S-f": 100,
|
| 507 |
+
"S-gov": 44,
|
| 508 |
+
"S-group": 28,
|
| 509 |
+
"S-hy": 152,
|
| 510 |
+
"S-ins": 136,
|
| 511 |
+
"S-loc": 64,
|
| 512 |
+
"S-m": 88,
|
| 513 |
+
"S-means": 140,
|
| 514 |
+
"S-med": 68,
|
| 515 |
+
"S-meet": 16,
|
| 516 |
+
"S-microbe": 148,
|
| 517 |
+
"S-n": 56,
|
| 518 |
+
"S-ord": 116,
|
| 519 |
+
"S-org": 144,
|
| 520 |
+
"S-p": 104,
|
| 521 |
+
"S-per": 80,
|
| 522 |
+
"S-phe": 92,
|
| 523 |
+
"S-plant": 156,
|
| 524 |
+
"S-pol": 96,
|
| 525 |
+
"S-policy": 112,
|
| 526 |
+
"S-pro": 60,
|
| 527 |
+
"S-q": 120,
|
| 528 |
+
"S-r": 124,
|
| 529 |
+
"S-time": 36,
|
| 530 |
+
"S-u": 40,
|
| 531 |
+
"S-v": 12,
|
| 532 |
+
"S-vn": 72,
|
| 533 |
+
"S-w": 20,
|
| 534 |
+
"S-xc": 160
|
| 535 |
+
},
|
| 536 |
+
"labels": [
|
| 537 |
+
"O",
|
| 538 |
+
"B-doc",
|
| 539 |
+
"I-doc",
|
| 540 |
+
"E-doc",
|
| 541 |
+
"S-doc",
|
| 542 |
+
"B-d",
|
| 543 |
+
"I-d",
|
| 544 |
+
"E-d",
|
| 545 |
+
"S-d",
|
| 546 |
+
"B-v",
|
| 547 |
+
"I-v",
|
| 548 |
+
"E-v",
|
| 549 |
+
"S-v",
|
| 550 |
+
"B-meet",
|
| 551 |
+
"I-meet",
|
| 552 |
+
"E-meet",
|
| 553 |
+
"S-meet",
|
| 554 |
+
"B-w",
|
| 555 |
+
"I-w",
|
| 556 |
+
"E-w",
|
| 557 |
+
"S-w",
|
| 558 |
+
"B-conj",
|
| 559 |
+
"I-conj",
|
| 560 |
+
"E-conj",
|
| 561 |
+
"S-conj",
|
| 562 |
+
"B-group",
|
| 563 |
+
"I-group",
|
| 564 |
+
"E-group",
|
| 565 |
+
"S-group",
|
| 566 |
+
"B-env",
|
| 567 |
+
"I-env",
|
| 568 |
+
"E-env",
|
| 569 |
+
"S-env",
|
| 570 |
+
"B-time",
|
| 571 |
+
"I-time",
|
| 572 |
+
"E-time",
|
| 573 |
+
"S-time",
|
| 574 |
+
"B-u",
|
| 575 |
+
"I-u",
|
| 576 |
+
"E-u",
|
| 577 |
+
"S-u",
|
| 578 |
+
"B-gov",
|
| 579 |
+
"I-gov",
|
| 580 |
+
"E-gov",
|
| 581 |
+
"S-gov",
|
| 582 |
+
"B-act",
|
| 583 |
+
"I-act",
|
| 584 |
+
"E-act",
|
| 585 |
+
"S-act",
|
| 586 |
+
"B-a",
|
| 587 |
+
"I-a",
|
| 588 |
+
"E-a",
|
| 589 |
+
"S-a",
|
| 590 |
+
"B-n",
|
| 591 |
+
"I-n",
|
| 592 |
+
"E-n",
|
| 593 |
+
"S-n",
|
| 594 |
+
"B-pro",
|
| 595 |
+
"I-pro",
|
| 596 |
+
"E-pro",
|
| 597 |
+
"S-pro",
|
| 598 |
+
"B-loc",
|
| 599 |
+
"I-loc",
|
| 600 |
+
"E-loc",
|
| 601 |
+
"S-loc",
|
| 602 |
+
"B-med",
|
| 603 |
+
"I-med",
|
| 604 |
+
"E-med",
|
| 605 |
+
"S-med",
|
| 606 |
+
"B-vn",
|
| 607 |
+
"I-vn",
|
| 608 |
+
"E-vn",
|
| 609 |
+
"S-vn",
|
| 610 |
+
"B-event",
|
| 611 |
+
"I-event",
|
| 612 |
+
"E-event",
|
| 613 |
+
"S-event",
|
| 614 |
+
"B-per",
|
| 615 |
+
"I-per",
|
| 616 |
+
"E-per",
|
| 617 |
+
"S-per",
|
| 618 |
+
"B-c",
|
| 619 |
+
"I-c",
|
| 620 |
+
"E-c",
|
| 621 |
+
"S-c",
|
| 622 |
+
"B-m",
|
| 623 |
+
"I-m",
|
| 624 |
+
"E-m",
|
| 625 |
+
"S-m",
|
| 626 |
+
"B-phe",
|
| 627 |
+
"I-phe",
|
| 628 |
+
"E-phe",
|
| 629 |
+
"S-phe",
|
| 630 |
+
"B-pol",
|
| 631 |
+
"I-pol",
|
| 632 |
+
"E-pol",
|
| 633 |
+
"S-pol",
|
| 634 |
+
"B-f",
|
| 635 |
+
"I-f",
|
| 636 |
+
"E-f",
|
| 637 |
+
"S-f",
|
| 638 |
+
"B-p",
|
| 639 |
+
"I-p",
|
| 640 |
+
"E-p",
|
| 641 |
+
"S-p",
|
| 642 |
+
"B-com",
|
| 643 |
+
"I-com",
|
| 644 |
+
"E-com",
|
| 645 |
+
"S-com",
|
| 646 |
+
"B-policy",
|
| 647 |
+
"I-policy",
|
| 648 |
+
"E-policy",
|
| 649 |
+
"S-policy",
|
| 650 |
+
"B-ord",
|
| 651 |
+
"I-ord",
|
| 652 |
+
"E-ord",
|
| 653 |
+
"S-ord",
|
| 654 |
+
"B-q",
|
| 655 |
+
"I-q",
|
| 656 |
+
"E-q",
|
| 657 |
+
"S-q",
|
| 658 |
+
"B-r",
|
| 659 |
+
"I-r",
|
| 660 |
+
"E-r",
|
| 661 |
+
"S-r",
|
| 662 |
+
"B-desease",
|
| 663 |
+
"I-desease",
|
| 664 |
+
"E-desease",
|
| 665 |
+
"S-desease",
|
| 666 |
+
"B-b",
|
| 667 |
+
"I-b",
|
| 668 |
+
"E-b",
|
| 669 |
+
"S-b",
|
| 670 |
+
"B-ins",
|
| 671 |
+
"I-ins",
|
| 672 |
+
"E-ins",
|
| 673 |
+
"S-ins",
|
| 674 |
+
"B-means",
|
| 675 |
+
"I-means",
|
| 676 |
+
"E-means",
|
| 677 |
+
"S-means",
|
| 678 |
+
"B-org",
|
| 679 |
+
"I-org",
|
| 680 |
+
"E-org",
|
| 681 |
+
"S-org",
|
| 682 |
+
"B-microbe",
|
| 683 |
+
"I-microbe",
|
| 684 |
+
"E-microbe",
|
| 685 |
+
"S-microbe",
|
| 686 |
+
"B-hy",
|
| 687 |
+
"I-hy",
|
| 688 |
+
"E-hy",
|
| 689 |
+
"S-hy",
|
| 690 |
+
"B-plant",
|
| 691 |
+
"I-plant",
|
| 692 |
+
"E-plant",
|
| 693 |
+
"S-plant",
|
| 694 |
+
"B-xc",
|
| 695 |
+
"I-xc",
|
| 696 |
+
"E-xc",
|
| 697 |
+
"S-xc",
|
| 698 |
+
"B-code",
|
| 699 |
+
"I-code",
|
| 700 |
+
"E-code",
|
| 701 |
+
"S-code",
|
| 702 |
+
"B-animal",
|
| 703 |
+
"I-animal",
|
| 704 |
+
"E-animal",
|
| 705 |
+
"S-animal"
|
| 706 |
+
],
|
| 707 |
+
"layer_norm_eps": 1e-12,
|
| 708 |
+
"liscence": "Apache Lisence",
|
| 709 |
+
"max_length": 510,
|
| 710 |
+
"max_position_embeddings": 512,
|
| 711 |
+
"model_type": "bert",
|
| 712 |
+
"ner_encoding": "BIOES",
|
| 713 |
+
"num_attention_heads": 12,
|
| 714 |
+
"num_entities": 42,
|
| 715 |
+
"num_hidden_layers": 12,
|
| 716 |
+
"num_test_texts": 0,
|
| 717 |
+
"num_train_texts": 3441,
|
| 718 |
+
"num_valid_texts": 408,
|
| 719 |
+
"output_past": true,
|
| 720 |
+
"package": "envtext",
|
| 721 |
+
"pad_token_id": 0,
|
| 722 |
+
"pooler_fc_size": 768,
|
| 723 |
+
"pooler_num_attention_heads": 12,
|
| 724 |
+
"pooler_num_fc_layers": 3,
|
| 725 |
+
"pooler_size_per_head": 128,
|
| 726 |
+
"pooler_type": "first_token_transform",
|
| 727 |
+
"position_embedding_type": "absolute",
|
| 728 |
+
"resampling": true,
|
| 729 |
+
"resampling_ratio": 4,
|
| 730 |
+
"torch_dtype": "float32",
|
| 731 |
+
"transformers_version": "4.22.1",
|
| 732 |
+
"type_vocab_size": 2,
|
| 733 |
+
"use_cache": true,
|
| 734 |
+
"visualizer": "entity",
|
| 735 |
+
"vocab_size": 21128
|
| 736 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c8e84c835bd4e78c9908942bf743eae32b0527048dadc39771887e111f876c6b
|
| 3 |
+
size 448096037
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "[CLS]",
|
| 3 |
+
"mask_token": "[MASK]",
|
| 4 |
+
"pad_token": "[PAD]",
|
| 5 |
+
"sep_token": "[SEP]",
|
| 6 |
+
"unk_token": "[UNK]"
|
| 7 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "[CLS]",
|
| 3 |
+
"do_basic_tokenize": true,
|
| 4 |
+
"do_lower_case": true,
|
| 5 |
+
"mask_token": "[MASK]",
|
| 6 |
+
"name_or_path": "../models/envbert/",
|
| 7 |
+
"never_split": null,
|
| 8 |
+
"pad_token": "[PAD]",
|
| 9 |
+
"sep_token": "[SEP]",
|
| 10 |
+
"special_tokens_map_file": null,
|
| 11 |
+
"strip_accents": null,
|
| 12 |
+
"tokenize_chinese_chars": true,
|
| 13 |
+
"tokenizer_class": "BertTokenizer",
|
| 14 |
+
"unk_token": "[UNK]"
|
| 15 |
+
}
|
vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|