Train 10k + CRF for open-pii-masking-500k-ai4privacy. F1: 0.9434
Browse files
open-pii-masking-500k-ai4privacy/TokenBased-BERT/config.json
ADDED
|
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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-AGE",
|
| 14 |
+
"1": "B-BUILDINGNUM",
|
| 15 |
+
"2": "B-CITY",
|
| 16 |
+
"3": "B-CREDITCARDNUMBER",
|
| 17 |
+
"4": "B-DATE",
|
| 18 |
+
"5": "B-DRIVERLICENSENUM",
|
| 19 |
+
"6": "B-EMAIL",
|
| 20 |
+
"7": "B-GENDER",
|
| 21 |
+
"8": "B-GIVENNAME",
|
| 22 |
+
"9": "B-IDCARDNUM",
|
| 23 |
+
"10": "B-PASSPORTNUM",
|
| 24 |
+
"11": "B-SOCIALNUM",
|
| 25 |
+
"12": "B-STREET",
|
| 26 |
+
"13": "B-SURNAME",
|
| 27 |
+
"14": "B-TAXNUM",
|
| 28 |
+
"15": "B-TELEPHONENUM",
|
| 29 |
+
"16": "B-TIME",
|
| 30 |
+
"17": "B-TITLE",
|
| 31 |
+
"18": "B-ZIPCODE",
|
| 32 |
+
"19": "I-AGE",
|
| 33 |
+
"20": "I-BUILDINGNUM",
|
| 34 |
+
"21": "I-CITY",
|
| 35 |
+
"22": "I-CREDITCARDNUMBER",
|
| 36 |
+
"23": "I-DATE",
|
| 37 |
+
"24": "I-DRIVERLICENSENUM",
|
| 38 |
+
"25": "I-EMAIL",
|
| 39 |
+
"26": "I-GENDER",
|
| 40 |
+
"27": "I-GIVENNAME",
|
| 41 |
+
"28": "I-IDCARDNUM",
|
| 42 |
+
"29": "I-PASSPORTNUM",
|
| 43 |
+
"30": "I-SOCIALNUM",
|
| 44 |
+
"31": "I-STREET",
|
| 45 |
+
"32": "I-SURNAME",
|
| 46 |
+
"33": "I-TAXNUM",
|
| 47 |
+
"34": "I-TELEPHONENUM",
|
| 48 |
+
"35": "I-TIME",
|
| 49 |
+
"36": "I-TITLE",
|
| 50 |
+
"37": "I-ZIPCODE",
|
| 51 |
+
"38": "O"
|
| 52 |
+
},
|
| 53 |
+
"initializer_range": 0.02,
|
| 54 |
+
"intermediate_size": 3072,
|
| 55 |
+
"label2id": {
|
| 56 |
+
"B-AGE": 0,
|
| 57 |
+
"B-BUILDINGNUM": 1,
|
| 58 |
+
"B-CITY": 2,
|
| 59 |
+
"B-CREDITCARDNUMBER": 3,
|
| 60 |
+
"B-DATE": 4,
|
| 61 |
+
"B-DRIVERLICENSENUM": 5,
|
| 62 |
+
"B-EMAIL": 6,
|
| 63 |
+
"B-GENDER": 7,
|
| 64 |
+
"B-GIVENNAME": 8,
|
| 65 |
+
"B-IDCARDNUM": 9,
|
| 66 |
+
"B-PASSPORTNUM": 10,
|
| 67 |
+
"B-SOCIALNUM": 11,
|
| 68 |
+
"B-STREET": 12,
|
| 69 |
+
"B-SURNAME": 13,
|
| 70 |
+
"B-TAXNUM": 14,
|
| 71 |
+
"B-TELEPHONENUM": 15,
|
| 72 |
+
"B-TIME": 16,
|
| 73 |
+
"B-TITLE": 17,
|
| 74 |
+
"B-ZIPCODE": 18,
|
| 75 |
+
"I-AGE": 19,
|
| 76 |
+
"I-BUILDINGNUM": 20,
|
| 77 |
+
"I-CITY": 21,
|
| 78 |
+
"I-CREDITCARDNUMBER": 22,
|
| 79 |
+
"I-DATE": 23,
|
| 80 |
+
"I-DRIVERLICENSENUM": 24,
|
| 81 |
+
"I-EMAIL": 25,
|
| 82 |
+
"I-GENDER": 26,
|
| 83 |
+
"I-GIVENNAME": 27,
|
| 84 |
+
"I-IDCARDNUM": 28,
|
| 85 |
+
"I-PASSPORTNUM": 29,
|
| 86 |
+
"I-SOCIALNUM": 30,
|
| 87 |
+
"I-STREET": 31,
|
| 88 |
+
"I-SURNAME": 32,
|
| 89 |
+
"I-TAXNUM": 33,
|
| 90 |
+
"I-TELEPHONENUM": 34,
|
| 91 |
+
"I-TIME": 35,
|
| 92 |
+
"I-TITLE": 36,
|
| 93 |
+
"I-ZIPCODE": 37,
|
| 94 |
+
"O": 38
|
| 95 |
+
},
|
| 96 |
+
"layer_norm_eps": 1e-12,
|
| 97 |
+
"max_position_embeddings": 512,
|
| 98 |
+
"model_type": "bert",
|
| 99 |
+
"num_attention_heads": 12,
|
| 100 |
+
"num_hidden_layers": 12,
|
| 101 |
+
"pad_token_id": 0,
|
| 102 |
+
"pooler_fc_size": 768,
|
| 103 |
+
"pooler_num_attention_heads": 12,
|
| 104 |
+
"pooler_num_fc_layers": 3,
|
| 105 |
+
"pooler_size_per_head": 128,
|
| 106 |
+
"pooler_type": "first_token_transform",
|
| 107 |
+
"position_embedding_type": "absolute",
|
| 108 |
+
"transformers_version": "4.56.0",
|
| 109 |
+
"type_vocab_size": 2,
|
| 110 |
+
"use_cache": false,
|
| 111 |
+
"vocab_size": 119547
|
| 112 |
+
}
|
open-pii-masking-500k-ai4privacy/TokenBased-BERT/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 711563936
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0ff79d6ebd3b9bc6ea03de767ad718be091c4b463a279f6369a617c0666bc7d5
|
| 3 |
size 711563936
|
open-pii-masking-500k-ai4privacy/TokenBased-BERT/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:def6267931d73d1f804e0c0cc1e93d5e4d94b7c47f7308907df437ea0178a4f2
|
| 3 |
+
size 5841
|