Keerthikl commited on
Commit
6669a2b
·
verified ·
1 Parent(s): 8e628d7

Upload 5 files

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
config.json ADDED
@@ -0,0 +1,406 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "ModernBertForTokenClassification"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 2,
8
+ "classifier_activation": "gelu",
9
+ "classifier_bias": false,
10
+ "classifier_dropout": 0.0,
11
+ "classifier_pooling": "mean",
12
+ "cls_token_id": 1,
13
+ "decoder_bias": true,
14
+ "deterministic_flash_attn": false,
15
+ "dtype": "float32",
16
+ "embedding_dropout": 0.0,
17
+ "eos_token_id": 1,
18
+ "global_attn_every_n_layers": 3,
19
+ "gradient_checkpointing": false,
20
+ "hidden_activation": "gelu",
21
+ "hidden_size": 768,
22
+ "id2label": {
23
+ "0": "O",
24
+ "1": "B-ACCOUNT_BALANCE",
25
+ "2": "I-ACCOUNT_BALANCE",
26
+ "3": "B-ACCOUNT_NUMBER",
27
+ "4": "I-ACCOUNT_NUMBER",
28
+ "5": "B-ADVERTISING_ID",
29
+ "6": "I-ADVERTISING_ID",
30
+ "7": "B-AGE",
31
+ "8": "I-AGE",
32
+ "9": "B-API_KEY",
33
+ "10": "I-API_KEY",
34
+ "11": "B-ASSET_INFORMATION",
35
+ "12": "I-ASSET_INFORMATION",
36
+ "13": "B-BANK_ACCOUNT_STATEMENT",
37
+ "14": "I-BANK_ACCOUNT_STATEMENT",
38
+ "15": "B-BANK_ROUTING_NUMBER",
39
+ "16": "I-BANK_ROUTING_NUMBER",
40
+ "17": "B-BENEFICIARY_NUMBER",
41
+ "18": "I-BENEFICIARY_NUMBER",
42
+ "19": "B-BIOMETRIC_IDENTIFIER",
43
+ "20": "I-BIOMETRIC_IDENTIFIER",
44
+ "21": "B-BLOOD_TYPE",
45
+ "22": "I-BLOOD_TYPE",
46
+ "23": "B-BROWSING_HISTORY_BEHAVIOUR_DATA",
47
+ "24": "I-BROWSING_HISTORY_BEHAVIOUR_DATA",
48
+ "25": "B-CHILD_INFORMATION",
49
+ "26": "I-CHILD_INFORMATION",
50
+ "27": "B-COOKIE",
51
+ "28": "I-COOKIE",
52
+ "29": "B-COORDINATES",
53
+ "30": "I-COORDINATES",
54
+ "31": "B-CREDIT_CARD",
55
+ "32": "I-CREDIT_CARD",
56
+ "33": "B-CREDIT_HISTORY",
57
+ "34": "I-CREDIT_HISTORY",
58
+ "35": "B-CREDIT_SCORE",
59
+ "36": "I-CREDIT_SCORE",
60
+ "37": "B-CRIMINAL_LEGAL_RECORDS",
61
+ "38": "I-CRIMINAL_LEGAL_RECORDS",
62
+ "39": "B-CUSTOMER_ID",
63
+ "40": "I-CUSTOMER_ID",
64
+ "41": "B-CVV",
65
+ "42": "I-CVV",
66
+ "43": "B-DATE_TIME",
67
+ "44": "I-DATE_TIME",
68
+ "45": "B-DEBT_INFORMATION",
69
+ "46": "I-DEBT_INFORMATION",
70
+ "47": "B-DEVICE_ID",
71
+ "48": "I-DEVICE_ID",
72
+ "49": "B-DEVICE_IDENTIFIER",
73
+ "50": "I-DEVICE_IDENTIFIER",
74
+ "51": "B-DOB",
75
+ "52": "I-DOB",
76
+ "53": "B-DOMAIN_NAME",
77
+ "54": "I-DOMAIN_NAME",
78
+ "55": "B-EDUCATION_LEVEL",
79
+ "56": "I-EDUCATION_LEVEL",
80
+ "57": "B-EMAIL_ADDRESS",
81
+ "58": "I-EMAIL_ADDRESS",
82
+ "59": "B-EMPLOYEE_ID",
83
+ "60": "I-EMPLOYEE_ID",
84
+ "61": "B-EMPLOYMENT_STATUS",
85
+ "62": "I-EMPLOYMENT_STATUS",
86
+ "63": "B-FAX_NUMBER",
87
+ "64": "I-FAX_NUMBER",
88
+ "65": "B-FINANCIAL_STATEMENTS",
89
+ "66": "I-FINANCIAL_STATEMENTS",
90
+ "67": "B-GENDER",
91
+ "68": "I-GENDER",
92
+ "69": "B-GENETIC_DATA",
93
+ "70": "I-GENETIC_DATA",
94
+ "71": "B-GPE",
95
+ "72": "I-GPE",
96
+ "73": "B-HEALTH_INSURANCE_ID",
97
+ "74": "I-HEALTH_INSURANCE_ID",
98
+ "75": "B-HEALTH_INSURANCE_ID_NUMBER",
99
+ "76": "I-HEALTH_INSURANCE_ID_NUMBER",
100
+ "77": "B-HEALTH_PLAN_BENEFICIARY_NUMBER",
101
+ "78": "I-HEALTH_PLAN_BENEFICIARY_NUMBER",
102
+ "79": "B-IBAN_CODE",
103
+ "80": "I-IBAN_CODE",
104
+ "81": "B-INCOME_INFORMATION",
105
+ "82": "I-INCOME_INFORMATION",
106
+ "83": "B-IP_ADDRESS",
107
+ "84": "I-IP_ADDRESS",
108
+ "85": "B-LANGUAGE",
109
+ "86": "I-LANGUAGE",
110
+ "87": "B-LICENSE_NUMBER",
111
+ "88": "I-LICENSE_NUMBER",
112
+ "89": "B-MAC_ADDRESS",
113
+ "90": "I-MAC_ADDRESS",
114
+ "91": "B-MARITAL_RELATIONSHIP_STATUS",
115
+ "92": "I-MARITAL_RELATIONSHIP_STATUS",
116
+ "93": "B-MEDIA_RECORDING",
117
+ "94": "I-MEDIA_RECORDING",
118
+ "95": "B-MEDICAL_RECORD_NUMBER",
119
+ "96": "I-MEDICAL_RECORD_NUMBER",
120
+ "97": "B-NATIONAL_ID",
121
+ "98": "I-NATIONAL_ID",
122
+ "99": "B-NETWORK_ACTIVITY_LOGS",
123
+ "100": "I-NETWORK_ACTIVITY_LOGS",
124
+ "101": "B-NET_WORTH_INFORMATION",
125
+ "102": "I-NET_WORTH_INFORMATION",
126
+ "103": "B-NRP",
127
+ "104": "I-NRP",
128
+ "105": "B-OCCUPATION",
129
+ "106": "I-OCCUPATION",
130
+ "107": "B-ORGANIZATION",
131
+ "108": "I-ORGANIZATION",
132
+ "109": "B-PAN_NUMBER",
133
+ "110": "I-PAN_NUMBER",
134
+ "111": "B-PARENT_INFORMATION",
135
+ "112": "I-PARENT_INFORMATION",
136
+ "113": "B-PASSPORT_NUMBER",
137
+ "114": "I-PASSPORT_NUMBER",
138
+ "115": "B-PAYMENT_HISTORY",
139
+ "116": "I-PAYMENT_HISTORY",
140
+ "117": "B-PERSON",
141
+ "118": "I-PERSON",
142
+ "119": "B-PHONE_NUMBER",
143
+ "120": "I-PHONE_NUMBER",
144
+ "121": "B-PIN",
145
+ "122": "I-PIN",
146
+ "123": "B-PRIVATE_DATE",
147
+ "124": "I-PRIVATE_DATE",
148
+ "125": "B-SECRET",
149
+ "126": "I-SECRET",
150
+ "127": "B-SEXUALITY",
151
+ "128": "I-SEXUALITY",
152
+ "129": "B-SOCIAL_MEDIA_POSTS",
153
+ "130": "I-SOCIAL_MEDIA_POSTS",
154
+ "131": "B-STREET_ADDRESS",
155
+ "132": "I-STREET_ADDRESS",
156
+ "133": "B-STUDENT_ID",
157
+ "134": "I-STUDENT_ID",
158
+ "135": "B-SWIFT_CODE",
159
+ "136": "I-SWIFT_CODE",
160
+ "137": "B-TAX_ID",
161
+ "138": "I-TAX_ID",
162
+ "139": "B-TIME",
163
+ "140": "I-TIME",
164
+ "141": "B-TITLE",
165
+ "142": "I-TITLE",
166
+ "143": "B-TRADE_UNION_MEMBERSHIP",
167
+ "144": "I-TRADE_UNION_MEMBERSHIP",
168
+ "145": "B-TRANSACTION_HISTORY",
169
+ "146": "I-TRANSACTION_HISTORY",
170
+ "147": "B-UNIQUE_ID",
171
+ "148": "I-UNIQUE_ID",
172
+ "149": "B-USERNAME",
173
+ "150": "I-USERNAME",
174
+ "151": "B-US_DRIVER_LICENSE",
175
+ "152": "I-US_DRIVER_LICENSE",
176
+ "153": "B-US_SSN",
177
+ "154": "I-US_SSN",
178
+ "155": "B-VEHICLE_IDENTIFIER",
179
+ "156": "I-VEHICLE_IDENTIFIER",
180
+ "157": "B-WIRE_TRANSFER_DETAILS",
181
+ "158": "I-WIRE_TRANSFER_DETAILS",
182
+ "159": "B-ZIP_CODE",
183
+ "160": "I-ZIP_CODE"
184
+ },
185
+ "initializer_cutoff_factor": 2.0,
186
+ "initializer_range": 0.02,
187
+ "intermediate_size": 1152,
188
+ "label2id": {
189
+ "B-ACCOUNT_BALANCE": 1,
190
+ "B-ACCOUNT_NUMBER": 3,
191
+ "B-ADVERTISING_ID": 5,
192
+ "B-AGE": 7,
193
+ "B-API_KEY": 9,
194
+ "B-ASSET_INFORMATION": 11,
195
+ "B-BANK_ACCOUNT_STATEMENT": 13,
196
+ "B-BANK_ROUTING_NUMBER": 15,
197
+ "B-BENEFICIARY_NUMBER": 17,
198
+ "B-BIOMETRIC_IDENTIFIER": 19,
199
+ "B-BLOOD_TYPE": 21,
200
+ "B-BROWSING_HISTORY_BEHAVIOUR_DATA": 23,
201
+ "B-CHILD_INFORMATION": 25,
202
+ "B-COOKIE": 27,
203
+ "B-COORDINATES": 29,
204
+ "B-CREDIT_CARD": 31,
205
+ "B-CREDIT_HISTORY": 33,
206
+ "B-CREDIT_SCORE": 35,
207
+ "B-CRIMINAL_LEGAL_RECORDS": 37,
208
+ "B-CUSTOMER_ID": 39,
209
+ "B-CVV": 41,
210
+ "B-DATE_TIME": 43,
211
+ "B-DEBT_INFORMATION": 45,
212
+ "B-DEVICE_ID": 47,
213
+ "B-DEVICE_IDENTIFIER": 49,
214
+ "B-DOB": 51,
215
+ "B-DOMAIN_NAME": 53,
216
+ "B-EDUCATION_LEVEL": 55,
217
+ "B-EMAIL_ADDRESS": 57,
218
+ "B-EMPLOYEE_ID": 59,
219
+ "B-EMPLOYMENT_STATUS": 61,
220
+ "B-FAX_NUMBER": 63,
221
+ "B-FINANCIAL_STATEMENTS": 65,
222
+ "B-GENDER": 67,
223
+ "B-GENETIC_DATA": 69,
224
+ "B-GPE": 71,
225
+ "B-HEALTH_INSURANCE_ID": 73,
226
+ "B-HEALTH_INSURANCE_ID_NUMBER": 75,
227
+ "B-HEALTH_PLAN_BENEFICIARY_NUMBER": 77,
228
+ "B-IBAN_CODE": 79,
229
+ "B-INCOME_INFORMATION": 81,
230
+ "B-IP_ADDRESS": 83,
231
+ "B-LANGUAGE": 85,
232
+ "B-LICENSE_NUMBER": 87,
233
+ "B-MAC_ADDRESS": 89,
234
+ "B-MARITAL_RELATIONSHIP_STATUS": 91,
235
+ "B-MEDIA_RECORDING": 93,
236
+ "B-MEDICAL_RECORD_NUMBER": 95,
237
+ "B-NATIONAL_ID": 97,
238
+ "B-NETWORK_ACTIVITY_LOGS": 99,
239
+ "B-NET_WORTH_INFORMATION": 101,
240
+ "B-NRP": 103,
241
+ "B-OCCUPATION": 105,
242
+ "B-ORGANIZATION": 107,
243
+ "B-PAN_NUMBER": 109,
244
+ "B-PARENT_INFORMATION": 111,
245
+ "B-PASSPORT_NUMBER": 113,
246
+ "B-PAYMENT_HISTORY": 115,
247
+ "B-PERSON": 117,
248
+ "B-PHONE_NUMBER": 119,
249
+ "B-PIN": 121,
250
+ "B-PRIVATE_DATE": 123,
251
+ "B-SECRET": 125,
252
+ "B-SEXUALITY": 127,
253
+ "B-SOCIAL_MEDIA_POSTS": 129,
254
+ "B-STREET_ADDRESS": 131,
255
+ "B-STUDENT_ID": 133,
256
+ "B-SWIFT_CODE": 135,
257
+ "B-TAX_ID": 137,
258
+ "B-TIME": 139,
259
+ "B-TITLE": 141,
260
+ "B-TRADE_UNION_MEMBERSHIP": 143,
261
+ "B-TRANSACTION_HISTORY": 145,
262
+ "B-UNIQUE_ID": 147,
263
+ "B-USERNAME": 149,
264
+ "B-US_DRIVER_LICENSE": 151,
265
+ "B-US_SSN": 153,
266
+ "B-VEHICLE_IDENTIFIER": 155,
267
+ "B-WIRE_TRANSFER_DETAILS": 157,
268
+ "B-ZIP_CODE": 159,
269
+ "I-ACCOUNT_BALANCE": 2,
270
+ "I-ACCOUNT_NUMBER": 4,
271
+ "I-ADVERTISING_ID": 6,
272
+ "I-AGE": 8,
273
+ "I-API_KEY": 10,
274
+ "I-ASSET_INFORMATION": 12,
275
+ "I-BANK_ACCOUNT_STATEMENT": 14,
276
+ "I-BANK_ROUTING_NUMBER": 16,
277
+ "I-BENEFICIARY_NUMBER": 18,
278
+ "I-BIOMETRIC_IDENTIFIER": 20,
279
+ "I-BLOOD_TYPE": 22,
280
+ "I-BROWSING_HISTORY_BEHAVIOUR_DATA": 24,
281
+ "I-CHILD_INFORMATION": 26,
282
+ "I-COOKIE": 28,
283
+ "I-COORDINATES": 30,
284
+ "I-CREDIT_CARD": 32,
285
+ "I-CREDIT_HISTORY": 34,
286
+ "I-CREDIT_SCORE": 36,
287
+ "I-CRIMINAL_LEGAL_RECORDS": 38,
288
+ "I-CUSTOMER_ID": 40,
289
+ "I-CVV": 42,
290
+ "I-DATE_TIME": 44,
291
+ "I-DEBT_INFORMATION": 46,
292
+ "I-DEVICE_ID": 48,
293
+ "I-DEVICE_IDENTIFIER": 50,
294
+ "I-DOB": 52,
295
+ "I-DOMAIN_NAME": 54,
296
+ "I-EDUCATION_LEVEL": 56,
297
+ "I-EMAIL_ADDRESS": 58,
298
+ "I-EMPLOYEE_ID": 60,
299
+ "I-EMPLOYMENT_STATUS": 62,
300
+ "I-FAX_NUMBER": 64,
301
+ "I-FINANCIAL_STATEMENTS": 66,
302
+ "I-GENDER": 68,
303
+ "I-GENETIC_DATA": 70,
304
+ "I-GPE": 72,
305
+ "I-HEALTH_INSURANCE_ID": 74,
306
+ "I-HEALTH_INSURANCE_ID_NUMBER": 76,
307
+ "I-HEALTH_PLAN_BENEFICIARY_NUMBER": 78,
308
+ "I-IBAN_CODE": 80,
309
+ "I-INCOME_INFORMATION": 82,
310
+ "I-IP_ADDRESS": 84,
311
+ "I-LANGUAGE": 86,
312
+ "I-LICENSE_NUMBER": 88,
313
+ "I-MAC_ADDRESS": 90,
314
+ "I-MARITAL_RELATIONSHIP_STATUS": 92,
315
+ "I-MEDIA_RECORDING": 94,
316
+ "I-MEDICAL_RECORD_NUMBER": 96,
317
+ "I-NATIONAL_ID": 98,
318
+ "I-NETWORK_ACTIVITY_LOGS": 100,
319
+ "I-NET_WORTH_INFORMATION": 102,
320
+ "I-NRP": 104,
321
+ "I-OCCUPATION": 106,
322
+ "I-ORGANIZATION": 108,
323
+ "I-PAN_NUMBER": 110,
324
+ "I-PARENT_INFORMATION": 112,
325
+ "I-PASSPORT_NUMBER": 114,
326
+ "I-PAYMENT_HISTORY": 116,
327
+ "I-PERSON": 118,
328
+ "I-PHONE_NUMBER": 120,
329
+ "I-PIN": 122,
330
+ "I-PRIVATE_DATE": 124,
331
+ "I-SECRET": 126,
332
+ "I-SEXUALITY": 128,
333
+ "I-SOCIAL_MEDIA_POSTS": 130,
334
+ "I-STREET_ADDRESS": 132,
335
+ "I-STUDENT_ID": 134,
336
+ "I-SWIFT_CODE": 136,
337
+ "I-TAX_ID": 138,
338
+ "I-TIME": 140,
339
+ "I-TITLE": 142,
340
+ "I-TRADE_UNION_MEMBERSHIP": 144,
341
+ "I-TRANSACTION_HISTORY": 146,
342
+ "I-UNIQUE_ID": 148,
343
+ "I-USERNAME": 150,
344
+ "I-US_DRIVER_LICENSE": 152,
345
+ "I-US_SSN": 154,
346
+ "I-VEHICLE_IDENTIFIER": 156,
347
+ "I-WIRE_TRANSFER_DETAILS": 158,
348
+ "I-ZIP_CODE": 160,
349
+ "O": 0
350
+ },
351
+ "layer_norm_eps": 1e-05,
352
+ "layer_types": [
353
+ "full_attention",
354
+ "sliding_attention",
355
+ "sliding_attention",
356
+ "full_attention",
357
+ "sliding_attention",
358
+ "sliding_attention",
359
+ "full_attention",
360
+ "sliding_attention",
361
+ "sliding_attention",
362
+ "full_attention",
363
+ "sliding_attention",
364
+ "sliding_attention",
365
+ "full_attention",
366
+ "sliding_attention",
367
+ "sliding_attention",
368
+ "full_attention",
369
+ "sliding_attention",
370
+ "sliding_attention",
371
+ "full_attention",
372
+ "sliding_attention",
373
+ "sliding_attention",
374
+ "full_attention"
375
+ ],
376
+ "local_attention": 128,
377
+ "mask_token_id": 4,
378
+ "max_position_embeddings": 32768,
379
+ "mlp_bias": false,
380
+ "mlp_dropout": 0.0,
381
+ "model_type": "modernbert",
382
+ "norm_bias": false,
383
+ "norm_eps": 1e-05,
384
+ "num_attention_heads": 12,
385
+ "num_hidden_layers": 22,
386
+ "pad_token_id": 0,
387
+ "position_embedding_type": "sans_pos",
388
+ "repad_logits_with_grad": false,
389
+ "rope_parameters": {
390
+ "full_attention": {
391
+ "rope_theta": 160000,
392
+ "rope_type": "default"
393
+ },
394
+ "sliding_attention": {
395
+ "rope_theta": 160000,
396
+ "rope_type": "default"
397
+ }
398
+ },
399
+ "sep_token_id": 1,
400
+ "sparse_pred_ignore_index": -100,
401
+ "sparse_prediction": false,
402
+ "tie_word_embeddings": true,
403
+ "transformers_version": "5.5.0",
404
+ "use_cache": false,
405
+ "vocab_size": 256000
406
+ }
label_mapping.json ADDED
@@ -0,0 +1,497 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "label2id": {
3
+ "O": 0,
4
+ "B-ACCOUNT_BALANCE": 1,
5
+ "I-ACCOUNT_BALANCE": 2,
6
+ "B-ACCOUNT_NUMBER": 3,
7
+ "I-ACCOUNT_NUMBER": 4,
8
+ "B-ADVERTISING_ID": 5,
9
+ "I-ADVERTISING_ID": 6,
10
+ "B-AGE": 7,
11
+ "I-AGE": 8,
12
+ "B-API_KEY": 9,
13
+ "I-API_KEY": 10,
14
+ "B-ASSET_INFORMATION": 11,
15
+ "I-ASSET_INFORMATION": 12,
16
+ "B-BANK_ACCOUNT_STATEMENT": 13,
17
+ "I-BANK_ACCOUNT_STATEMENT": 14,
18
+ "B-BANK_ROUTING_NUMBER": 15,
19
+ "I-BANK_ROUTING_NUMBER": 16,
20
+ "B-BENEFICIARY_NUMBER": 17,
21
+ "I-BENEFICIARY_NUMBER": 18,
22
+ "B-BIOMETRIC_IDENTIFIER": 19,
23
+ "I-BIOMETRIC_IDENTIFIER": 20,
24
+ "B-BLOOD_TYPE": 21,
25
+ "I-BLOOD_TYPE": 22,
26
+ "B-BROWSING_HISTORY_BEHAVIOUR_DATA": 23,
27
+ "I-BROWSING_HISTORY_BEHAVIOUR_DATA": 24,
28
+ "B-CHILD_INFORMATION": 25,
29
+ "I-CHILD_INFORMATION": 26,
30
+ "B-COOKIE": 27,
31
+ "I-COOKIE": 28,
32
+ "B-COORDINATES": 29,
33
+ "I-COORDINATES": 30,
34
+ "B-CREDIT_CARD": 31,
35
+ "I-CREDIT_CARD": 32,
36
+ "B-CREDIT_HISTORY": 33,
37
+ "I-CREDIT_HISTORY": 34,
38
+ "B-CREDIT_SCORE": 35,
39
+ "I-CREDIT_SCORE": 36,
40
+ "B-CRIMINAL_LEGAL_RECORDS": 37,
41
+ "I-CRIMINAL_LEGAL_RECORDS": 38,
42
+ "B-CUSTOMER_ID": 39,
43
+ "I-CUSTOMER_ID": 40,
44
+ "B-CVV": 41,
45
+ "I-CVV": 42,
46
+ "B-DATE_TIME": 43,
47
+ "I-DATE_TIME": 44,
48
+ "B-DEBT_INFORMATION": 45,
49
+ "I-DEBT_INFORMATION": 46,
50
+ "B-DEVICE_ID": 47,
51
+ "I-DEVICE_ID": 48,
52
+ "B-DEVICE_IDENTIFIER": 49,
53
+ "I-DEVICE_IDENTIFIER": 50,
54
+ "B-DOB": 51,
55
+ "I-DOB": 52,
56
+ "B-DOMAIN_NAME": 53,
57
+ "I-DOMAIN_NAME": 54,
58
+ "B-EDUCATION_LEVEL": 55,
59
+ "I-EDUCATION_LEVEL": 56,
60
+ "B-EMAIL_ADDRESS": 57,
61
+ "I-EMAIL_ADDRESS": 58,
62
+ "B-EMPLOYEE_ID": 59,
63
+ "I-EMPLOYEE_ID": 60,
64
+ "B-EMPLOYMENT_STATUS": 61,
65
+ "I-EMPLOYMENT_STATUS": 62,
66
+ "B-FAX_NUMBER": 63,
67
+ "I-FAX_NUMBER": 64,
68
+ "B-FINANCIAL_STATEMENTS": 65,
69
+ "I-FINANCIAL_STATEMENTS": 66,
70
+ "B-GENDER": 67,
71
+ "I-GENDER": 68,
72
+ "B-GENETIC_DATA": 69,
73
+ "I-GENETIC_DATA": 70,
74
+ "B-GPE": 71,
75
+ "I-GPE": 72,
76
+ "B-HEALTH_INSURANCE_ID": 73,
77
+ "I-HEALTH_INSURANCE_ID": 74,
78
+ "B-HEALTH_INSURANCE_ID_NUMBER": 75,
79
+ "I-HEALTH_INSURANCE_ID_NUMBER": 76,
80
+ "B-HEALTH_PLAN_BENEFICIARY_NUMBER": 77,
81
+ "I-HEALTH_PLAN_BENEFICIARY_NUMBER": 78,
82
+ "B-IBAN_CODE": 79,
83
+ "I-IBAN_CODE": 80,
84
+ "B-INCOME_INFORMATION": 81,
85
+ "I-INCOME_INFORMATION": 82,
86
+ "B-IP_ADDRESS": 83,
87
+ "I-IP_ADDRESS": 84,
88
+ "B-LANGUAGE": 85,
89
+ "I-LANGUAGE": 86,
90
+ "B-LICENSE_NUMBER": 87,
91
+ "I-LICENSE_NUMBER": 88,
92
+ "B-MAC_ADDRESS": 89,
93
+ "I-MAC_ADDRESS": 90,
94
+ "B-MARITAL_RELATIONSHIP_STATUS": 91,
95
+ "I-MARITAL_RELATIONSHIP_STATUS": 92,
96
+ "B-MEDIA_RECORDING": 93,
97
+ "I-MEDIA_RECORDING": 94,
98
+ "B-MEDICAL_RECORD_NUMBER": 95,
99
+ "I-MEDICAL_RECORD_NUMBER": 96,
100
+ "B-NATIONAL_ID": 97,
101
+ "I-NATIONAL_ID": 98,
102
+ "B-NETWORK_ACTIVITY_LOGS": 99,
103
+ "I-NETWORK_ACTIVITY_LOGS": 100,
104
+ "B-NET_WORTH_INFORMATION": 101,
105
+ "I-NET_WORTH_INFORMATION": 102,
106
+ "B-NRP": 103,
107
+ "I-NRP": 104,
108
+ "B-OCCUPATION": 105,
109
+ "I-OCCUPATION": 106,
110
+ "B-ORGANIZATION": 107,
111
+ "I-ORGANIZATION": 108,
112
+ "B-PAN_NUMBER": 109,
113
+ "I-PAN_NUMBER": 110,
114
+ "B-PARENT_INFORMATION": 111,
115
+ "I-PARENT_INFORMATION": 112,
116
+ "B-PASSPORT_NUMBER": 113,
117
+ "I-PASSPORT_NUMBER": 114,
118
+ "B-PAYMENT_HISTORY": 115,
119
+ "I-PAYMENT_HISTORY": 116,
120
+ "B-PERSON": 117,
121
+ "I-PERSON": 118,
122
+ "B-PHONE_NUMBER": 119,
123
+ "I-PHONE_NUMBER": 120,
124
+ "B-PIN": 121,
125
+ "I-PIN": 122,
126
+ "B-PRIVATE_DATE": 123,
127
+ "I-PRIVATE_DATE": 124,
128
+ "B-SECRET": 125,
129
+ "I-SECRET": 126,
130
+ "B-SEXUALITY": 127,
131
+ "I-SEXUALITY": 128,
132
+ "B-SOCIAL_MEDIA_POSTS": 129,
133
+ "I-SOCIAL_MEDIA_POSTS": 130,
134
+ "B-STREET_ADDRESS": 131,
135
+ "I-STREET_ADDRESS": 132,
136
+ "B-STUDENT_ID": 133,
137
+ "I-STUDENT_ID": 134,
138
+ "B-SWIFT_CODE": 135,
139
+ "I-SWIFT_CODE": 136,
140
+ "B-TAX_ID": 137,
141
+ "I-TAX_ID": 138,
142
+ "B-TIME": 139,
143
+ "I-TIME": 140,
144
+ "B-TITLE": 141,
145
+ "I-TITLE": 142,
146
+ "B-TRADE_UNION_MEMBERSHIP": 143,
147
+ "I-TRADE_UNION_MEMBERSHIP": 144,
148
+ "B-TRANSACTION_HISTORY": 145,
149
+ "I-TRANSACTION_HISTORY": 146,
150
+ "B-UNIQUE_ID": 147,
151
+ "I-UNIQUE_ID": 148,
152
+ "B-USERNAME": 149,
153
+ "I-USERNAME": 150,
154
+ "B-US_DRIVER_LICENSE": 151,
155
+ "I-US_DRIVER_LICENSE": 152,
156
+ "B-US_SSN": 153,
157
+ "I-US_SSN": 154,
158
+ "B-VEHICLE_IDENTIFIER": 155,
159
+ "I-VEHICLE_IDENTIFIER": 156,
160
+ "B-WIRE_TRANSFER_DETAILS": 157,
161
+ "I-WIRE_TRANSFER_DETAILS": 158,
162
+ "B-ZIP_CODE": 159,
163
+ "I-ZIP_CODE": 160
164
+ },
165
+ "id2label": {
166
+ "0": "O",
167
+ "1": "B-ACCOUNT_BALANCE",
168
+ "2": "I-ACCOUNT_BALANCE",
169
+ "3": "B-ACCOUNT_NUMBER",
170
+ "4": "I-ACCOUNT_NUMBER",
171
+ "5": "B-ADVERTISING_ID",
172
+ "6": "I-ADVERTISING_ID",
173
+ "7": "B-AGE",
174
+ "8": "I-AGE",
175
+ "9": "B-API_KEY",
176
+ "10": "I-API_KEY",
177
+ "11": "B-ASSET_INFORMATION",
178
+ "12": "I-ASSET_INFORMATION",
179
+ "13": "B-BANK_ACCOUNT_STATEMENT",
180
+ "14": "I-BANK_ACCOUNT_STATEMENT",
181
+ "15": "B-BANK_ROUTING_NUMBER",
182
+ "16": "I-BANK_ROUTING_NUMBER",
183
+ "17": "B-BENEFICIARY_NUMBER",
184
+ "18": "I-BENEFICIARY_NUMBER",
185
+ "19": "B-BIOMETRIC_IDENTIFIER",
186
+ "20": "I-BIOMETRIC_IDENTIFIER",
187
+ "21": "B-BLOOD_TYPE",
188
+ "22": "I-BLOOD_TYPE",
189
+ "23": "B-BROWSING_HISTORY_BEHAVIOUR_DATA",
190
+ "24": "I-BROWSING_HISTORY_BEHAVIOUR_DATA",
191
+ "25": "B-CHILD_INFORMATION",
192
+ "26": "I-CHILD_INFORMATION",
193
+ "27": "B-COOKIE",
194
+ "28": "I-COOKIE",
195
+ "29": "B-COORDINATES",
196
+ "30": "I-COORDINATES",
197
+ "31": "B-CREDIT_CARD",
198
+ "32": "I-CREDIT_CARD",
199
+ "33": "B-CREDIT_HISTORY",
200
+ "34": "I-CREDIT_HISTORY",
201
+ "35": "B-CREDIT_SCORE",
202
+ "36": "I-CREDIT_SCORE",
203
+ "37": "B-CRIMINAL_LEGAL_RECORDS",
204
+ "38": "I-CRIMINAL_LEGAL_RECORDS",
205
+ "39": "B-CUSTOMER_ID",
206
+ "40": "I-CUSTOMER_ID",
207
+ "41": "B-CVV",
208
+ "42": "I-CVV",
209
+ "43": "B-DATE_TIME",
210
+ "44": "I-DATE_TIME",
211
+ "45": "B-DEBT_INFORMATION",
212
+ "46": "I-DEBT_INFORMATION",
213
+ "47": "B-DEVICE_ID",
214
+ "48": "I-DEVICE_ID",
215
+ "49": "B-DEVICE_IDENTIFIER",
216
+ "50": "I-DEVICE_IDENTIFIER",
217
+ "51": "B-DOB",
218
+ "52": "I-DOB",
219
+ "53": "B-DOMAIN_NAME",
220
+ "54": "I-DOMAIN_NAME",
221
+ "55": "B-EDUCATION_LEVEL",
222
+ "56": "I-EDUCATION_LEVEL",
223
+ "57": "B-EMAIL_ADDRESS",
224
+ "58": "I-EMAIL_ADDRESS",
225
+ "59": "B-EMPLOYEE_ID",
226
+ "60": "I-EMPLOYEE_ID",
227
+ "61": "B-EMPLOYMENT_STATUS",
228
+ "62": "I-EMPLOYMENT_STATUS",
229
+ "63": "B-FAX_NUMBER",
230
+ "64": "I-FAX_NUMBER",
231
+ "65": "B-FINANCIAL_STATEMENTS",
232
+ "66": "I-FINANCIAL_STATEMENTS",
233
+ "67": "B-GENDER",
234
+ "68": "I-GENDER",
235
+ "69": "B-GENETIC_DATA",
236
+ "70": "I-GENETIC_DATA",
237
+ "71": "B-GPE",
238
+ "72": "I-GPE",
239
+ "73": "B-HEALTH_INSURANCE_ID",
240
+ "74": "I-HEALTH_INSURANCE_ID",
241
+ "75": "B-HEALTH_INSURANCE_ID_NUMBER",
242
+ "76": "I-HEALTH_INSURANCE_ID_NUMBER",
243
+ "77": "B-HEALTH_PLAN_BENEFICIARY_NUMBER",
244
+ "78": "I-HEALTH_PLAN_BENEFICIARY_NUMBER",
245
+ "79": "B-IBAN_CODE",
246
+ "80": "I-IBAN_CODE",
247
+ "81": "B-INCOME_INFORMATION",
248
+ "82": "I-INCOME_INFORMATION",
249
+ "83": "B-IP_ADDRESS",
250
+ "84": "I-IP_ADDRESS",
251
+ "85": "B-LANGUAGE",
252
+ "86": "I-LANGUAGE",
253
+ "87": "B-LICENSE_NUMBER",
254
+ "88": "I-LICENSE_NUMBER",
255
+ "89": "B-MAC_ADDRESS",
256
+ "90": "I-MAC_ADDRESS",
257
+ "91": "B-MARITAL_RELATIONSHIP_STATUS",
258
+ "92": "I-MARITAL_RELATIONSHIP_STATUS",
259
+ "93": "B-MEDIA_RECORDING",
260
+ "94": "I-MEDIA_RECORDING",
261
+ "95": "B-MEDICAL_RECORD_NUMBER",
262
+ "96": "I-MEDICAL_RECORD_NUMBER",
263
+ "97": "B-NATIONAL_ID",
264
+ "98": "I-NATIONAL_ID",
265
+ "99": "B-NETWORK_ACTIVITY_LOGS",
266
+ "100": "I-NETWORK_ACTIVITY_LOGS",
267
+ "101": "B-NET_WORTH_INFORMATION",
268
+ "102": "I-NET_WORTH_INFORMATION",
269
+ "103": "B-NRP",
270
+ "104": "I-NRP",
271
+ "105": "B-OCCUPATION",
272
+ "106": "I-OCCUPATION",
273
+ "107": "B-ORGANIZATION",
274
+ "108": "I-ORGANIZATION",
275
+ "109": "B-PAN_NUMBER",
276
+ "110": "I-PAN_NUMBER",
277
+ "111": "B-PARENT_INFORMATION",
278
+ "112": "I-PARENT_INFORMATION",
279
+ "113": "B-PASSPORT_NUMBER",
280
+ "114": "I-PASSPORT_NUMBER",
281
+ "115": "B-PAYMENT_HISTORY",
282
+ "116": "I-PAYMENT_HISTORY",
283
+ "117": "B-PERSON",
284
+ "118": "I-PERSON",
285
+ "119": "B-PHONE_NUMBER",
286
+ "120": "I-PHONE_NUMBER",
287
+ "121": "B-PIN",
288
+ "122": "I-PIN",
289
+ "123": "B-PRIVATE_DATE",
290
+ "124": "I-PRIVATE_DATE",
291
+ "125": "B-SECRET",
292
+ "126": "I-SECRET",
293
+ "127": "B-SEXUALITY",
294
+ "128": "I-SEXUALITY",
295
+ "129": "B-SOCIAL_MEDIA_POSTS",
296
+ "130": "I-SOCIAL_MEDIA_POSTS",
297
+ "131": "B-STREET_ADDRESS",
298
+ "132": "I-STREET_ADDRESS",
299
+ "133": "B-STUDENT_ID",
300
+ "134": "I-STUDENT_ID",
301
+ "135": "B-SWIFT_CODE",
302
+ "136": "I-SWIFT_CODE",
303
+ "137": "B-TAX_ID",
304
+ "138": "I-TAX_ID",
305
+ "139": "B-TIME",
306
+ "140": "I-TIME",
307
+ "141": "B-TITLE",
308
+ "142": "I-TITLE",
309
+ "143": "B-TRADE_UNION_MEMBERSHIP",
310
+ "144": "I-TRADE_UNION_MEMBERSHIP",
311
+ "145": "B-TRANSACTION_HISTORY",
312
+ "146": "I-TRANSACTION_HISTORY",
313
+ "147": "B-UNIQUE_ID",
314
+ "148": "I-UNIQUE_ID",
315
+ "149": "B-USERNAME",
316
+ "150": "I-USERNAME",
317
+ "151": "B-US_DRIVER_LICENSE",
318
+ "152": "I-US_DRIVER_LICENSE",
319
+ "153": "B-US_SSN",
320
+ "154": "I-US_SSN",
321
+ "155": "B-VEHICLE_IDENTIFIER",
322
+ "156": "I-VEHICLE_IDENTIFIER",
323
+ "157": "B-WIRE_TRANSFER_DETAILS",
324
+ "158": "I-WIRE_TRANSFER_DETAILS",
325
+ "159": "B-ZIP_CODE",
326
+ "160": "I-ZIP_CODE"
327
+ },
328
+ "original_entity_types": [
329
+ "PERSON",
330
+ "EMAIL_ADDRESS",
331
+ "PHONE_NUMBER",
332
+ "STREET_ADDRESS",
333
+ "CREDIT_CARD",
334
+ "US_SSN",
335
+ "US_DRIVER_LICENSE",
336
+ "IBAN_CODE",
337
+ "IP_ADDRESS",
338
+ "DATE_TIME",
339
+ "AGE",
340
+ "ORGANIZATION",
341
+ "GPE",
342
+ "ZIP_CODE",
343
+ "DOMAIN_NAME",
344
+ "NRP",
345
+ "TITLE"
346
+ ],
347
+ "new_entity_types": [
348
+ "ACCOUNT_BALANCE",
349
+ "ACCOUNT_NUMBER",
350
+ "ADVERTISING_ID",
351
+ "API_KEY",
352
+ "ASSET_INFORMATION",
353
+ "BANK_ACCOUNT_STATEMENT",
354
+ "BANK_ROUTING_NUMBER",
355
+ "BENEFICIARY_NUMBER",
356
+ "BIOMETRIC_IDENTIFIER",
357
+ "BLOOD_TYPE",
358
+ "BROWSING_HISTORY_BEHAVIOUR_DATA",
359
+ "CHILD_INFORMATION",
360
+ "COOKIE",
361
+ "COORDINATES",
362
+ "CREDIT_HISTORY",
363
+ "CREDIT_SCORE",
364
+ "CRIMINAL_LEGAL_RECORDS",
365
+ "CUSTOMER_ID",
366
+ "CVV",
367
+ "DEBT_INFORMATION",
368
+ "DEVICE_ID",
369
+ "DEVICE_IDENTIFIER",
370
+ "DOB",
371
+ "EDUCATION_LEVEL",
372
+ "EMPLOYEE_ID",
373
+ "EMPLOYMENT_STATUS",
374
+ "FAX_NUMBER",
375
+ "FINANCIAL_STATEMENTS",
376
+ "GENDER",
377
+ "GENETIC_DATA",
378
+ "HEALTH_INSURANCE_ID",
379
+ "HEALTH_INSURANCE_ID_NUMBER",
380
+ "HEALTH_PLAN_BENEFICIARY_NUMBER",
381
+ "INCOME_INFORMATION",
382
+ "LANGUAGE",
383
+ "LICENSE_NUMBER",
384
+ "MAC_ADDRESS",
385
+ "MARITAL_RELATIONSHIP_STATUS",
386
+ "MEDIA_RECORDING",
387
+ "MEDICAL_RECORD_NUMBER",
388
+ "NATIONAL_ID",
389
+ "NETWORK_ACTIVITY_LOGS",
390
+ "NET_WORTH_INFORMATION",
391
+ "OCCUPATION",
392
+ "PAN_NUMBER",
393
+ "PARENT_INFORMATION",
394
+ "PASSPORT_NUMBER",
395
+ "PAYMENT_HISTORY",
396
+ "PIN",
397
+ "PRIVATE_DATE",
398
+ "SECRET",
399
+ "SEXUALITY",
400
+ "SOCIAL_MEDIA_POSTS",
401
+ "STUDENT_ID",
402
+ "SWIFT_CODE",
403
+ "TAX_ID",
404
+ "TIME",
405
+ "TRADE_UNION_MEMBERSHIP",
406
+ "TRANSACTION_HISTORY",
407
+ "UNIQUE_ID",
408
+ "USERNAME",
409
+ "VEHICLE_IDENTIFIER",
410
+ "WIRE_TRANSFER_DETAILS"
411
+ ],
412
+ "all_entity_types": [
413
+ "ACCOUNT_BALANCE",
414
+ "ACCOUNT_NUMBER",
415
+ "ADVERTISING_ID",
416
+ "AGE",
417
+ "API_KEY",
418
+ "ASSET_INFORMATION",
419
+ "BANK_ACCOUNT_STATEMENT",
420
+ "BANK_ROUTING_NUMBER",
421
+ "BENEFICIARY_NUMBER",
422
+ "BIOMETRIC_IDENTIFIER",
423
+ "BLOOD_TYPE",
424
+ "BROWSING_HISTORY_BEHAVIOUR_DATA",
425
+ "CHILD_INFORMATION",
426
+ "COOKIE",
427
+ "COORDINATES",
428
+ "CREDIT_CARD",
429
+ "CREDIT_HISTORY",
430
+ "CREDIT_SCORE",
431
+ "CRIMINAL_LEGAL_RECORDS",
432
+ "CUSTOMER_ID",
433
+ "CVV",
434
+ "DATE_TIME",
435
+ "DEBT_INFORMATION",
436
+ "DEVICE_ID",
437
+ "DEVICE_IDENTIFIER",
438
+ "DOB",
439
+ "DOMAIN_NAME",
440
+ "EDUCATION_LEVEL",
441
+ "EMAIL_ADDRESS",
442
+ "EMPLOYEE_ID",
443
+ "EMPLOYMENT_STATUS",
444
+ "FAX_NUMBER",
445
+ "FINANCIAL_STATEMENTS",
446
+ "GENDER",
447
+ "GENETIC_DATA",
448
+ "GPE",
449
+ "HEALTH_INSURANCE_ID",
450
+ "HEALTH_INSURANCE_ID_NUMBER",
451
+ "HEALTH_PLAN_BENEFICIARY_NUMBER",
452
+ "IBAN_CODE",
453
+ "INCOME_INFORMATION",
454
+ "IP_ADDRESS",
455
+ "LANGUAGE",
456
+ "LICENSE_NUMBER",
457
+ "MAC_ADDRESS",
458
+ "MARITAL_RELATIONSHIP_STATUS",
459
+ "MEDIA_RECORDING",
460
+ "MEDICAL_RECORD_NUMBER",
461
+ "NATIONAL_ID",
462
+ "NETWORK_ACTIVITY_LOGS",
463
+ "NET_WORTH_INFORMATION",
464
+ "NRP",
465
+ "OCCUPATION",
466
+ "ORGANIZATION",
467
+ "PAN_NUMBER",
468
+ "PARENT_INFORMATION",
469
+ "PASSPORT_NUMBER",
470
+ "PAYMENT_HISTORY",
471
+ "PERSON",
472
+ "PHONE_NUMBER",
473
+ "PIN",
474
+ "PRIVATE_DATE",
475
+ "SECRET",
476
+ "SEXUALITY",
477
+ "SOCIAL_MEDIA_POSTS",
478
+ "STREET_ADDRESS",
479
+ "STUDENT_ID",
480
+ "SWIFT_CODE",
481
+ "TAX_ID",
482
+ "TIME",
483
+ "TITLE",
484
+ "TRADE_UNION_MEMBERSHIP",
485
+ "TRANSACTION_HISTORY",
486
+ "UNIQUE_ID",
487
+ "USERNAME",
488
+ "US_DRIVER_LICENSE",
489
+ "US_SSN",
490
+ "VEHICLE_IDENTIFIER",
491
+ "WIRE_TRANSFER_DETAILS",
492
+ "ZIP_CODE"
493
+ ],
494
+ "missing_original_types": [
495
+ "TITLE"
496
+ ]
497
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f8afad52c3980d599c548be9b2687efda3a46a9c6254a8541b0319b758f5486
3
+ size 1230630524
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:609d8f4c067cd3950f88594c5a802616cea245823836ef5848ee4fc40aab5b6f
3
+ size 34363188
tokenizer_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "bos_token": "<bos>",
4
+ "clean_up_tokenization_spaces": false,
5
+ "cls_token": "<bos>",
6
+ "eos_token": "<eos>",
7
+ "is_local": true,
8
+ "local_files_only": false,
9
+ "mask_token": "<mask>",
10
+ "model_input_names": [
11
+ "input_ids",
12
+ "attention_mask"
13
+ ],
14
+ "model_max_length": 32768,
15
+ "pad_token": "<pad>",
16
+ "padding_side": "right",
17
+ "sep_token": "<eos>",
18
+ "spaces_between_special_tokens": false,
19
+ "tokenizer_class": "TokenizersBackend",
20
+ "unk_token": "<unk>"
21
+ }