redact / config.json
finnvoorhees's picture
v0.2.1: retrained model (context-robust; fixes punctuation-dependent city detection under int8/int4). 85-label head (IMEI now model-supported).
18df68d verified
Raw
History Blame Contribute Delete
4.78 kB
{
"add_cross_attention": false,
"architectures": [
"BertForTokenClassification"
],
"attention_probs_dropout_prob": 0.1,
"bos_token_id": 0,
"classifier_dropout": null,
"dtype": "float32",
"eos_token_id": 2,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 384,
"id2label": {
"0": "O",
"1": "B-SSN",
"2": "I-SSN",
"3": "E-SSN",
"4": "S-SSN",
"5": "B-CREDIT_CARD",
"6": "I-CREDIT_CARD",
"7": "E-CREDIT_CARD",
"8": "S-CREDIT_CARD",
"9": "B-EMAIL",
"10": "I-EMAIL",
"11": "E-EMAIL",
"12": "S-EMAIL",
"13": "B-URL",
"14": "I-URL",
"15": "E-URL",
"16": "S-URL",
"17": "B-IP_ADDRESS",
"18": "I-IP_ADDRESS",
"19": "E-IP_ADDRESS",
"20": "S-IP_ADDRESS",
"21": "B-GIVEN_NAME",
"22": "I-GIVEN_NAME",
"23": "E-GIVEN_NAME",
"24": "S-GIVEN_NAME",
"25": "B-SURNAME",
"26": "I-SURNAME",
"27": "E-SURNAME",
"28": "S-SURNAME",
"29": "B-PHONE",
"30": "I-PHONE",
"31": "E-PHONE",
"32": "S-PHONE",
"33": "B-TAX_ID",
"34": "I-TAX_ID",
"35": "E-TAX_ID",
"36": "S-TAX_ID",
"37": "B-BANK_ACCOUNT",
"38": "I-BANK_ACCOUNT",
"39": "E-BANK_ACCOUNT",
"40": "S-BANK_ACCOUNT",
"41": "B-ROUTING_NUMBER",
"42": "I-ROUTING_NUMBER",
"43": "E-ROUTING_NUMBER",
"44": "S-ROUTING_NUMBER",
"45": "B-GOVERNMENT_ID",
"46": "I-GOVERNMENT_ID",
"47": "E-GOVERNMENT_ID",
"48": "S-GOVERNMENT_ID",
"49": "B-PASSPORT",
"50": "I-PASSPORT",
"51": "E-PASSPORT",
"52": "S-PASSPORT",
"53": "B-DRIVERS_LICENSE",
"54": "I-DRIVERS_LICENSE",
"55": "E-DRIVERS_LICENSE",
"56": "S-DRIVERS_LICENSE",
"57": "B-BUILDING_NUMBER",
"58": "I-BUILDING_NUMBER",
"59": "E-BUILDING_NUMBER",
"60": "S-BUILDING_NUMBER",
"61": "B-STREET_NAME",
"62": "I-STREET_NAME",
"63": "E-STREET_NAME",
"64": "S-STREET_NAME",
"65": "B-SECONDARY_ADDRESS",
"66": "I-SECONDARY_ADDRESS",
"67": "E-SECONDARY_ADDRESS",
"68": "S-SECONDARY_ADDRESS",
"69": "B-CITY",
"70": "I-CITY",
"71": "E-CITY",
"72": "S-CITY",
"73": "B-STATE",
"74": "I-STATE",
"75": "E-STATE",
"76": "S-STATE",
"77": "B-ZIP_CODE",
"78": "I-ZIP_CODE",
"79": "E-ZIP_CODE",
"80": "S-ZIP_CODE",
"81": "B-IMEI",
"82": "I-IMEI",
"83": "E-IMEI",
"84": "S-IMEI"
},
"initializer_range": 0.02,
"intermediate_size": 1536,
"is_decoder": false,
"label2id": {
"B-BANK_ACCOUNT": 37,
"B-BUILDING_NUMBER": 57,
"B-CITY": 69,
"B-CREDIT_CARD": 5,
"B-DRIVERS_LICENSE": 53,
"B-EMAIL": 9,
"B-GIVEN_NAME": 21,
"B-GOVERNMENT_ID": 45,
"B-IMEI": 81,
"B-IP_ADDRESS": 17,
"B-PASSPORT": 49,
"B-PHONE": 29,
"B-ROUTING_NUMBER": 41,
"B-SECONDARY_ADDRESS": 65,
"B-SSN": 1,
"B-STATE": 73,
"B-STREET_NAME": 61,
"B-SURNAME": 25,
"B-TAX_ID": 33,
"B-URL": 13,
"B-ZIP_CODE": 77,
"E-BANK_ACCOUNT": 39,
"E-BUILDING_NUMBER": 59,
"E-CITY": 71,
"E-CREDIT_CARD": 7,
"E-DRIVERS_LICENSE": 55,
"E-EMAIL": 11,
"E-GIVEN_NAME": 23,
"E-GOVERNMENT_ID": 47,
"E-IMEI": 83,
"E-IP_ADDRESS": 19,
"E-PASSPORT": 51,
"E-PHONE": 31,
"E-ROUTING_NUMBER": 43,
"E-SECONDARY_ADDRESS": 67,
"E-SSN": 3,
"E-STATE": 75,
"E-STREET_NAME": 63,
"E-SURNAME": 27,
"E-TAX_ID": 35,
"E-URL": 15,
"E-ZIP_CODE": 79,
"I-BANK_ACCOUNT": 38,
"I-BUILDING_NUMBER": 58,
"I-CITY": 70,
"I-CREDIT_CARD": 6,
"I-DRIVERS_LICENSE": 54,
"I-EMAIL": 10,
"I-GIVEN_NAME": 22,
"I-GOVERNMENT_ID": 46,
"I-IMEI": 82,
"I-IP_ADDRESS": 18,
"I-PASSPORT": 50,
"I-PHONE": 30,
"I-ROUTING_NUMBER": 42,
"I-SECONDARY_ADDRESS": 66,
"I-SSN": 2,
"I-STATE": 74,
"I-STREET_NAME": 62,
"I-SURNAME": 26,
"I-TAX_ID": 34,
"I-URL": 14,
"I-ZIP_CODE": 78,
"O": 0,
"S-BANK_ACCOUNT": 40,
"S-BUILDING_NUMBER": 60,
"S-CITY": 72,
"S-CREDIT_CARD": 8,
"S-DRIVERS_LICENSE": 56,
"S-EMAIL": 12,
"S-GIVEN_NAME": 24,
"S-GOVERNMENT_ID": 48,
"S-IMEI": 84,
"S-IP_ADDRESS": 20,
"S-PASSPORT": 52,
"S-PHONE": 32,
"S-ROUTING_NUMBER": 44,
"S-SECONDARY_ADDRESS": 68,
"S-SSN": 4,
"S-STATE": 76,
"S-STREET_NAME": 64,
"S-SURNAME": 28,
"S-TAX_ID": 36,
"S-URL": 16,
"S-ZIP_CODE": 80
},
"layer_norm_eps": 1e-12,
"max_position_embeddings": 512,
"model_type": "bert",
"num_attention_heads": 12,
"num_hidden_layers": 6,
"pad_token_id": 1,
"tie_word_embeddings": true,
"tokenizer_class": "XLMRobertaTokenizer",
"transformers_version": "5.12.1",
"type_vocab_size": 2,
"use_cache": false,
"vocab_size": 31475
}