PuxAI commited on
Commit
b5d71eb
·
verified ·
1 Parent(s): ab58273

Train 10k + CRF for nemotron-pii-ready. F1: 0.9618

Browse files
nemotron-pii-ready/TokenBased-BERT/config.json ADDED
@@ -0,0 +1,252 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "BertForTokenClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "classifier_dropout": null,
7
+ "directionality": "bidi",
8
+ "dtype": "float32",
9
+ "hidden_act": "gelu",
10
+ "hidden_dropout_prob": 0.1,
11
+ "hidden_size": 768,
12
+ "id2label": {
13
+ "0": "B-ACCOUNT_NUMBER",
14
+ "1": "B-AGE",
15
+ "2": "B-API_KEY",
16
+ "3": "B-BANK_ROUTING_NUMBER",
17
+ "4": "B-BIOMETRIC_IDENTIFIER",
18
+ "5": "B-BLOOD_TYPE",
19
+ "6": "B-CERTIFICATE_LICENSE_NUMBER",
20
+ "7": "B-CITY",
21
+ "8": "B-COMPANY_NAME",
22
+ "9": "B-COORDINATE",
23
+ "10": "B-COUNTRY",
24
+ "11": "B-COUNTY",
25
+ "12": "B-CREDIT_DEBIT_CARD",
26
+ "13": "B-CUSTOMER_ID",
27
+ "14": "B-CVV",
28
+ "15": "B-DATE",
29
+ "16": "B-DATE_OF_BIRTH",
30
+ "17": "B-DATE_TIME",
31
+ "18": "B-DEVICE_IDENTIFIER",
32
+ "19": "B-EDUCATION_LEVEL",
33
+ "20": "B-EMAIL",
34
+ "21": "B-EMPLOYEE_ID",
35
+ "22": "B-EMPLOYMENT_STATUS",
36
+ "23": "B-FAX_NUMBER",
37
+ "24": "B-FIRST_NAME",
38
+ "25": "B-GENDER",
39
+ "26": "B-HEALTH_PLAN_BENEFICIARY_NUMBER",
40
+ "27": "B-HTTP_COOKIE",
41
+ "28": "B-IPADDRESS",
42
+ "29": "B-LANGUAGE",
43
+ "30": "B-LAST_NAME",
44
+ "31": "B-LICENSE_PLATE",
45
+ "32": "B-MAC_ADDRESS",
46
+ "33": "B-MEDICAL_RECORD_NUMBER",
47
+ "34": "B-NATIONAL_ID",
48
+ "35": "B-OCCUPATION",
49
+ "36": "B-PASSWORD",
50
+ "37": "B-PHONE_NUMBER",
51
+ "38": "B-PIN",
52
+ "39": "B-POLITICAL_VIEW",
53
+ "40": "B-POSTCODE",
54
+ "41": "B-RACE_ETHNICITY",
55
+ "42": "B-RELIGIOUS_BELIEF",
56
+ "43": "B-SEXUALITY",
57
+ "44": "B-SSN/CCCD",
58
+ "45": "B-STATE",
59
+ "46": "B-STREET_ADDRESS",
60
+ "47": "B-SWIFT_BIC",
61
+ "48": "B-TAX_ID",
62
+ "49": "B-TIME",
63
+ "50": "B-UNIQUE_ID",
64
+ "51": "B-URL",
65
+ "52": "B-USER_NAME",
66
+ "53": "B-VEHICLE_IDENTIFIER",
67
+ "54": "I-ACCOUNT_NUMBER",
68
+ "55": "I-AGE",
69
+ "56": "I-API_KEY",
70
+ "57": "I-BANK_ROUTING_NUMBER",
71
+ "58": "I-BIOMETRIC_IDENTIFIER",
72
+ "59": "I-BLOOD_TYPE",
73
+ "60": "I-CERTIFICATE_LICENSE_NUMBER",
74
+ "61": "I-CITY",
75
+ "62": "I-COMPANY_NAME",
76
+ "63": "I-COORDINATE",
77
+ "64": "I-COUNTRY",
78
+ "65": "I-COUNTY",
79
+ "66": "I-CREDIT_DEBIT_CARD",
80
+ "67": "I-CUSTOMER_ID",
81
+ "68": "I-CVV",
82
+ "69": "I-DATE",
83
+ "70": "I-DATE_OF_BIRTH",
84
+ "71": "I-DATE_TIME",
85
+ "72": "I-DEVICE_IDENTIFIER",
86
+ "73": "I-EDUCATION_LEVEL",
87
+ "74": "I-EMAIL",
88
+ "75": "I-EMPLOYEE_ID",
89
+ "76": "I-EMPLOYMENT_STATUS",
90
+ "77": "I-FAX_NUMBER",
91
+ "78": "I-FIRST_NAME",
92
+ "79": "I-GENDER",
93
+ "80": "I-HEALTH_PLAN_BENEFICIARY_NUMBER",
94
+ "81": "I-HTTP_COOKIE",
95
+ "82": "I-IPADDRESS",
96
+ "83": "I-LANGUAGE",
97
+ "84": "I-LAST_NAME",
98
+ "85": "I-LICENSE_PLATE",
99
+ "86": "I-MAC_ADDRESS",
100
+ "87": "I-MEDICAL_RECORD_NUMBER",
101
+ "88": "I-NATIONAL_ID",
102
+ "89": "I-OCCUPATION",
103
+ "90": "I-PASSWORD",
104
+ "91": "I-PHONE_NUMBER",
105
+ "92": "I-PIN",
106
+ "93": "I-POLITICAL_VIEW",
107
+ "94": "I-POSTCODE",
108
+ "95": "I-RACE_ETHNICITY",
109
+ "96": "I-RELIGIOUS_BELIEF",
110
+ "97": "I-SEXUALITY",
111
+ "98": "I-SSN/CCCD",
112
+ "99": "I-STATE",
113
+ "100": "I-STREET_ADDRESS",
114
+ "101": "I-SWIFT_BIC",
115
+ "102": "I-TAX_ID",
116
+ "103": "I-TIME",
117
+ "104": "I-UNIQUE_ID",
118
+ "105": "I-URL",
119
+ "106": "I-USER_NAME",
120
+ "107": "I-VEHICLE_IDENTIFIER",
121
+ "108": "O"
122
+ },
123
+ "initializer_range": 0.02,
124
+ "intermediate_size": 3072,
125
+ "label2id": {
126
+ "B-ACCOUNT_NUMBER": 0,
127
+ "B-AGE": 1,
128
+ "B-API_KEY": 2,
129
+ "B-BANK_ROUTING_NUMBER": 3,
130
+ "B-BIOMETRIC_IDENTIFIER": 4,
131
+ "B-BLOOD_TYPE": 5,
132
+ "B-CERTIFICATE_LICENSE_NUMBER": 6,
133
+ "B-CITY": 7,
134
+ "B-COMPANY_NAME": 8,
135
+ "B-COORDINATE": 9,
136
+ "B-COUNTRY": 10,
137
+ "B-COUNTY": 11,
138
+ "B-CREDIT_DEBIT_CARD": 12,
139
+ "B-CUSTOMER_ID": 13,
140
+ "B-CVV": 14,
141
+ "B-DATE": 15,
142
+ "B-DATE_OF_BIRTH": 16,
143
+ "B-DATE_TIME": 17,
144
+ "B-DEVICE_IDENTIFIER": 18,
145
+ "B-EDUCATION_LEVEL": 19,
146
+ "B-EMAIL": 20,
147
+ "B-EMPLOYEE_ID": 21,
148
+ "B-EMPLOYMENT_STATUS": 22,
149
+ "B-FAX_NUMBER": 23,
150
+ "B-FIRST_NAME": 24,
151
+ "B-GENDER": 25,
152
+ "B-HEALTH_PLAN_BENEFICIARY_NUMBER": 26,
153
+ "B-HTTP_COOKIE": 27,
154
+ "B-IPADDRESS": 28,
155
+ "B-LANGUAGE": 29,
156
+ "B-LAST_NAME": 30,
157
+ "B-LICENSE_PLATE": 31,
158
+ "B-MAC_ADDRESS": 32,
159
+ "B-MEDICAL_RECORD_NUMBER": 33,
160
+ "B-NATIONAL_ID": 34,
161
+ "B-OCCUPATION": 35,
162
+ "B-PASSWORD": 36,
163
+ "B-PHONE_NUMBER": 37,
164
+ "B-PIN": 38,
165
+ "B-POLITICAL_VIEW": 39,
166
+ "B-POSTCODE": 40,
167
+ "B-RACE_ETHNICITY": 41,
168
+ "B-RELIGIOUS_BELIEF": 42,
169
+ "B-SEXUALITY": 43,
170
+ "B-SSN/CCCD": 44,
171
+ "B-STATE": 45,
172
+ "B-STREET_ADDRESS": 46,
173
+ "B-SWIFT_BIC": 47,
174
+ "B-TAX_ID": 48,
175
+ "B-TIME": 49,
176
+ "B-UNIQUE_ID": 50,
177
+ "B-URL": 51,
178
+ "B-USER_NAME": 52,
179
+ "B-VEHICLE_IDENTIFIER": 53,
180
+ "I-ACCOUNT_NUMBER": 54,
181
+ "I-AGE": 55,
182
+ "I-API_KEY": 56,
183
+ "I-BANK_ROUTING_NUMBER": 57,
184
+ "I-BIOMETRIC_IDENTIFIER": 58,
185
+ "I-BLOOD_TYPE": 59,
186
+ "I-CERTIFICATE_LICENSE_NUMBER": 60,
187
+ "I-CITY": 61,
188
+ "I-COMPANY_NAME": 62,
189
+ "I-COORDINATE": 63,
190
+ "I-COUNTRY": 64,
191
+ "I-COUNTY": 65,
192
+ "I-CREDIT_DEBIT_CARD": 66,
193
+ "I-CUSTOMER_ID": 67,
194
+ "I-CVV": 68,
195
+ "I-DATE": 69,
196
+ "I-DATE_OF_BIRTH": 70,
197
+ "I-DATE_TIME": 71,
198
+ "I-DEVICE_IDENTIFIER": 72,
199
+ "I-EDUCATION_LEVEL": 73,
200
+ "I-EMAIL": 74,
201
+ "I-EMPLOYEE_ID": 75,
202
+ "I-EMPLOYMENT_STATUS": 76,
203
+ "I-FAX_NUMBER": 77,
204
+ "I-FIRST_NAME": 78,
205
+ "I-GENDER": 79,
206
+ "I-HEALTH_PLAN_BENEFICIARY_NUMBER": 80,
207
+ "I-HTTP_COOKIE": 81,
208
+ "I-IPADDRESS": 82,
209
+ "I-LANGUAGE": 83,
210
+ "I-LAST_NAME": 84,
211
+ "I-LICENSE_PLATE": 85,
212
+ "I-MAC_ADDRESS": 86,
213
+ "I-MEDICAL_RECORD_NUMBER": 87,
214
+ "I-NATIONAL_ID": 88,
215
+ "I-OCCUPATION": 89,
216
+ "I-PASSWORD": 90,
217
+ "I-PHONE_NUMBER": 91,
218
+ "I-PIN": 92,
219
+ "I-POLITICAL_VIEW": 93,
220
+ "I-POSTCODE": 94,
221
+ "I-RACE_ETHNICITY": 95,
222
+ "I-RELIGIOUS_BELIEF": 96,
223
+ "I-SEXUALITY": 97,
224
+ "I-SSN/CCCD": 98,
225
+ "I-STATE": 99,
226
+ "I-STREET_ADDRESS": 100,
227
+ "I-SWIFT_BIC": 101,
228
+ "I-TAX_ID": 102,
229
+ "I-TIME": 103,
230
+ "I-UNIQUE_ID": 104,
231
+ "I-URL": 105,
232
+ "I-USER_NAME": 106,
233
+ "I-VEHICLE_IDENTIFIER": 107,
234
+ "O": 108
235
+ },
236
+ "layer_norm_eps": 1e-12,
237
+ "max_position_embeddings": 512,
238
+ "model_type": "bert",
239
+ "num_attention_heads": 12,
240
+ "num_hidden_layers": 12,
241
+ "pad_token_id": 0,
242
+ "pooler_fc_size": 768,
243
+ "pooler_num_attention_heads": 12,
244
+ "pooler_num_fc_layers": 3,
245
+ "pooler_size_per_head": 128,
246
+ "pooler_type": "first_token_transform",
247
+ "position_embedding_type": "absolute",
248
+ "transformers_version": "4.56.0",
249
+ "type_vocab_size": 2,
250
+ "use_cache": false,
251
+ "vocab_size": 119547
252
+ }
nemotron-pii-ready/TokenBased-BERT/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4649138c23283fe799aab9937ded575e6bb58db7787e7f039aaf963dd511e993
3
  size 711821256
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e3b1c7ee6eba638fe02ddb6377c45baa8b85edbb7c6817a5b9894ce0949fe00
3
  size 711821256
nemotron-pii-ready/TokenBased-BERT/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e8a60c4b5ef8d85a9f06594d38fa25eeb77e83dafcbfc5b502ca6b4157986510
3
  size 5777
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c42a27d9755be130f1e33b02c56d544b73c29c68df62f18464429769b372eab0
3
  size 5777