toghrultahirov commited on
Commit
87fc268
1 Parent(s): af69b2e

Training in progress, epoch 1

Browse files
Files changed (3) hide show
  1. config.json +221 -221
  2. model.safetensors +2 -2
  3. training_args.bin +1 -1
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "prajjwal1/bert-mini",
3
  "architectures": [
4
  "BertForTokenClassification"
5
  ],
@@ -7,244 +7,244 @@
7
  "classifier_dropout": null,
8
  "hidden_act": "gelu",
9
  "hidden_dropout_prob": 0.1,
10
- "hidden_size": 256,
11
  "id2label": {
12
  "0": "O",
13
- "1": "B-IPV4",
14
- "2": "I-IPV4",
15
  "3": "B-BIC",
16
  "4": "I-BIC",
17
- "5": "B-MASKEDNUMBER",
18
- "6": "I-MASKEDNUMBER",
19
- "7": "B-USERAGENT",
20
- "8": "I-USERAGENT",
21
- "9": "B-ZIPCODE",
22
- "10": "I-ZIPCODE",
23
- "11": "B-HEIGHT",
24
- "12": "I-HEIGHT",
25
  "13": "B-VEHICLEVRM",
26
  "14": "I-VEHICLEVRM",
27
- "15": "B-MAC",
28
- "16": "I-MAC",
29
- "17": "B-URL",
30
- "18": "I-URL",
31
- "19": "B-IP",
32
- "20": "I-IP",
33
- "21": "B-GENDER",
34
- "22": "I-GENDER",
35
- "23": "B-PREFIX",
36
- "24": "I-PREFIX",
37
- "25": "B-STATE",
38
- "26": "I-STATE",
39
- "27": "B-AGE",
40
- "28": "I-AGE",
41
- "29": "B-PIN",
42
- "30": "I-PIN",
43
- "31": "B-IBAN",
44
- "32": "I-IBAN",
45
- "33": "B-COMPANYNAME",
46
- "34": "I-COMPANYNAME",
47
- "35": "B-STREET",
48
- "36": "I-STREET",
49
- "37": "B-CREDITCARDNUMBER",
50
- "38": "I-CREDITCARDNUMBER",
51
- "39": "B-DOB",
52
- "40": "I-DOB",
53
- "41": "B-IPV6",
54
- "42": "I-IPV6",
55
- "43": "B-VEHICLEVIN",
56
- "44": "I-VEHICLEVIN",
57
- "45": "B-CURRENCYCODE",
58
- "46": "I-CURRENCYCODE",
59
- "47": "B-USERNAME",
60
- "48": "I-USERNAME",
61
- "49": "B-JOBAREA",
62
- "50": "I-JOBAREA",
63
- "51": "B-EYECOLOR",
64
- "52": "I-EYECOLOR",
65
- "53": "B-MIDDLENAME",
66
- "54": "I-MIDDLENAME",
67
- "55": "B-EMAIL",
68
- "56": "I-EMAIL",
69
- "57": "B-CURRENCYNAME",
70
- "58": "I-CURRENCYNAME",
71
- "59": "B-COUNTY",
72
- "60": "I-COUNTY",
73
- "61": "B-CURRENCY",
74
- "62": "I-CURRENCY",
75
- "63": "B-ORDINALDIRECTION",
76
- "64": "I-ORDINALDIRECTION",
77
- "65": "B-ACCOUNTNUMBER",
78
- "66": "I-ACCOUNTNUMBER",
79
- "67": "B-TIME",
80
- "68": "I-TIME",
81
- "69": "B-ACCOUNTNAME",
82
- "70": "I-ACCOUNTNAME",
83
- "71": "B-CREDITCARDCVV",
84
- "72": "I-CREDITCARDCVV",
85
- "73": "B-DATE",
86
- "74": "I-DATE",
87
- "75": "B-LITECOINADDRESS",
88
- "76": "I-LITECOINADDRESS",
89
- "77": "B-AMOUNT",
90
- "78": "I-AMOUNT",
91
- "79": "B-ETHEREUMADDRESS",
92
- "80": "I-ETHEREUMADDRESS",
93
- "81": "B-PHONENUMBER",
94
- "82": "I-PHONENUMBER",
95
- "83": "B-CITY",
96
- "84": "I-CITY",
97
- "85": "B-SEX",
98
- "86": "I-SEX",
99
- "87": "B-CREDITCARDISSUER",
100
- "88": "I-CREDITCARDISSUER",
101
- "89": "B-SSN",
102
- "90": "I-SSN",
103
- "91": "B-PHONEIMEI",
104
- "92": "I-PHONEIMEI",
105
- "93": "B-NEARBYGPSCOORDINATE",
106
- "94": "I-NEARBYGPSCOORDINATE",
107
- "95": "B-CURRENCYSYMBOL",
108
- "96": "I-CURRENCYSYMBOL",
109
- "97": "B-BITCOINADDRESS",
110
- "98": "I-BITCOINADDRESS",
111
- "99": "B-PASSWORD",
112
- "100": "I-PASSWORD",
113
- "101": "B-JOBTYPE",
114
- "102": "I-JOBTYPE",
115
- "103": "B-JOBTITLE",
116
- "104": "I-JOBTITLE",
117
- "105": "B-LASTNAME",
118
- "106": "I-LASTNAME",
119
- "107": "B-FIRSTNAME",
120
- "108": "I-FIRSTNAME",
121
- "109": "B-BUILDINGNUMBER",
122
- "110": "I-BUILDINGNUMBER",
123
- "111": "B-SECONDARYADDRESS",
124
- "112": "I-SECONDARYADDRESS"
125
  },
126
  "initializer_range": 0.02,
127
- "intermediate_size": 1024,
128
  "label2id": {
129
- "B-ACCOUNTNAME": 69,
130
- "B-ACCOUNTNUMBER": 65,
131
- "B-AGE": 27,
132
- "B-AMOUNT": 77,
133
  "B-BIC": 3,
134
- "B-BITCOINADDRESS": 97,
135
- "B-BUILDINGNUMBER": 109,
136
- "B-CITY": 83,
137
- "B-COMPANYNAME": 33,
138
- "B-COUNTY": 59,
139
- "B-CREDITCARDCVV": 71,
140
- "B-CREDITCARDISSUER": 87,
141
- "B-CREDITCARDNUMBER": 37,
142
- "B-CURRENCY": 61,
143
- "B-CURRENCYCODE": 45,
144
- "B-CURRENCYNAME": 57,
145
- "B-CURRENCYSYMBOL": 95,
146
- "B-DATE": 73,
147
- "B-DOB": 39,
148
- "B-EMAIL": 55,
149
- "B-ETHEREUMADDRESS": 79,
150
- "B-EYECOLOR": 51,
151
- "B-FIRSTNAME": 107,
152
- "B-GENDER": 21,
153
- "B-HEIGHT": 11,
154
- "B-IBAN": 31,
155
- "B-IP": 19,
156
- "B-IPV4": 1,
157
- "B-IPV6": 41,
158
- "B-JOBAREA": 49,
159
- "B-JOBTITLE": 103,
160
- "B-JOBTYPE": 101,
161
- "B-LASTNAME": 105,
162
- "B-LITECOINADDRESS": 75,
163
- "B-MAC": 15,
164
- "B-MASKEDNUMBER": 5,
165
- "B-MIDDLENAME": 53,
166
- "B-NEARBYGPSCOORDINATE": 93,
167
- "B-ORDINALDIRECTION": 63,
168
- "B-PASSWORD": 99,
169
- "B-PHONEIMEI": 91,
170
- "B-PHONENUMBER": 81,
171
- "B-PIN": 29,
172
- "B-PREFIX": 23,
173
- "B-SECONDARYADDRESS": 111,
174
- "B-SEX": 85,
175
- "B-SSN": 89,
176
- "B-STATE": 25,
177
- "B-STREET": 35,
178
- "B-TIME": 67,
179
- "B-URL": 17,
180
- "B-USERAGENT": 7,
181
- "B-USERNAME": 47,
182
- "B-VEHICLEVIN": 43,
183
  "B-VEHICLEVRM": 13,
184
- "B-ZIPCODE": 9,
185
- "I-ACCOUNTNAME": 70,
186
- "I-ACCOUNTNUMBER": 66,
187
- "I-AGE": 28,
188
- "I-AMOUNT": 78,
189
  "I-BIC": 4,
190
- "I-BITCOINADDRESS": 98,
191
- "I-BUILDINGNUMBER": 110,
192
- "I-CITY": 84,
193
- "I-COMPANYNAME": 34,
194
- "I-COUNTY": 60,
195
- "I-CREDITCARDCVV": 72,
196
- "I-CREDITCARDISSUER": 88,
197
- "I-CREDITCARDNUMBER": 38,
198
- "I-CURRENCY": 62,
199
- "I-CURRENCYCODE": 46,
200
- "I-CURRENCYNAME": 58,
201
- "I-CURRENCYSYMBOL": 96,
202
- "I-DATE": 74,
203
- "I-DOB": 40,
204
- "I-EMAIL": 56,
205
- "I-ETHEREUMADDRESS": 80,
206
- "I-EYECOLOR": 52,
207
- "I-FIRSTNAME": 108,
208
- "I-GENDER": 22,
209
- "I-HEIGHT": 12,
210
- "I-IBAN": 32,
211
- "I-IP": 20,
212
- "I-IPV4": 2,
213
- "I-IPV6": 42,
214
- "I-JOBAREA": 50,
215
- "I-JOBTITLE": 104,
216
- "I-JOBTYPE": 102,
217
- "I-LASTNAME": 106,
218
- "I-LITECOINADDRESS": 76,
219
- "I-MAC": 16,
220
- "I-MASKEDNUMBER": 6,
221
- "I-MIDDLENAME": 54,
222
- "I-NEARBYGPSCOORDINATE": 94,
223
- "I-ORDINALDIRECTION": 64,
224
- "I-PASSWORD": 100,
225
- "I-PHONEIMEI": 92,
226
- "I-PHONENUMBER": 82,
227
- "I-PIN": 30,
228
- "I-PREFIX": 24,
229
- "I-SECONDARYADDRESS": 112,
230
- "I-SEX": 86,
231
- "I-SSN": 90,
232
- "I-STATE": 26,
233
- "I-STREET": 36,
234
- "I-TIME": 68,
235
- "I-URL": 18,
236
- "I-USERAGENT": 8,
237
- "I-USERNAME": 48,
238
- "I-VEHICLEVIN": 44,
239
  "I-VEHICLEVRM": 14,
240
- "I-ZIPCODE": 10,
241
  "O": 0
242
  },
243
  "layer_norm_eps": 1e-12,
244
  "max_position_embeddings": 512,
245
  "model_type": "bert",
246
- "num_attention_heads": 4,
247
- "num_hidden_layers": 4,
248
  "pad_token_id": 0,
249
  "position_embedding_type": "absolute",
250
  "torch_dtype": "float32",
 
1
  {
2
+ "_name_or_path": "google/bert_uncased_L-8_H-512_A-8",
3
  "architectures": [
4
  "BertForTokenClassification"
5
  ],
 
7
  "classifier_dropout": null,
8
  "hidden_act": "gelu",
9
  "hidden_dropout_prob": 0.1,
10
+ "hidden_size": 512,
11
  "id2label": {
12
  "0": "O",
13
+ "1": "B-BUILDINGNUMBER",
14
+ "2": "I-BUILDINGNUMBER",
15
  "3": "B-BIC",
16
  "4": "I-BIC",
17
+ "5": "B-CREDITCARDISSUER",
18
+ "6": "I-CREDITCARDISSUER",
19
+ "7": "B-STATE",
20
+ "8": "I-STATE",
21
+ "9": "B-COMPANYNAME",
22
+ "10": "I-COMPANYNAME",
23
+ "11": "B-USERNAME",
24
+ "12": "I-USERNAME",
25
  "13": "B-VEHICLEVRM",
26
  "14": "I-VEHICLEVRM",
27
+ "15": "B-JOBTYPE",
28
+ "16": "I-JOBTYPE",
29
+ "17": "B-IPV4",
30
+ "18": "I-IPV4",
31
+ "19": "B-FIRSTNAME",
32
+ "20": "I-FIRSTNAME",
33
+ "21": "B-COUNTY",
34
+ "22": "I-COUNTY",
35
+ "23": "B-IBAN",
36
+ "24": "I-IBAN",
37
+ "25": "B-SECONDARYADDRESS",
38
+ "26": "I-SECONDARYADDRESS",
39
+ "27": "B-SEX",
40
+ "28": "I-SEX",
41
+ "29": "B-DOB",
42
+ "30": "I-DOB",
43
+ "31": "B-VEHICLEVIN",
44
+ "32": "I-VEHICLEVIN",
45
+ "33": "B-ZIPCODE",
46
+ "34": "I-ZIPCODE",
47
+ "35": "B-EYECOLOR",
48
+ "36": "I-EYECOLOR",
49
+ "37": "B-LASTNAME",
50
+ "38": "I-LASTNAME",
51
+ "39": "B-STREET",
52
+ "40": "I-STREET",
53
+ "41": "B-PREFIX",
54
+ "42": "I-PREFIX",
55
+ "43": "B-ORDINALDIRECTION",
56
+ "44": "I-ORDINALDIRECTION",
57
+ "45": "B-CURRENCY",
58
+ "46": "I-CURRENCY",
59
+ "47": "B-JOBAREA",
60
+ "48": "I-JOBAREA",
61
+ "49": "B-MASKEDNUMBER",
62
+ "50": "I-MASKEDNUMBER",
63
+ "51": "B-EMAIL",
64
+ "52": "I-EMAIL",
65
+ "53": "B-ACCOUNTNUMBER",
66
+ "54": "I-ACCOUNTNUMBER",
67
+ "55": "B-CREDITCARDCVV",
68
+ "56": "I-CREDITCARDCVV",
69
+ "57": "B-CURRENCYSYMBOL",
70
+ "58": "I-CURRENCYSYMBOL",
71
+ "59": "B-TIME",
72
+ "60": "I-TIME",
73
+ "61": "B-HEIGHT",
74
+ "62": "I-HEIGHT",
75
+ "63": "B-IP",
76
+ "64": "I-IP",
77
+ "65": "B-ACCOUNTNAME",
78
+ "66": "I-ACCOUNTNAME",
79
+ "67": "B-DATE",
80
+ "68": "I-DATE",
81
+ "69": "B-GENDER",
82
+ "70": "I-GENDER",
83
+ "71": "B-AMOUNT",
84
+ "72": "I-AMOUNT",
85
+ "73": "B-JOBTITLE",
86
+ "74": "I-JOBTITLE",
87
+ "75": "B-CREDITCARDNUMBER",
88
+ "76": "I-CREDITCARDNUMBER",
89
+ "77": "B-AGE",
90
+ "78": "I-AGE",
91
+ "79": "B-SSN",
92
+ "80": "I-SSN",
93
+ "81": "B-PIN",
94
+ "82": "I-PIN",
95
+ "83": "B-PHONEIMEI",
96
+ "84": "I-PHONEIMEI",
97
+ "85": "B-BITCOINADDRESS",
98
+ "86": "I-BITCOINADDRESS",
99
+ "87": "B-LITECOINADDRESS",
100
+ "88": "I-LITECOINADDRESS",
101
+ "89": "B-USERAGENT",
102
+ "90": "I-USERAGENT",
103
+ "91": "B-URL",
104
+ "92": "I-URL",
105
+ "93": "B-PASSWORD",
106
+ "94": "I-PASSWORD",
107
+ "95": "B-IPV6",
108
+ "96": "I-IPV6",
109
+ "97": "B-CURRENCYCODE",
110
+ "98": "I-CURRENCYCODE",
111
+ "99": "B-MIDDLENAME",
112
+ "100": "I-MIDDLENAME",
113
+ "101": "B-CITY",
114
+ "102": "I-CITY",
115
+ "103": "B-NEARBYGPSCOORDINATE",
116
+ "104": "I-NEARBYGPSCOORDINATE",
117
+ "105": "B-ETHEREUMADDRESS",
118
+ "106": "I-ETHEREUMADDRESS",
119
+ "107": "B-PHONENUMBER",
120
+ "108": "I-PHONENUMBER",
121
+ "109": "B-CURRENCYNAME",
122
+ "110": "I-CURRENCYNAME",
123
+ "111": "B-MAC",
124
+ "112": "I-MAC"
125
  },
126
  "initializer_range": 0.02,
127
+ "intermediate_size": 2048,
128
  "label2id": {
129
+ "B-ACCOUNTNAME": 65,
130
+ "B-ACCOUNTNUMBER": 53,
131
+ "B-AGE": 77,
132
+ "B-AMOUNT": 71,
133
  "B-BIC": 3,
134
+ "B-BITCOINADDRESS": 85,
135
+ "B-BUILDINGNUMBER": 1,
136
+ "B-CITY": 101,
137
+ "B-COMPANYNAME": 9,
138
+ "B-COUNTY": 21,
139
+ "B-CREDITCARDCVV": 55,
140
+ "B-CREDITCARDISSUER": 5,
141
+ "B-CREDITCARDNUMBER": 75,
142
+ "B-CURRENCY": 45,
143
+ "B-CURRENCYCODE": 97,
144
+ "B-CURRENCYNAME": 109,
145
+ "B-CURRENCYSYMBOL": 57,
146
+ "B-DATE": 67,
147
+ "B-DOB": 29,
148
+ "B-EMAIL": 51,
149
+ "B-ETHEREUMADDRESS": 105,
150
+ "B-EYECOLOR": 35,
151
+ "B-FIRSTNAME": 19,
152
+ "B-GENDER": 69,
153
+ "B-HEIGHT": 61,
154
+ "B-IBAN": 23,
155
+ "B-IP": 63,
156
+ "B-IPV4": 17,
157
+ "B-IPV6": 95,
158
+ "B-JOBAREA": 47,
159
+ "B-JOBTITLE": 73,
160
+ "B-JOBTYPE": 15,
161
+ "B-LASTNAME": 37,
162
+ "B-LITECOINADDRESS": 87,
163
+ "B-MAC": 111,
164
+ "B-MASKEDNUMBER": 49,
165
+ "B-MIDDLENAME": 99,
166
+ "B-NEARBYGPSCOORDINATE": 103,
167
+ "B-ORDINALDIRECTION": 43,
168
+ "B-PASSWORD": 93,
169
+ "B-PHONEIMEI": 83,
170
+ "B-PHONENUMBER": 107,
171
+ "B-PIN": 81,
172
+ "B-PREFIX": 41,
173
+ "B-SECONDARYADDRESS": 25,
174
+ "B-SEX": 27,
175
+ "B-SSN": 79,
176
+ "B-STATE": 7,
177
+ "B-STREET": 39,
178
+ "B-TIME": 59,
179
+ "B-URL": 91,
180
+ "B-USERAGENT": 89,
181
+ "B-USERNAME": 11,
182
+ "B-VEHICLEVIN": 31,
183
  "B-VEHICLEVRM": 13,
184
+ "B-ZIPCODE": 33,
185
+ "I-ACCOUNTNAME": 66,
186
+ "I-ACCOUNTNUMBER": 54,
187
+ "I-AGE": 78,
188
+ "I-AMOUNT": 72,
189
  "I-BIC": 4,
190
+ "I-BITCOINADDRESS": 86,
191
+ "I-BUILDINGNUMBER": 2,
192
+ "I-CITY": 102,
193
+ "I-COMPANYNAME": 10,
194
+ "I-COUNTY": 22,
195
+ "I-CREDITCARDCVV": 56,
196
+ "I-CREDITCARDISSUER": 6,
197
+ "I-CREDITCARDNUMBER": 76,
198
+ "I-CURRENCY": 46,
199
+ "I-CURRENCYCODE": 98,
200
+ "I-CURRENCYNAME": 110,
201
+ "I-CURRENCYSYMBOL": 58,
202
+ "I-DATE": 68,
203
+ "I-DOB": 30,
204
+ "I-EMAIL": 52,
205
+ "I-ETHEREUMADDRESS": 106,
206
+ "I-EYECOLOR": 36,
207
+ "I-FIRSTNAME": 20,
208
+ "I-GENDER": 70,
209
+ "I-HEIGHT": 62,
210
+ "I-IBAN": 24,
211
+ "I-IP": 64,
212
+ "I-IPV4": 18,
213
+ "I-IPV6": 96,
214
+ "I-JOBAREA": 48,
215
+ "I-JOBTITLE": 74,
216
+ "I-JOBTYPE": 16,
217
+ "I-LASTNAME": 38,
218
+ "I-LITECOINADDRESS": 88,
219
+ "I-MAC": 112,
220
+ "I-MASKEDNUMBER": 50,
221
+ "I-MIDDLENAME": 100,
222
+ "I-NEARBYGPSCOORDINATE": 104,
223
+ "I-ORDINALDIRECTION": 44,
224
+ "I-PASSWORD": 94,
225
+ "I-PHONEIMEI": 84,
226
+ "I-PHONENUMBER": 108,
227
+ "I-PIN": 82,
228
+ "I-PREFIX": 42,
229
+ "I-SECONDARYADDRESS": 26,
230
+ "I-SEX": 28,
231
+ "I-SSN": 80,
232
+ "I-STATE": 8,
233
+ "I-STREET": 40,
234
+ "I-TIME": 60,
235
+ "I-URL": 92,
236
+ "I-USERAGENT": 90,
237
+ "I-USERNAME": 12,
238
+ "I-VEHICLEVIN": 32,
239
  "I-VEHICLEVRM": 14,
240
+ "I-ZIPCODE": 34,
241
  "O": 0
242
  },
243
  "layer_norm_eps": 1e-12,
244
  "max_position_embeddings": 512,
245
  "model_type": "bert",
246
+ "num_attention_heads": 8,
247
+ "num_hidden_layers": 8,
248
  "pad_token_id": 0,
249
  "position_embedding_type": "absolute",
250
  "torch_dtype": "float32",
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:64b35549be667c1549d862037915c5ab65bef0f34959b026a5a694ea5f28010e
3
- size 44543364
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d87fc414acf1c35a092e7916b123b2c9f13b170a776d57085ca4a1fc75dad92
3
+ size 164689684
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:66fadb07e9910f11b222583ea9852888ce5978e69f6c23574216c6c9123a197f
3
  size 4664
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8b5f8d535a948fd87ab65334e081ce03e2bdfebfc2be73110b4f9a5ad73b99f
3
  size 4664