PuxAI commited on
Commit
928f28d
·
verified ·
1 Parent(s): 6f624e5

Train 10k + CRF for gretel-pii-ready. F1: 0.9576

Browse files
gretel-pii-ready/TokenBased-BERT/config.json ADDED
@@ -0,0 +1,200 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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-ADDRESS",
15
+ "2": "B-API_KEY",
16
+ "3": "B-BANK_ROUTING_NUMBER",
17
+ "4": "B-BIOMETRIC_IDENTIFIER",
18
+ "5": "B-CERTIFICATE_LICENSE_NUMBER",
19
+ "6": "B-CITY",
20
+ "7": "B-COMPANY_NAME",
21
+ "8": "B-COORDINATE",
22
+ "9": "B-COUNTRY",
23
+ "10": "B-CREDIT_CARD_NUMBER",
24
+ "11": "B-CUSTOMER_ID",
25
+ "12": "B-CVV",
26
+ "13": "B-DATE",
27
+ "14": "B-DATE_OF_BIRTH",
28
+ "15": "B-DATE_TIME",
29
+ "16": "B-DEVICE_IDENTIFIER",
30
+ "17": "B-EMAIL",
31
+ "18": "B-EMPLOYEE_ID",
32
+ "19": "B-FIRST_NAME",
33
+ "20": "B-HEALTH_PLAN_BENEFICIARY_NUMBER",
34
+ "21": "B-IPADDRESS",
35
+ "22": "B-LAST_NAME",
36
+ "23": "B-LICENSE_PLATE",
37
+ "24": "B-MEDICAL_RECORD_NUMBER",
38
+ "25": "B-NAME",
39
+ "26": "B-NATIONAL_ID",
40
+ "27": "B-PASSWORD",
41
+ "28": "B-PHONE_NUMBER",
42
+ "29": "B-PIN",
43
+ "30": "B-POSTCODE",
44
+ "31": "B-SSN/CCCD",
45
+ "32": "B-STATE",
46
+ "33": "B-STREET_ADDRESS",
47
+ "34": "B-SWIFT_BIC",
48
+ "35": "B-TAX_ID",
49
+ "36": "B-TIME",
50
+ "37": "B-UNIQUE_IDENTIFIER",
51
+ "38": "B-URL",
52
+ "39": "B-USER_NAME",
53
+ "40": "B-VEHICLE_IDENTIFIER",
54
+ "41": "I-ACCOUNT_NUMBER",
55
+ "42": "I-ADDRESS",
56
+ "43": "I-API_KEY",
57
+ "44": "I-BANK_ROUTING_NUMBER",
58
+ "45": "I-BIOMETRIC_IDENTIFIER",
59
+ "46": "I-CERTIFICATE_LICENSE_NUMBER",
60
+ "47": "I-CITY",
61
+ "48": "I-COMPANY_NAME",
62
+ "49": "I-COORDINATE",
63
+ "50": "I-COUNTRY",
64
+ "51": "I-CREDIT_CARD_NUMBER",
65
+ "52": "I-CUSTOMER_ID",
66
+ "53": "I-CVV",
67
+ "54": "I-DATE",
68
+ "55": "I-DATE_OF_BIRTH",
69
+ "56": "I-DATE_TIME",
70
+ "57": "I-DEVICE_IDENTIFIER",
71
+ "58": "I-EMAIL",
72
+ "59": "I-EMPLOYEE_ID",
73
+ "60": "I-FIRST_NAME",
74
+ "61": "I-HEALTH_PLAN_BENEFICIARY_NUMBER",
75
+ "62": "I-IPADDRESS",
76
+ "63": "I-LAST_NAME",
77
+ "64": "I-LICENSE_PLATE",
78
+ "65": "I-MEDICAL_RECORD_NUMBER",
79
+ "66": "I-NAME",
80
+ "67": "I-NATIONAL_ID",
81
+ "68": "I-PASSWORD",
82
+ "69": "I-PHONE_NUMBER",
83
+ "70": "I-PIN",
84
+ "71": "I-POSTCODE",
85
+ "72": "I-SSN/CCCD",
86
+ "73": "I-STATE",
87
+ "74": "I-STREET_ADDRESS",
88
+ "75": "I-SWIFT_BIC",
89
+ "76": "I-TAX_ID",
90
+ "77": "I-TIME",
91
+ "78": "I-UNIQUE_IDENTIFIER",
92
+ "79": "I-URL",
93
+ "80": "I-USER_NAME",
94
+ "81": "I-VEHICLE_IDENTIFIER",
95
+ "82": "O"
96
+ },
97
+ "initializer_range": 0.02,
98
+ "intermediate_size": 3072,
99
+ "label2id": {
100
+ "B-ACCOUNT_NUMBER": 0,
101
+ "B-ADDRESS": 1,
102
+ "B-API_KEY": 2,
103
+ "B-BANK_ROUTING_NUMBER": 3,
104
+ "B-BIOMETRIC_IDENTIFIER": 4,
105
+ "B-CERTIFICATE_LICENSE_NUMBER": 5,
106
+ "B-CITY": 6,
107
+ "B-COMPANY_NAME": 7,
108
+ "B-COORDINATE": 8,
109
+ "B-COUNTRY": 9,
110
+ "B-CREDIT_CARD_NUMBER": 10,
111
+ "B-CUSTOMER_ID": 11,
112
+ "B-CVV": 12,
113
+ "B-DATE": 13,
114
+ "B-DATE_OF_BIRTH": 14,
115
+ "B-DATE_TIME": 15,
116
+ "B-DEVICE_IDENTIFIER": 16,
117
+ "B-EMAIL": 17,
118
+ "B-EMPLOYEE_ID": 18,
119
+ "B-FIRST_NAME": 19,
120
+ "B-HEALTH_PLAN_BENEFICIARY_NUMBER": 20,
121
+ "B-IPADDRESS": 21,
122
+ "B-LAST_NAME": 22,
123
+ "B-LICENSE_PLATE": 23,
124
+ "B-MEDICAL_RECORD_NUMBER": 24,
125
+ "B-NAME": 25,
126
+ "B-NATIONAL_ID": 26,
127
+ "B-PASSWORD": 27,
128
+ "B-PHONE_NUMBER": 28,
129
+ "B-PIN": 29,
130
+ "B-POSTCODE": 30,
131
+ "B-SSN/CCCD": 31,
132
+ "B-STATE": 32,
133
+ "B-STREET_ADDRESS": 33,
134
+ "B-SWIFT_BIC": 34,
135
+ "B-TAX_ID": 35,
136
+ "B-TIME": 36,
137
+ "B-UNIQUE_IDENTIFIER": 37,
138
+ "B-URL": 38,
139
+ "B-USER_NAME": 39,
140
+ "B-VEHICLE_IDENTIFIER": 40,
141
+ "I-ACCOUNT_NUMBER": 41,
142
+ "I-ADDRESS": 42,
143
+ "I-API_KEY": 43,
144
+ "I-BANK_ROUTING_NUMBER": 44,
145
+ "I-BIOMETRIC_IDENTIFIER": 45,
146
+ "I-CERTIFICATE_LICENSE_NUMBER": 46,
147
+ "I-CITY": 47,
148
+ "I-COMPANY_NAME": 48,
149
+ "I-COORDINATE": 49,
150
+ "I-COUNTRY": 50,
151
+ "I-CREDIT_CARD_NUMBER": 51,
152
+ "I-CUSTOMER_ID": 52,
153
+ "I-CVV": 53,
154
+ "I-DATE": 54,
155
+ "I-DATE_OF_BIRTH": 55,
156
+ "I-DATE_TIME": 56,
157
+ "I-DEVICE_IDENTIFIER": 57,
158
+ "I-EMAIL": 58,
159
+ "I-EMPLOYEE_ID": 59,
160
+ "I-FIRST_NAME": 60,
161
+ "I-HEALTH_PLAN_BENEFICIARY_NUMBER": 61,
162
+ "I-IPADDRESS": 62,
163
+ "I-LAST_NAME": 63,
164
+ "I-LICENSE_PLATE": 64,
165
+ "I-MEDICAL_RECORD_NUMBER": 65,
166
+ "I-NAME": 66,
167
+ "I-NATIONAL_ID": 67,
168
+ "I-PASSWORD": 68,
169
+ "I-PHONE_NUMBER": 69,
170
+ "I-PIN": 70,
171
+ "I-POSTCODE": 71,
172
+ "I-SSN/CCCD": 72,
173
+ "I-STATE": 73,
174
+ "I-STREET_ADDRESS": 74,
175
+ "I-SWIFT_BIC": 75,
176
+ "I-TAX_ID": 76,
177
+ "I-TIME": 77,
178
+ "I-UNIQUE_IDENTIFIER": 78,
179
+ "I-URL": 79,
180
+ "I-USER_NAME": 80,
181
+ "I-VEHICLE_IDENTIFIER": 81,
182
+ "O": 82
183
+ },
184
+ "layer_norm_eps": 1e-12,
185
+ "max_position_embeddings": 512,
186
+ "model_type": "bert",
187
+ "num_attention_heads": 12,
188
+ "num_hidden_layers": 12,
189
+ "pad_token_id": 0,
190
+ "pooler_fc_size": 768,
191
+ "pooler_num_attention_heads": 12,
192
+ "pooler_num_fc_layers": 3,
193
+ "pooler_size_per_head": 128,
194
+ "pooler_type": "first_token_transform",
195
+ "position_embedding_type": "absolute",
196
+ "transformers_version": "4.56.0",
197
+ "type_vocab_size": 2,
198
+ "use_cache": false,
199
+ "vocab_size": 119547
200
+ }
gretel-pii-ready/TokenBased-BERT/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:615a01d8f5a1ccbf233fe6e2effb060899e14af8433b20baf9bde3b088793778
3
  size 711721104
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9373a4d407993ae77d6c24a301f5a8dccc7ee3e7fec365d4e7f282b4f8b9d55a
3
  size 711721104
gretel-pii-ready/TokenBased-BERT/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0408505de4e49c1e6352ad3bfc9e4bf6cdaadae5346c991a6e07d0054acf1d80
3
  size 5777
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f3fca44ea2c3269af512f4e8fcac91564061df8d0a811814cfddb25dfea554b
3
  size 5777