Upload 13 files

added_tokens.json

@@ -0,0 +1,7 @@
+{
+  "[CLS]": 1,
+  "[MASK]": 50264,
+  "[PAD]": 0,
+  "[SEP]": 2,
+  "[UNK]": 3
+}

config.json

@@ -0,0 +1,268 @@
+{
+  "_name_or_path": "microsoft/deberta-base",
+  "architectures": [
+    "DebertaForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "B-PREFIX",
+    "1": "I-PREFIX",
+    "2": "B-FIRSTNAME",
+    "3": "I-FIRSTNAME",
+    "4": "B-MIDDLENAME",
+    "5": "B-LASTNAME",
+    "6": "I-LASTNAME",
+    "7": "O",
+    "8": "B-JOBDESCRIPTOR",
+    "9": "B-JOBTITLE",
+    "10": "I-JOBTITLE",
+    "11": "B-COMPANY_NAME",
+    "12": "I-COMPANY_NAME",
+    "13": "B-JOBAREA",
+    "14": "B-EMAIL",
+    "15": "I-EMAIL",
+    "16": "B-TIME",
+    "17": "I-TIME",
+    "18": "B-DATE",
+    "19": "I-DATE",
+    "20": "B-URL",
+    "21": "I-URL",
+    "22": "B-BITCOINADDRESS",
+    "23": "I-BITCOINADDRESS",
+    "24": "B-ETHEREUMADDRESS",
+    "25": "I-ETHEREUMADDRESS",
+    "26": "B-ACCOUNTNAME",
+    "27": "I-ACCOUNTNAME",
+    "28": "B-IBAN",
+    "29": "I-IBAN",
+    "30": "B-ACCOUNTNUMBER",
+    "31": "I-ACCOUNTNUMBER",
+    "32": "B-BIC",
+    "33": "I-BIC",
+    "34": "B-IPV4",
+    "35": "I-IPV4",
+    "36": "B-STREETADDRESS",
+    "37": "I-STREETADDRESS",
+    "38": "B-CITY",
+    "39": "I-CITY",
+    "40": "B-ZIPCODE",
+    "41": "I-ZIPCODE",
+    "42": "B-USERNAME",
+    "43": "I-USERNAME",
+    "44": "B-IPV6",
+    "45": "I-IPV6",
+    "46": "B-CREDITCARDNUMBER",
+    "47": "I-CREDITCARDNUMBER",
+    "48": "B-VEHICLEVIN",
+    "49": "I-VEHICLEVIN",
+    "50": "B-SUFFIX",
+    "51": "I-SUFFIX",
+    "52": "B-AMOUNT",
+    "53": "I-AMOUNT",
+    "54": "B-CURRENCY",
+    "55": "I-CURRENCY",
+    "56": "B-PASSWORD",
+    "57": "I-PASSWORD",
+    "58": "B-JOBTYPE",
+    "59": "B-STATE",
+    "60": "B-BUILDINGNUMBER",
+    "61": "I-BUILDINGNUMBER",
+    "62": "B-VEHICLEVRM",
+    "63": "I-VEHICLEVRM",
+    "64": "B-PHONEIMEI",
+    "65": "I-PHONEIMEI",
+    "66": "I-JOBAREA",
+    "67": "I-STATE",
+    "68": "B-COUNTY",
+    "69": "B-CURRENCYNAME",
+    "70": "I-CURRENCYNAME",
+    "71": "B-CURRENCYSYMBOL",
+    "72": "B-MASKEDNUMBER",
+    "73": "I-MASKEDNUMBER",
+    "74": "B-PHONE_NUMBER",
+    "75": "I-PHONE_NUMBER",
+    "76": "B-SECONDARYADDRESS",
+    "77": "I-SECONDARYADDRESS",
+    "78": "B-SSN",
+    "79": "I-SSN",
+    "80": "B-CURRENCYCODE",
+    "81": "B-LITECOINADDRESS",
+    "82": "I-LITECOINADDRESS",
+    "83": "B-MAC",
+    "84": "I-MAC",
+    "85": "B-CREDITCARDISSUER",
+    "86": "I-CREDITCARDISSUER",
+    "87": "B-CREDITCARDCVV",
+    "88": "I-CREDITCARDCVV",
+    "89": "B-USERAGENT",
+    "90": "I-USERAGENT",
+    "91": "B-IP",
+    "92": "I-IP",
+    "93": "B-SEX",
+    "94": "B-STREET",
+    "95": "I-STREET",
+    "96": "B-PIN",
+    "97": "I-PIN",
+    "98": "I-JOBTYPE",
+    "99": "I-MIDDLENAME",
+    "100": "I-CURRENCYCODE",
+    "101": "I-CURRENCYSYMBOL",
+    "102": "B-FULLNAME",
+    "103": "I-FULLNAME",
+    "104": "B-NAME",
+    "105": "I-NAME",
+    "106": "B-GENDER",
+    "107": "B-NUMBER",
+    "108": "I-NUMBER",
+    "109": "I-GENDER",
+    "110": "B-NEARBYGPSCOORDINATE",
+    "111": "I-NEARBYGPSCOORDINATE",
+    "112": "B-DISPLAYNAME",
+    "113": "I-DISPLAYNAME",
+    "114": "B-SEXTYPE",
+    "115": "B-ORDINALDIRECTION"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "B-ACCOUNTNAME": 26,
+    "B-ACCOUNTNUMBER": 30,
+    "B-AMOUNT": 52,
+    "B-BIC": 32,
+    "B-BITCOINADDRESS": 22,
+    "B-BUILDINGNUMBER": 60,
+    "B-CITY": 38,
+    "B-COMPANY_NAME": 11,
+    "B-COUNTY": 68,
+    "B-CREDITCARDCVV": 87,
+    "B-CREDITCARDISSUER": 85,
+    "B-CREDITCARDNUMBER": 46,
+    "B-CURRENCY": 54,
+    "B-CURRENCYCODE": 80,
+    "B-CURRENCYNAME": 69,
+    "B-CURRENCYSYMBOL": 71,
+    "B-DATE": 18,
+    "B-DISPLAYNAME": 112,
+    "B-EMAIL": 14,
+    "B-ETHEREUMADDRESS": 24,
+    "B-FIRSTNAME": 2,
+    "B-FULLNAME": 102,
+    "B-GENDER": 106,
+    "B-IBAN": 28,
+    "B-IP": 91,
+    "B-IPV4": 34,
+    "B-IPV6": 44,
+    "B-JOBAREA": 13,
+    "B-JOBDESCRIPTOR": 8,
+    "B-JOBTITLE": 9,
+    "B-JOBTYPE": 58,
+    "B-LASTNAME": 5,
+    "B-LITECOINADDRESS": 81,
+    "B-MAC": 83,
+    "B-MASKEDNUMBER": 72,
+    "B-MIDDLENAME": 4,
+    "B-NAME": 104,
+    "B-NEARBYGPSCOORDINATE": 110,
+    "B-NUMBER": 107,
+    "B-ORDINALDIRECTION": 115,
+    "B-PASSWORD": 56,
+    "B-PHONEIMEI": 64,
+    "B-PHONE_NUMBER": 74,
+    "B-PIN": 96,
+    "B-PREFIX": 0,
+    "B-SECONDARYADDRESS": 76,
+    "B-SEX": 93,
+    "B-SEXTYPE": 114,
+    "B-SSN": 78,
+    "B-STATE": 59,
+    "B-STREET": 94,
+    "B-STREETADDRESS": 36,
+    "B-SUFFIX": 50,
+    "B-TIME": 16,
+    "B-URL": 20,
+    "B-USERAGENT": 89,
+    "B-USERNAME": 42,
+    "B-VEHICLEVIN": 48,
+    "B-VEHICLEVRM": 62,
+    "B-ZIPCODE": 40,
+    "I-ACCOUNTNAME": 27,
+    "I-ACCOUNTNUMBER": 31,
+    "I-AMOUNT": 53,
+    "I-BIC": 33,
+    "I-BITCOINADDRESS": 23,
+    "I-BUILDINGNUMBER": 61,
+    "I-CITY": 39,
+    "I-COMPANY_NAME": 12,
+    "I-CREDITCARDCVV": 88,
+    "I-CREDITCARDISSUER": 86,
+    "I-CREDITCARDNUMBER": 47,
+    "I-CURRENCY": 55,
+    "I-CURRENCYCODE": 100,
+    "I-CURRENCYNAME": 70,
+    "I-CURRENCYSYMBOL": 101,
+    "I-DATE": 19,
+    "I-DISPLAYNAME": 113,
+    "I-EMAIL": 15,
+    "I-ETHEREUMADDRESS": 25,
+    "I-FIRSTNAME": 3,
+    "I-FULLNAME": 103,
+    "I-GENDER": 109,
+    "I-IBAN": 29,
+    "I-IP": 92,
+    "I-IPV4": 35,
+    "I-IPV6": 45,
+    "I-JOBAREA": 66,
+    "I-JOBTITLE": 10,
+    "I-JOBTYPE": 98,
+    "I-LASTNAME": 6,
+    "I-LITECOINADDRESS": 82,
+    "I-MAC": 84,
+    "I-MASKEDNUMBER": 73,
+    "I-MIDDLENAME": 99,
+    "I-NAME": 105,
+    "I-NEARBYGPSCOORDINATE": 111,
+    "I-NUMBER": 108,
+    "I-PASSWORD": 57,
+    "I-PHONEIMEI": 65,
+    "I-PHONE_NUMBER": 75,
+    "I-PIN": 97,
+    "I-PREFIX": 1,
+    "I-SECONDARYADDRESS": 77,
+    "I-SSN": 79,
+    "I-STATE": 67,
+    "I-STREET": 95,
+    "I-STREETADDRESS": 37,
+    "I-SUFFIX": 51,
+    "I-TIME": 17,
+    "I-URL": 21,
+    "I-USERAGENT": 90,
+    "I-USERNAME": 43,
+    "I-VEHICLEVIN": 49,
+    "I-VEHICLEVRM": 63,
+    "I-ZIPCODE": 41,
+    "O": 7
+  },
+  "layer_norm_eps": 1e-07,
+  "max_position_embeddings": 512,
+  "max_relative_positions": -1,
+  "model_type": "deberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "pooler_dropout": 0,
+  "pooler_hidden_act": "gelu",
+  "pooler_hidden_size": 768,
+  "pos_att_type": [
+    "c2p",
+    "p2c"
+  ],
+  "position_biased_input": false,
+  "relative_attention": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.34.0",
+  "type_vocab_size": 0,
+  "vocab_size": 50265
+}

optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:549cfaf98d70c5d1b9228fde16d7f5049e4b6a75915460ced13aab0217da656e
+size 1109692613

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6539218f24a2ebcaf67665049f657d0ee61afb44b856588a0efe3ca8c33a8cf2
+size 554831781

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:463ca5ebc7cb3d40f145ee439ba1e616e46262e1cf05052699c423b2150a17cd
+size 14575

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:257d4e5b87ed6c145ab2d02ea57f1cb8736e2bb139099f5b69120642f82b0a20
+size 627

special_tokens_map.json

@@ -0,0 +1,9 @@
+{
+  "bos_token": "[CLS]",
+  "cls_token": "[CLS]",
+  "eos_token": "[SEP]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json

@@ -0,0 +1,60 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": true,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50264": {
+      "content": "[MASK]",
+      "lstrip": true,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "bos_token": "[CLS]",
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_lower_case": false,
+  "eos_token": "[SEP]",
+  "errors": "replace",
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "tokenizer_class": "DebertaTokenizer",
+  "unk_token": "[UNK]",
+  "vocab_type": "gpt2"
+}

trainer_state.json

@@ -0,0 +1,618 @@
+{
+  "best_metric": 0.008737307973206043,
+  "best_model_checkpoint": "drive/MyDrive/colabFiles/safellm/aiPII/model/deberta_finetuned/checkpoint-9359",
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 9359,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3354700854700855e-05,
+      "loss": 1.5328,
+      "step": 500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.670940170940171e-05,
+      "loss": 0.3157,
+      "step": 1000
+    },
+    {
+      "epoch": 1.0,
+      "eval_ACCOUNTNAME_f1": 0.8498583569405098,
+      "eval_ACCOUNTNUMBER_f1": 0.7777777777777778,
+      "eval_AMOUNT_f1": 0.7830423940149626,
+      "eval_BIC_f1": 0.10999999999999999,
+      "eval_BITCOINADDRESS_f1": 0.0489795918367347,
+      "eval_BUILDINGNUMBER_f1": 0.2996941896024465,
+      "eval_CITY_f1": 0.9399141630901287,
+      "eval_COMPANY_NAME_f1": 0.3222222222222222,
+      "eval_COUNTY_f1": 0.9363207547169812,
+      "eval_CREDITCARDCVV_f1": 0.0,
+      "eval_CREDITCARDISSUER_f1": 0.0916030534351145,
+      "eval_CREDITCARDNUMBER_f1": 0.6247379454926625,
+      "eval_CURRENCYCODE_f1": 0.0,
+      "eval_CURRENCYNAME_f1": 0.008658008658008658,
+      "eval_CURRENCYSYMBOL_f1": 0.0,
+      "eval_CURRENCY_f1": 0.4955156950672646,
+      "eval_DATE_f1": 0.8909952606635071,
+      "eval_DISPLAYNAME_f1": 0.0,
+      "eval_EMAIL_f1": 0.9967564060979566,
+      "eval_ETHEREUMADDRESS_f1": 0.7978142076502732,
+      "eval_FIRSTNAME_f1": 0.7487401007919366,
+      "eval_FULLNAME_f1": 0.9725754176015956,
+      "eval_GENDER_f1": 0.5737704918032787,
+      "eval_IBAN_f1": 0.5320197044334977,
+      "eval_IPV4_f1": 0.04864864864864866,
+      "eval_IPV6_f1": 0.6883720930232559,
+      "eval_IP_f1": 0.004514672686230249,
+      "eval_JOBAREA_f1": 0.8751714677640604,
+      "eval_JOBDESCRIPTOR_f1": 0.08,
+      "eval_JOBTITLE_f1": 0.7678571428571428,
+      "eval_JOBTYPE_f1": 0.18320610687022904,
+      "eval_LASTNAME_f1": 0.3830787309048179,
+      "eval_LITECOINADDRESS_f1": 0.03690036900369004,
+      "eval_MAC_f1": 0.588235294117647,
+      "eval_MASKEDNUMBER_f1": 0.0,
+      "eval_MIDDLENAME_f1": 0.0,
+      "eval_NAME_f1": 0.9682713347921226,
+      "eval_NEARBYGPSCOORDINATE_f1": 0.0,
+      "eval_NUMBER_f1": 0.0,
+      "eval_PASSWORD_f1": 0.6082289803220037,
+      "eval_PHONEIMEI_f1": 0.911242603550296,
+      "eval_PHONE_NUMBER_f1": 0.42622950819672123,
+      "eval_PIN_f1": 0.014184397163120567,
+      "eval_PREFIX_f1": 0.8070175438596491,
+      "eval_SECONDARYADDRESS_f1": 0.7699530516431925,
+      "eval_SEXTYPE_f1": 0.0,
+      "eval_SEX_f1": 0.8055555555555556,
+      "eval_SSN_f1": 0.013605442176870748,
+      "eval_STATE_f1": 0.9252669039145908,
+      "eval_STREETADDRESS_f1": 0.7314578005115089,
+      "eval_STREET_f1": 0.38073394495412843,
+      "eval_SUFFIX_f1": 0.0,
+      "eval_TIME_f1": 0.9922480620155039,
+      "eval_URL_f1": 0.9927431059506532,
+      "eval_USERAGENT_f1": 0.9321266968325792,
+      "eval_USERNAME_f1": 0.7885597548518897,
+      "eval_VEHICLEVIN_f1": 0.0,
+      "eval_VEHICLEVRM_f1": 0.2533333333333333,
+      "eval_ZIPCODE_f1": 0.7258064516129034,
+      "eval_loss": 0.14694246649742126,
+      "eval_overall_accuracy": 0.9484817248102932,
+      "eval_overall_f1": 0.8294381247845569,
+      "eval_overall_precision": 0.8040499899752723,
+      "eval_overall_recall": 0.8564818110628604,
+      "eval_runtime": 90.4464,
+      "eval_samples_per_second": 118.225,
+      "eval_steps_per_second": 14.782,
+      "step": 1337
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.006410256410257e-05,
+      "loss": 0.2134,
+      "step": 1500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.914518498731134e-05,
+      "loss": 0.1542,
+      "step": 2000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.580606384399626e-05,
+      "loss": 0.1231,
+      "step": 2500
+    },
+    {
+      "epoch": 2.0,
+      "eval_ACCOUNTNAME_f1": 0.9712460063897763,
+      "eval_ACCOUNTNUMBER_f1": 0.9777777777777777,
+      "eval_AMOUNT_f1": 0.9312169312169312,
+      "eval_BIC_f1": 0.8363636363636363,
+      "eval_BITCOINADDRESS_f1": 0.7272727272727273,
+      "eval_BUILDINGNUMBER_f1": 0.6622950819672131,
+      "eval_CITY_f1": 0.968736239542052,
+      "eval_COMPANY_NAME_f1": 0.874074074074074,
+      "eval_COUNTY_f1": 0.994413407821229,
+      "eval_CREDITCARDCVV_f1": 0.8474576271186439,
+      "eval_CREDITCARDISSUER_f1": 0.7811158798283261,
+      "eval_CREDITCARDNUMBER_f1": 0.09195402298850575,
+      "eval_CURRENCYCODE_f1": 0.564102564102564,
+      "eval_CURRENCYNAME_f1": 0.24761904761904763,
+      "eval_CURRENCYSYMBOL_f1": 0.5306122448979592,
+      "eval_CURRENCY_f1": 0.6920152091254752,
+      "eval_DATE_f1": 0.9928057553956835,
+      "eval_DISPLAYNAME_f1": 0.0,
+      "eval_EMAIL_f1": 0.9993498049414824,
+      "eval_ETHEREUMADDRESS_f1": 0.9528795811518325,
+      "eval_FIRSTNAME_f1": 0.7688584106494555,
+      "eval_FULLNAME_f1": 0.9868637110016422,
+      "eval_GENDER_f1": 0.9075630252100839,
+      "eval_IBAN_f1": 0.9468599033816426,
+      "eval_IPV4_f1": 0.84375,
+      "eval_IPV6_f1": 0.7846153846153846,
+      "eval_IP_f1": 0.0,
+      "eval_JOBAREA_f1": 0.9410050983248361,
+      "eval_JOBDESCRIPTOR_f1": 0.5128205128205128,
+      "eval_JOBTITLE_f1": 0.9441233140655106,
+      "eval_JOBTYPE_f1": 0.7578125000000001,
+      "eval_LASTNAME_f1": 0.5283422459893049,
+      "eval_LITECOINADDRESS_f1": 0.7647058823529412,
+      "eval_MAC_f1": 0.9702970297029703,
+      "eval_MASKEDNUMBER_f1": 0.3492063492063492,
+      "eval_MIDDLENAME_f1": 0.3691460055096419,
+      "eval_NAME_f1": 0.988691888873377,
+      "eval_NUMBER_f1": 0.47393364928909953,
+      "eval_PASSWORD_f1": 0.9886363636363638,
+      "eval_PHONEIMEI_f1": 0.9629629629629629,
+      "eval_PHONE_NUMBER_f1": 0.8366533864541832,
+      "eval_PIN_f1": 0.7967479674796748,
+      "eval_PREFIX_f1": 0.8198757763975155,
+      "eval_SECONDARYADDRESS_f1": 0.9265536723163842,
+      "eval_SEXTYPE_f1": 0.0,
+      "eval_SEX_f1": 0.8941176470588235,
+      "eval_SSN_f1": 0.7486033519553071,
+      "eval_STATE_f1": 0.9692487349163099,
+      "eval_STREETADDRESS_f1": 0.8785310734463277,
+      "eval_STREET_f1": 0.5085714285714286,
+      "eval_SUFFIX_f1": 0.5306122448979592,
+      "eval_TIME_f1": 0.942528735632184,
+      "eval_URL_f1": 0.9738372093023255,
+      "eval_USERAGENT_f1": 0.912442396313364,
+      "eval_USERNAME_f1": 0.9172625127681308,
+      "eval_VEHICLEVIN_f1": 0.8368794326241135,
+      "eval_VEHICLEVRM_f1": 0.8051948051948051,
+      "eval_ZIPCODE_f1": 0.8036117381489842,
+      "eval_loss": 0.08253061771392822,
+      "eval_overall_accuracy": 0.9728585799624934,
+      "eval_overall_f1": 0.9135649808334365,
+      "eval_overall_precision": 0.9068209714185517,
+      "eval_overall_recall": 0.9204100519683919,
+      "eval_runtime": 83.6232,
+      "eval_samples_per_second": 127.871,
+      "eval_steps_per_second": 15.988,
+      "step": 2674
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.2466942700681185e-05,
+      "loss": 0.099,
+      "step": 3000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.91278215573661e-05,
+      "loss": 0.0815,
+      "step": 3500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.578870041405102e-05,
+      "loss": 0.0745,
+      "step": 4000
+    },
+    {
+      "epoch": 3.0,
+      "eval_ACCOUNTNAME_f1": 0.9967426710097721,
+      "eval_ACCOUNTNUMBER_f1": 0.9821428571428572,
+      "eval_AMOUNT_f1": 0.957983193277311,
+      "eval_BIC_f1": 0.9054054054054054,
+      "eval_BITCOINADDRESS_f1": 0.8984375,
+      "eval_BUILDINGNUMBER_f1": 0.7807807807807808,
+      "eval_CITY_f1": 0.9926682959342368,
+      "eval_COMPANY_NAME_f1": 0.9468354430379746,
+      "eval_COUNTY_f1": 0.9977728285077951,
+      "eval_CREDITCARDCVV_f1": 0.912,
+      "eval_CREDITCARDISSUER_f1": 0.9665071770334929,
+      "eval_CREDITCARDNUMBER_f1": 0.9044585987261147,
+      "eval_CURRENCYCODE_f1": 0.9079754601226994,
+      "eval_CURRENCYNAME_f1": 0.25837320574162675,
+      "eval_CURRENCYSYMBOL_f1": 0.762589928057554,
+      "eval_CURRENCY_f1": 0.8146341463414634,
+      "eval_DATE_f1": 0.9976019184652278,
+      "eval_DISPLAYNAME_f1": 0.5544554455445545,
+      "eval_EMAIL_f1": 1.0,
+      "eval_ETHEREUMADDRESS_f1": 0.9837837837837838,
+      "eval_FIRSTNAME_f1": 0.8647106764466178,
+      "eval_FULLNAME_f1": 0.9867741529159844,
+      "eval_GENDER_f1": 0.9914529914529915,
+      "eval_IBAN_f1": 0.9751243781094527,
+      "eval_IPV4_f1": 0.852017937219731,
+      "eval_IPV6_f1": 0.44791666666666663,
+      "eval_IP_f1": 0.0,
+      "eval_JOBAREA_f1": 0.9867963863794301,
+      "eval_JOBDESCRIPTOR_f1": 0.8383838383838385,
+      "eval_JOBTITLE_f1": 0.9901380670611439,
+      "eval_JOBTYPE_f1": 0.7729468599033816,
+      "eval_LASTNAME_f1": 0.6699147381242387,
+      "eval_LITECOINADDRESS_f1": 0.84472049689441,
+      "eval_MAC_f1": 0.9565217391304348,
+      "eval_MASKEDNUMBER_f1": 0.8143712574850299,
+      "eval_MIDDLENAME_f1": 0.702355460385439,
+      "eval_NAME_f1": 0.9959355290819901,
+      "eval_NUMBER_f1": 0.7282051282051282,
+      "eval_PASSWORD_f1": 0.9725274725274725,
+      "eval_PHONEIMEI_f1": 0.963855421686747,
+      "eval_PHONE_NUMBER_f1": 0.8472222222222221,
+      "eval_PIN_f1": 0.8648648648648648,
+      "eval_PREFIX_f1": 0.5957446808510638,
+      "eval_SECONDARYADDRESS_f1": 0.9883720930232558,
+      "eval_SEXTYPE_f1": 0.8571428571428571,
+      "eval_SEX_f1": 0.945945945945946,
+      "eval_SSN_f1": 0.8701298701298701,
+      "eval_STATE_f1": 0.9960222752585521,
+      "eval_STREETADDRESS_f1": 0.904899135446686,
+      "eval_STREET_f1": 0.7334963325183375,
+      "eval_SUFFIX_f1": 0.7956989247311829,
+      "eval_TIME_f1": 0.9883268482490272,
+      "eval_URL_f1": 0.9941860465116279,
+      "eval_USERAGENT_f1": 0.9724770642201835,
+      "eval_USERNAME_f1": 0.9451476793248945,
+      "eval_VEHICLEVIN_f1": 0.9635036496350364,
+      "eval_VEHICLEVRM_f1": 0.9261744966442953,
+      "eval_ZIPCODE_f1": 0.9254498714652957,
+      "eval_loss": 0.04885776713490486,
+      "eval_overall_accuracy": 0.9795129099963984,
+      "eval_overall_f1": 0.9467927822013217,
+      "eval_overall_precision": 0.9476198965947584,
+      "eval_overall_recall": 0.9459671104150352,
+      "eval_runtime": 85.058,
+      "eval_samples_per_second": 125.714,
+      "eval_steps_per_second": 15.719,
+      "step": 4011
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.244957927073594e-05,
+      "loss": 0.0544,
+      "step": 4500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.9110458127420864e-05,
+      "loss": 0.0511,
+      "step": 5000
+    },
+    {
+      "epoch": 4.0,
+      "eval_ACCOUNTNAME_f1": 0.9901639344262295,
+      "eval_ACCOUNTNUMBER_f1": 1.0,
+      "eval_AMOUNT_f1": 0.24609374999999997,
+      "eval_BIC_f1": 0.9594594594594594,
+      "eval_BITCOINADDRESS_f1": 0.8837209302325583,
+      "eval_BUILDINGNUMBER_f1": 0.8328445747800587,
+      "eval_CITY_f1": 0.9973214285714286,
+      "eval_COMPANY_NAME_f1": 0.9815303430079155,
+      "eval_COUNTY_f1": 1.0,
+      "eval_CREDITCARDCVV_f1": 0.991869918699187,
+      "eval_CREDITCARDISSUER_f1": 0.9439252336448598,
+      "eval_CREDITCARDNUMBER_f1": 0.977198697068404,
+      "eval_CURRENCYCODE_f1": 0.9333333333333333,
+      "eval_CURRENCYNAME_f1": 0.7027027027027026,
+      "eval_CURRENCYSYMBOL_f1": 0.5191489361702128,
+      "eval_CURRENCY_f1": 0.8871181938911021,
+      "eval_DATE_f1": 0.9976019184652278,
+      "eval_DISPLAYNAME_f1": 0.6153846153846154,
+      "eval_EMAIL_f1": 1.0,
+      "eval_ETHEREUMADDRESS_f1": 0.9726775956284153,
+      "eval_FIRSTNAME_f1": 0.8918918918918919,
+      "eval_FULLNAME_f1": 0.9890151515151515,
+      "eval_GENDER_f1": 0.9827586206896551,
+      "eval_IBAN_f1": 0.9807692307692307,
+      "eval_IPV4_f1": 0.8565022421524664,
+      "eval_IPV6_f1": 0.851063829787234,
+      "eval_IP_f1": 0.031746031746031744,
+      "eval_JOBAREA_f1": 0.9901823281907434,
+      "eval_JOBDESCRIPTOR_f1": 0.9781659388646288,
+      "eval_JOBTITLE_f1": 0.9980119284294234,
+      "eval_JOBTYPE_f1": 0.9831932773109243,
+      "eval_LASTNAME_f1": 0.8038507821901323,
+      "eval_LITECOINADDRESS_f1": 0.782122905027933,
+      "eval_MAC_f1": 0.9753694581280787,
+      "eval_MASKEDNUMBER_f1": 0.9554140127388535,
+      "eval_MIDDLENAME_f1": 0.8237232289950577,
+      "eval_NAME_f1": 0.9990197451337348,
+      "eval_NUMBER_f1": 0.9807692307692307,
+      "eval_PASSWORD_f1": 1.0,
+      "eval_PHONEIMEI_f1": 1.0,
+      "eval_PHONE_NUMBER_f1": 0.9880478087649401,
+      "eval_PIN_f1": 0.8679245283018869,
+      "eval_PREFIX_f1": 0.920152091254753,
+      "eval_SECONDARYADDRESS_f1": 0.9826589595375722,
+      "eval_SEXTYPE_f1": 0.8571428571428571,
+      "eval_SEX_f1": 0.9673202614379085,
+      "eval_SSN_f1": 0.987012987012987,
+      "eval_STATE_f1": 0.9948186528497409,
+      "eval_STREETADDRESS_f1": 0.911208151382824,
+      "eval_STREET_f1": 0.7952941176470588,
+      "eval_SUFFIX_f1": 0.9655172413793104,
+      "eval_TIME_f1": 0.9921875,
+      "eval_URL_f1": 1.0,
+      "eval_USERAGENT_f1": 1.0,
+      "eval_USERNAME_f1": 0.9560557341907825,
+      "eval_VEHICLEVIN_f1": 0.962962962962963,
+      "eval_VEHICLEVRM_f1": 0.9784172661870504,
+      "eval_ZIPCODE_f1": 0.9973045822102425,
+      "eval_loss": 0.033664677292108536,
+      "eval_overall_accuracy": 0.9870067934276382,
+      "eval_overall_f1": 0.9611536824180503,
+      "eval_overall_precision": 0.9579591259458313,
+      "eval_overall_recall": 0.9643696162881754,
+      "eval_runtime": 88.9305,
+      "eval_samples_per_second": 120.24,
+      "eval_steps_per_second": 15.034,
+      "step": 5348
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.577133698410579e-05,
+      "loss": 0.0491,
+      "step": 5500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.2432215840790704e-05,
+      "loss": 0.0349,
+      "step": 6000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.9093094697475625e-05,
+      "loss": 0.0324,
+      "step": 6500
+    },
+    {
+      "epoch": 5.0,
+      "eval_ACCOUNTNAME_f1": 0.9967426710097721,
+      "eval_ACCOUNTNUMBER_f1": 1.0,
+      "eval_AMOUNT_f1": 1.0,
+      "eval_BIC_f1": 1.0,
+      "eval_BITCOINADDRESS_f1": 0.982905982905983,
+      "eval_BUILDINGNUMBER_f1": 0.959349593495935,
+      "eval_CITY_f1": 0.9951197870452528,
+      "eval_COMPANY_NAME_f1": 0.9948186528497409,
+      "eval_COUNTY_f1": 1.0,
+      "eval_CREDITCARDCVV_f1": 1.0,
+      "eval_CREDITCARDISSUER_f1": 1.0,
+      "eval_CREDITCARDNUMBER_f1": 0.9803921568627451,
+      "eval_CURRENCYCODE_f1": 0.9473684210526315,
+      "eval_CURRENCYNAME_f1": 0.8990825688073395,
+      "eval_CURRENCYSYMBOL_f1": 0.9051094890510949,
+      "eval_CURRENCY_f1": 0.9527896995708155,
+      "eval_DATE_f1": 1.0,
+      "eval_DISPLAYNAME_f1": 0.7719298245614035,
+      "eval_EMAIL_f1": 0.9990247074122237,
+      "eval_ETHEREUMADDRESS_f1": 1.0,
+      "eval_FIRSTNAME_f1": 0.9494071146245059,
+      "eval_FULLNAME_f1": 0.9937848807711821,
+      "eval_GENDER_f1": 1.0,
+      "eval_IBAN_f1": 1.0,
+      "eval_IPV4_f1": 0.8397291196388262,
+      "eval_IPV6_f1": 0.8426666666666667,
+      "eval_IP_f1": 0.05194805194805195,
+      "eval_JOBAREA_f1": 0.9965059399021663,
+      "eval_JOBDESCRIPTOR_f1": 0.9736842105263158,
+      "eval_JOBTITLE_f1": 0.9960159362549801,
+      "eval_JOBTYPE_f1": 0.9957805907172996,
+      "eval_LASTNAME_f1": 0.8947368421052632,
+      "eval_LITECOINADDRESS_f1": 0.918918918918919,
+      "eval_MAC_f1": 0.9900990099009901,
+      "eval_MASKEDNUMBER_f1": 0.9620253164556961,
+      "eval_MIDDLENAME_f1": 0.9489603024574669,
+      "eval_NAME_f1": 0.9994396189408798,
+      "eval_NUMBER_f1": 0.9951690821256038,
+      "eval_PASSWORD_f1": 0.9752066115702479,
+      "eval_PHONEIMEI_f1": 0.9937888198757764,
+      "eval_PHONE_NUMBER_f1": 0.9960159362549801,
+      "eval_PIN_f1": 0.9906542056074767,
+      "eval_PREFIX_f1": 0.956989247311828,
+      "eval_SECONDARYADDRESS_f1": 0.9883720930232558,
+      "eval_SEXTYPE_f1": 0.8888888888888888,
+      "eval_SEX_f1": 0.9871794871794872,
+      "eval_SSN_f1": 1.0,
+      "eval_STATE_f1": 0.9988023952095808,
+      "eval_STREETADDRESS_f1": 0.9765258215962442,
+      "eval_STREET_f1": 0.9037656903765691,
+      "eval_SUFFIX_f1": 0.970873786407767,
+      "eval_TIME_f1": 0.9961089494163424,
+      "eval_URL_f1": 1.0,
+      "eval_USERAGENT_f1": 0.9908256880733946,
+      "eval_USERNAME_f1": 0.9837133550488599,
+      "eval_VEHICLEVIN_f1": 1.0,
+      "eval_VEHICLEVRM_f1": 1.0,
+      "eval_ZIPCODE_f1": 0.9973045822102425,
+      "eval_loss": 0.022747300565242767,
+      "eval_overall_accuracy": 0.9911275599547933,
+      "eval_overall_f1": 0.9828670955817012,
+      "eval_overall_precision": 0.9815064603152066,
+      "eval_overall_recall": 0.9842315085071546,
+      "eval_runtime": 84.4935,
+      "eval_samples_per_second": 126.554,
+      "eval_steps_per_second": 15.824,
+      "step": 6685
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5753973554160546e-05,
+      "loss": 0.0297,
+      "step": 7000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.2414852410845466e-05,
+      "loss": 0.0283,
+      "step": 7500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 9.075731267530386e-06,
+      "loss": 0.0211,
+      "step": 8000
+    },
+    {
+      "epoch": 6.0,
+      "eval_ACCOUNTNAME_f1": 1.0,
+      "eval_ACCOUNTNUMBER_f1": 1.0,
+      "eval_AMOUNT_f1": 1.0,
+      "eval_BIC_f1": 1.0,
+      "eval_BITCOINADDRESS_f1": 1.0,
+      "eval_BUILDINGNUMBER_f1": 0.9973045822102425,
+      "eval_CITY_f1": 0.9988856697125027,
+      "eval_COMPANY_NAME_f1": 1.0,
+      "eval_COUNTY_f1": 1.0,
+      "eval_CREDITCARDCVV_f1": 1.0,
+      "eval_CREDITCARDISSUER_f1": 1.0,
+      "eval_CREDITCARDNUMBER_f1": 0.9803921568627451,
+      "eval_CURRENCYCODE_f1": 0.9878048780487805,
+      "eval_CURRENCYNAME_f1": 0.9207317073170731,
+      "eval_CURRENCYSYMBOL_f1": 0.9855072463768116,
+      "eval_CURRENCY_f1": 0.972818311874106,
+      "eval_DATE_f1": 1.0,
+      "eval_DISPLAYNAME_f1": 0.9291338582677166,
+      "eval_EMAIL_f1": 1.0,
+      "eval_ETHEREUMADDRESS_f1": 1.0,
+      "eval_FIRSTNAME_f1": 0.9910261412407336,
+      "eval_FULLNAME_f1": 0.9982799261005288,
+      "eval_GENDER_f1": 1.0,
+      "eval_IBAN_f1": 0.9901960784313726,
+      "eval_IPV4_f1": 0.8558352402745996,
+      "eval_IPV6_f1": 0.8465608465608466,
+      "eval_IP_f1": 0.14117647058823532,
+      "eval_JOBAREA_f1": 0.9985994397759104,
+      "eval_JOBDESCRIPTOR_f1": 1.0,
+      "eval_JOBTITLE_f1": 1.0,
+      "eval_JOBTYPE_f1": 0.9957805907172996,
+      "eval_LASTNAME_f1": 0.9753483386923902,
+      "eval_LITECOINADDRESS_f1": 0.9943502824858756,
+      "eval_MAC_f1": 1.0,
+      "eval_MASKEDNUMBER_f1": 0.9625,
+      "eval_MIDDLENAME_f1": 0.9834254143646408,
+      "eval_NAME_f1": 0.9994396189408798,
+      "eval_NUMBER_f1": 1.0,
+      "eval_PASSWORD_f1": 0.9971830985915493,
+      "eval_PHONEIMEI_f1": 1.0,
+      "eval_PHONE_NUMBER_f1": 1.0,
+      "eval_PIN_f1": 1.0,
+      "eval_PREFIX_f1": 0.9757009345794393,
+      "eval_SECONDARYADDRESS_f1": 1.0,
+      "eval_SEXTYPE_f1": 0.9473684210526316,
+      "eval_SEX_f1": 0.9871794871794872,
+      "eval_SSN_f1": 1.0,
+      "eval_STATE_f1": 0.9988023952095808,
+      "eval_STREETADDRESS_f1": 0.9984152139461173,
+      "eval_STREET_f1": 0.9878048780487805,
+      "eval_SUFFIX_f1": 0.9855072463768114,
+      "eval_TIME_f1": 1.0,
+      "eval_URL_f1": 1.0,
+      "eval_USERAGENT_f1": 1.0,
+      "eval_USERNAME_f1": 0.9944873208379272,
+      "eval_VEHICLEVIN_f1": 1.0,
+      "eval_VEHICLEVRM_f1": 1.0,
+      "eval_ZIPCODE_f1": 1.0,
+      "eval_loss": 0.01404440775513649,
+      "eval_overall_accuracy": 0.992913473838473,
+      "eval_overall_f1": 0.9916438502293495,
+      "eval_overall_precision": 0.9906223358908781,
+      "eval_overall_recall": 0.9926674734818822,
+      "eval_runtime": 88.6108,
+      "eval_samples_per_second": 120.674,
+      "eval_steps_per_second": 15.088,
+      "step": 8022
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.736610124215307e-06,
+      "loss": 0.0147,
+      "step": 8500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.3974889809002272e-06,
+      "loss": 0.0141,
+      "step": 9000
+    },
+    {
+      "epoch": 7.0,
+      "eval_ACCOUNTNAME_f1": 1.0,
+      "eval_ACCOUNTNUMBER_f1": 1.0,
+      "eval_AMOUNT_f1": 1.0,
+      "eval_BIC_f1": 1.0,
+      "eval_BITCOINADDRESS_f1": 1.0,
+      "eval_BUILDINGNUMBER_f1": 1.0,
+      "eval_CITY_f1": 0.9997771339425006,
+      "eval_COMPANY_NAME_f1": 1.0,
+      "eval_COUNTY_f1": 1.0,
+      "eval_CREDITCARDCVV_f1": 1.0,
+      "eval_CREDITCARDISSUER_f1": 1.0,
+      "eval_CREDITCARDNUMBER_f1": 0.977198697068404,
+      "eval_CURRENCYCODE_f1": 0.9879518072289156,
+      "eval_CURRENCYNAME_f1": 0.9970674486803519,
+      "eval_CURRENCYSYMBOL_f1": 0.9855072463768116,
+      "eval_CURRENCY_f1": 1.0,
+      "eval_DATE_f1": 1.0,
+      "eval_DISPLAYNAME_f1": 0.9763779527559054,
+      "eval_EMAIL_f1": 1.0,
+      "eval_ETHEREUMADDRESS_f1": 1.0,
+      "eval_FIRSTNAME_f1": 0.9965102753005042,
+      "eval_FULLNAME_f1": 0.9994899260392757,
+      "eval_GENDER_f1": 1.0,
+      "eval_IBAN_f1": 0.9901960784313726,
+      "eval_IPV4_f1": 0.9113924050632911,
+      "eval_IPV6_f1": 0.8241758241758242,
+      "eval_IP_f1": 0.44285714285714284,
+      "eval_JOBAREA_f1": 0.9992992291520673,
+      "eval_JOBDESCRIPTOR_f1": 1.0,
+      "eval_JOBTITLE_f1": 1.0,
+      "eval_JOBTYPE_f1": 1.0,
+      "eval_LASTNAME_f1": 0.9870410367170626,
+      "eval_LITECOINADDRESS_f1": 0.9943502824858756,
+      "eval_MAC_f1": 0.7984189723320159,
+      "eval_MASKEDNUMBER_f1": 0.9629629629629629,
+      "eval_MIDDLENAME_f1": 0.9944954128440368,
+      "eval_NAME_f1": 1.0,
+      "eval_NUMBER_f1": 1.0,
+      "eval_PASSWORD_f1": 1.0,
+      "eval_PHONEIMEI_f1": 0.9875,
+      "eval_PHONE_NUMBER_f1": 1.0,
+      "eval_PIN_f1": 1.0,
+      "eval_PREFIX_f1": 0.9962686567164178,
+      "eval_SECONDARYADDRESS_f1": 1.0,
+      "eval_SEXTYPE_f1": 0.9473684210526316,
+      "eval_SEX_f1": 0.9935483870967742,
+      "eval_SSN_f1": 0.9935483870967742,
+      "eval_STATE_f1": 0.9996004794246904,
+      "eval_STREETADDRESS_f1": 1.0,
+      "eval_STREET_f1": 0.9939393939393938,
+      "eval_SUFFIX_f1": 0.9855072463768114,
+      "eval_TIME_f1": 1.0,
+      "eval_URL_f1": 1.0,
+      "eval_USERAGENT_f1": 1.0,
+      "eval_USERNAME_f1": 1.0,
+      "eval_VEHICLEVIN_f1": 1.0,
+      "eval_VEHICLEVRM_f1": 1.0,
+      "eval_ZIPCODE_f1": 1.0,
+      "eval_loss": 0.008737307973206043,
+      "eval_overall_accuracy": 0.9963114296004669,
+      "eval_overall_f1": 0.9932909729864046,
+      "eval_overall_precision": 0.9905834041348061,
+      "eval_overall_recall": 0.996013383640635,
+      "eval_runtime": 85.9696,
+      "eval_samples_per_second": 124.381,
+      "eval_steps_per_second": 15.552,
+      "step": 9359
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 9359,
+  "num_train_epochs": 7,
+  "save_steps": 500,
+  "total_flos": 3293070338065680.0,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed8e8c274d68a9ab2c8cd8253471b18e6035136643f0483c06a7e2df27f9fd3b
+size 4155