Training in progress epoch 0

all_results.json

@@ -0,0 +1 @@
+{"eval_account_number_precision": 0.7672327672327672, "eval_account_number_recall": 0.7045871559633028, "eval_account_number_f1": 0.7345767575322812, "eval_account_number_number": 1090, "eval_address_precision": 0.8397394136807818, "eval_address_recall": 0.8469119579500657, "eval_address_f1": 0.8433104350670593, "eval_address_number": 1522, "eval_api_key_precision": 0.9230769230769231, "eval_api_key_recall": 0.935064935064935, "eval_api_key_f1": 0.9290322580645162, "eval_api_key_number": 77, "eval_bank_routing_number_precision": 0.915929203539823, "eval_bank_routing_number_recall": 0.9951923076923077, "eval_bank_routing_number_f1": 0.9539170506912442, "eval_bank_routing_number_number": 208, "eval_biometric_identifier_precision": 0.9148936170212766, "eval_biometric_identifier_recall": 1.0, "eval_biometric_identifier_f1": 0.9555555555555556, "eval_biometric_identifier_number": 129, "eval_certificate_license_number_precision": 0.9851851851851852, "eval_certificate_license_number_recall": 1.0, "eval_certificate_license_number_f1": 0.9925373134328358, "eval_certificate_license_number_number": 133, "eval_city_precision": 0.9478219243657252, "eval_city_recall": 0.953757225433526, "eval_city_f1": 0.95078031212485, "eval_city_number": 2076, "eval_company_name_precision": 0.8638743455497382, "eval_company_name_recall": 0.9482758620689655, "eval_company_name_f1": 0.9041095890410958, "eval_company_name_number": 174, "eval_coordinate_precision": 0.9365079365079365, "eval_coordinate_recall": 0.9516129032258065, "eval_coordinate_f1": 0.944, "eval_coordinate_number": 62, "eval_country_precision": 0.8148148148148148, "eval_country_recall": 0.9166666666666666, "eval_country_f1": 0.8627450980392156, "eval_country_number": 72, "eval_credit_card_number_precision": 0.8346525945470537, "eval_credit_card_number_recall": 0.9452191235059761, "eval_credit_card_number_f1": 0.8865016347501167, "eval_credit_card_number_number": 1004, "eval_customer_id_precision": 0.9577889447236181, "eval_customer_id_recall": 0.9473161033797217, "eval_customer_id_f1": 0.9525237381309345, "eval_customer_id_number": 1006, "eval_cvv_precision": 0.0, "eval_cvv_recall": 0.0, "eval_cvv_f1": 0.0, "eval_cvv_number": 11, "eval_date_precision": 0.94, "eval_date_recall": 0.935930735930736, "eval_date_f1": 0.9379609544468547, "eval_date_number": 1155, "eval_date_of_birth_precision": 0.9044410413476264, "eval_date_of_birth_recall": 0.9392493638676844, "eval_date_of_birth_f1": 0.9215166172569824, "eval_date_of_birth_number": 3144, "eval_date_time_precision": 0.9666666666666667, "eval_date_time_recall": 0.9854368932038835, "eval_date_time_f1": 0.9759615384615385, "eval_date_time_number": 206, "eval_device_identifier_precision": 0.9545454545454546, "eval_device_identifier_recall": 0.9130434782608695, "eval_device_identifier_f1": 0.9333333333333332, "eval_device_identifier_number": 207, "eval_email_precision": 0.994281357224552, "eval_email_recall": 0.994281357224552, "eval_email_f1": 0.994281357224552, "eval_email_number": 2623, "eval_employee_id_precision": 0.9439347604485219, "eval_employee_id_recall": 0.9497435897435897, "eval_employee_id_f1": 0.9468302658486707, "eval_employee_id_number": 975, "eval_first_name_precision": 0.8275536683179995, "eval_first_name_recall": 0.8910337820675641, "eval_first_name_f1": 0.8581213307240705, "eval_first_name_number": 3937, "eval_health_plan_beneficiary_number_precision": 1.0, "eval_health_plan_beneficiary_number_recall": 0.7708333333333334, "eval_health_plan_beneficiary_number_f1": 0.8705882352941177, "eval_health_plan_beneficiary_number_number": 48, "eval_ipv4_precision": 0.9950433705080545, "eval_ipv4_recall": 0.9975155279503105, "eval_ipv4_f1": 0.9962779156327544, "eval_ipv4_number": 805, "eval_ipv6_precision": 0.9491525423728814, "eval_ipv6_recall": 0.9491525423728814, "eval_ipv6_f1": 0.9491525423728814, "eval_ipv6_number": 59, "eval_last_name_precision": 0.8394557823129252, "eval_last_name_recall": 0.7892548768787976, "eval_last_name_f1": 0.8135816713367396, "eval_last_name_number": 3127, "eval_license_plate_precision": 0.9623706491063029, "eval_license_plate_recall": 0.899736147757256, "eval_license_plate_f1": 0.9299999999999999, "eval_license_plate_number": 1137, "eval_medical_record_number_precision": 0.9713289422704379, "eval_medical_record_number_recall": 0.9881750098541584, "eval_medical_record_number_f1": 0.9796795623290347, "eval_medical_record_number_number": 2537, "eval_name_precision": 0.8540145985401459, "eval_name_recall": 0.9276511397423192, "eval_name_f1": 0.8893111638954869, "eval_name_number": 1009, "eval_national_id_precision": 0.925, "eval_national_id_recall": 0.8409090909090909, "eval_national_id_f1": 0.8809523809523809, "eval_national_id_number": 44, "eval_password_precision": 0.8937329700272479, "eval_password_recall": 0.9123783031988874, "eval_password_f1": 0.9029593943565037, "eval_password_number": 719, "eval_phone_number_precision": 0.9796208530805687, "eval_phone_number_recall": 0.9828815977175464, "eval_phone_number_f1": 0.9812485164965583, "eval_phone_number_number": 2103, "eval_pin_precision": 0.0, "eval_pin_recall": 0.0, "eval_pin_f1": 0.0, "eval_pin_number": 4, "eval_postcode_precision": 0.8739495798319328, "eval_postcode_recall": 0.9425981873111783, "eval_postcode_f1": 0.9069767441860465, "eval_postcode_number": 993, "eval_ssn_precision": 0.9356557377049181, "eval_ssn_recall": 0.9457332228666114, "eval_ssn_f1": 0.9406674907292955, "eval_ssn_number": 2414, "eval_state_precision": 0.5555555555555556, "eval_state_recall": 0.22727272727272727, "eval_state_f1": 0.3225806451612903, "eval_state_number": 22, "eval_street_address_precision": 0.888030888030888, "eval_street_address_recall": 0.9317129629629629, "eval_street_address_f1": 0.9093476419090653, "eval_street_address_number": 1728, "eval_swift_bic_precision": 1.0, "eval_swift_bic_recall": 0.6842105263157895, "eval_swift_bic_f1": 0.8125000000000001, "eval_swift_bic_number": 19, "eval_tax_id_precision": 0.7963963963963964, "eval_tax_id_recall": 0.7453625632377741, "eval_tax_id_f1": 0.7700348432055748, "eval_tax_id_number": 593, "eval_time_precision": 0.9327731092436975, "eval_time_recall": 0.9910714285714286, "eval_time_f1": 0.961038961038961, "eval_time_number": 112, "eval_unique_identifier_precision": 0.8455349959774738, "eval_unique_identifier_recall": 0.8952299829642248, "eval_unique_identifier_f1": 0.869673148531237, "eval_unique_identifier_number": 1174, "eval_url_precision": 0.8837209302325582, "eval_url_recall": 0.9743589743589743, "eval_url_f1": 0.9268292682926831, "eval_url_number": 39, "eval_user_name_precision": 0.9459459459459459, "eval_user_name_recall": 0.8631528046421664, "eval_user_name_f1": 0.9026548672566372, "eval_user_name_number": 2068, "eval_vehicle_identifier_precision": 0.9523809523809523, "eval_vehicle_identifier_recall": 1.0, "eval_vehicle_identifier_f1": 0.975609756097561, "eval_vehicle_identifier_number": 100, "eval_overall_precision": 0.9050920857358343, "eval_overall_recall": 0.9148407721627936, "eval_overall_f1": 0.909940318951179, "eval_overall_accuracy": 0.9908545332539793}

config.json

@@ -10,178 +10,90 @@
   "hidden_size": 768,
   "id2label": {
     "0": "O",
-    "1": "B-medical_record_number",
-    "2": "I-medical_record_number",
+    "1": "B-unique_identifier",
+    "2": "I-unique_identifier",
     "3": "B-date_of_birth",
     "4": "I-date_of_birth",
     "5": "B-ssn",
     "6": "I-ssn",
-    "7": "B-date",
-    "8": "I-date",
-    "9": "B-first_name",
-    "10": "I-first_name",
+    "7": "B-date_time",
+    "8": "I-date_time",
+    "9": "B-person_name",
+    "10": "I-person_name",
     "11": "B-email",
     "12": "I-email",
-    "13": "B-last_name",
-    "14": "I-last_name",
-    "15": "B-customer_id",
-    "16": "I-customer_id",
-    "17": "B-employee_id",
-    "18": "I-employee_id",
-    "19": "B-name",
-    "20": "I-name",
-    "21": "B-street_address",
-    "22": "I-street_address",
-    "23": "B-phone_number",
-    "24": "I-phone_number",
-    "25": "B-ipv4",
-    "26": "I-ipv4",
-    "27": "B-credit_card_number",
-    "28": "I-credit_card_number",
-    "29": "B-license_plate",
-    "30": "I-license_plate",
-    "31": "B-address",
-    "32": "I-address",
-    "33": "B-user_name",
-    "34": "I-user_name",
-    "35": "B-device_identifier",
-    "36": "I-device_identifier",
-    "37": "B-bank_routing_number",
-    "38": "I-bank_routing_number",
-    "39": "B-date_time",
-    "40": "I-date_time",
-    "41": "B-company_name",
-    "42": "I-company_name",
-    "43": "B-unique_identifier",
-    "44": "I-unique_identifier",
-    "45": "B-biometric_identifier",
-    "46": "I-biometric_identifier",
-    "47": "B-account_number",
-    "48": "I-account_number",
-    "49": "B-city",
-    "50": "I-city",
-    "51": "B-certificate_license_number",
-    "52": "I-certificate_license_number",
-    "53": "B-time",
-    "54": "I-time",
-    "55": "B-postcode",
-    "56": "I-postcode",
-    "57": "B-vehicle_identifier",
-    "58": "I-vehicle_identifier",
-    "59": "B-coordinate",
-    "60": "I-coordinate",
-    "61": "B-country",
-    "62": "I-country",
-    "63": "B-api_key",
-    "64": "I-api_key",
-    "65": "B-ipv6",
-    "66": "I-ipv6",
-    "67": "B-password",
-    "68": "I-password",
-    "69": "B-health_plan_beneficiary_number",
-    "70": "I-health_plan_beneficiary_number",
-    "71": "B-national_id",
-    "72": "I-national_id",
-    "73": "B-tax_id",
-    "74": "I-tax_id",
-    "75": "B-url",
-    "76": "I-url",
-    "77": "B-state",
-    "78": "I-state",
-    "79": "B-swift_bic",
-    "80": "I-swift_bic",
-    "81": "B-cvv",
-    "82": "I-cvv",
-    "83": "B-pin",
-    "84": "I-pin"
+    "13": "B-location",
+    "14": "I-location",
+    "15": "B-phone_number",
+    "16": "I-phone_number",
+    "17": "B-ipv4",
+    "18": "I-ipv4",
+    "19": "B-credit_card_number",
+    "20": "I-credit_card_number",
+    "21": "B-user_name",
+    "22": "I-user_name",
+    "23": "B-company_name",
+    "24": "I-company_name",
+    "25": "B-account_number",
+    "26": "I-account_number",
+    "27": "B-api_key",
+    "28": "I-api_key",
+    "29": "B-ipv6",
+    "30": "I-ipv6",
+    "31": "B-password",
+    "32": "I-password",
+    "33": "B-url",
+    "34": "I-url",
+    "35": "B-swift_bic",
+    "36": "I-swift_bic",
+    "37": "B-cvv",
+    "38": "I-cvv",
+    "39": "B-pin",
+    "40": "I-pin"
   },
   "initializer_range": 0.02,
   "intermediate_size": 3072,
   "label2id": {
-    "B-account_number": 47,
-    "B-address": 31,
-    "B-api_key": 63,
-    "B-bank_routing_number": 37,
-    "B-biometric_identifier": 45,
-    "B-certificate_license_number": 51,
-    "B-city": 49,
-    "B-company_name": 41,
-    "B-coordinate": 59,
-    "B-country": 61,
-    "B-credit_card_number": 27,
-    "B-customer_id": 15,
-    "B-cvv": 81,
-    "B-date": 7,
+    "B-account_number": 25,
+    "B-api_key": 27,
+    "B-company_name": 23,
+    "B-credit_card_number": 19,
+    "B-cvv": 37,
     "B-date_of_birth": 3,
-    "B-date_time": 39,
-    "B-device_identifier": 35,
+    "B-date_time": 7,
     "B-email": 11,
-    "B-employee_id": 17,
-    "B-first_name": 9,
-    "B-health_plan_beneficiary_number": 69,
-    "B-ipv4": 25,
-    "B-ipv6": 65,
-    "B-last_name": 13,
-    "B-license_plate": 29,
-    "B-medical_record_number": 1,
-    "B-name": 19,
-    "B-national_id": 71,
-    "B-password": 67,
-    "B-phone_number": 23,
-    "B-pin": 83,
-    "B-postcode": 55,
+    "B-ipv4": 17,
+    "B-ipv6": 29,
+    "B-location": 13,
+    "B-password": 31,
+    "B-person_name": 9,
+    "B-phone_number": 15,
+    "B-pin": 39,
     "B-ssn": 5,
-    "B-state": 77,
-    "B-street_address": 21,
-    "B-swift_bic": 79,
-    "B-tax_id": 73,
-    "B-time": 53,
-    "B-unique_identifier": 43,
-    "B-url": 75,
-    "B-user_name": 33,
-    "B-vehicle_identifier": 57,
-    "I-account_number": 48,
-    "I-address": 32,
-    "I-api_key": 64,
-    "I-bank_routing_number": 38,
-    "I-biometric_identifier": 46,
-    "I-certificate_license_number": 52,
-    "I-city": 50,
-    "I-company_name": 42,
-    "I-coordinate": 60,
-    "I-country": 62,
-    "I-credit_card_number": 28,
-    "I-customer_id": 16,
-    "I-cvv": 82,
-    "I-date": 8,
+    "B-swift_bic": 35,
+    "B-unique_identifier": 1,
+    "B-url": 33,
+    "B-user_name": 21,
+    "I-account_number": 26,
+    "I-api_key": 28,
+    "I-company_name": 24,
+    "I-credit_card_number": 20,
+    "I-cvv": 38,
     "I-date_of_birth": 4,
-    "I-date_time": 40,
-    "I-device_identifier": 36,
+    "I-date_time": 8,
     "I-email": 12,
-    "I-employee_id": 18,
-    "I-first_name": 10,
-    "I-health_plan_beneficiary_number": 70,
-    "I-ipv4": 26,
-    "I-ipv6": 66,
-    "I-last_name": 14,
-    "I-license_plate": 30,
-    "I-medical_record_number": 2,
-    "I-name": 20,
-    "I-national_id": 72,
-    "I-password": 68,
-    "I-phone_number": 24,
-    "I-pin": 84,
-    "I-postcode": 56,
+    "I-ipv4": 18,
+    "I-ipv6": 30,
+    "I-location": 14,
+    "I-password": 32,
+    "I-person_name": 10,
+    "I-phone_number": 16,
+    "I-pin": 40,
     "I-ssn": 6,
-    "I-state": 78,
-    "I-street_address": 22,
-    "I-swift_bic": 80,
-    "I-tax_id": 74,
-    "I-time": 54,
-    "I-unique_identifier": 44,
-    "I-url": 76,
-    "I-user_name": 34,
-    "I-vehicle_identifier": 58,
+    "I-swift_bic": 36,
+    "I-unique_identifier": 2,
+    "I-url": 34,
+    "I-user_name": 22,
     "O": 0
   },
   "layer_norm_eps": 1e-12,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:23735799059e4eb802ba7e66530c94a74fdf50de40e4424b812a2643eb46fb26
-size 435851396
+oid sha256:bd0bb23d851b85c91efbe79377f333a7f1cc0d517a91c74983b1062199e300fd
+size 435716052