first model version

README.md

@@ -0,0 +1,4 @@
+# BERT-based deidentification model
+
+This repo contains model weights trained on the I2B2 dataset.
+See [OBI EHR deidentification] (https://github.com/obi-ds/ehr_deidentification) for more details and how to get started.

config.json

@@ -0,0 +1,119 @@
+{
+  "_name_or_path": "emilyalsentzer/Bio_ClinicalBERT",
+  "architectures": [
+    "BertForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "finetuning_task": "ner",
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "B-AGE",
+    "1": "B-DATE",
+    "2": "B-EMAIL",
+    "3": "B-HOSP",
+    "4": "B-ID",
+    "5": "B-LOC",
+    "6": "B-OTHERPHI",
+    "7": "B-PATIENT",
+    "8": "B-PATORG",
+    "9": "B-PHONE",
+    "10": "B-STAFF",
+    "11": "I-AGE",
+    "12": "I-DATE",
+    "13": "I-EMAIL",
+    "14": "I-HOSP",
+    "15": "I-ID",
+    "16": "I-LOC",
+    "17": "I-OTHERPHI",
+    "18": "I-PATIENT",
+    "19": "I-PATORG",
+    "20": "I-PHONE",
+    "21": "I-STAFF",
+    "22": "L-AGE",
+    "23": "L-DATE",
+    "24": "L-EMAIL",
+    "25": "L-HOSP",
+    "26": "L-ID",
+    "27": "L-LOC",
+    "28": "L-OTHERPHI",
+    "29": "L-PATIENT",
+    "30": "L-PATORG",
+    "31": "L-PHONE",
+    "32": "L-STAFF",
+    "33": "O",
+    "34": "U-AGE",
+    "35": "U-DATE",
+    "36": "U-EMAIL",
+    "37": "U-HOSP",
+    "38": "U-ID",
+    "39": "U-LOC",
+    "40": "U-OTHERPHI",
+    "41": "U-PATIENT",
+    "42": "U-PATORG",
+    "43": "U-PHONE",
+    "44": "U-STAFF"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "B-AGE": 0,
+    "B-DATE": 1,
+    "B-EMAIL": 2,
+    "B-HOSP": 3,
+    "B-ID": 4,
+    "B-LOC": 5,
+    "B-OTHERPHI": 6,
+    "B-PATIENT": 7,
+    "B-PATORG": 8,
+    "B-PHONE": 9,
+    "B-STAFF": 10,
+    "I-AGE": 11,
+    "I-DATE": 12,
+    "I-EMAIL": 13,
+    "I-HOSP": 14,
+    "I-ID": 15,
+    "I-LOC": 16,
+    "I-OTHERPHI": 17,
+    "I-PATIENT": 18,
+    "I-PATORG": 19,
+    "I-PHONE": 20,
+    "I-STAFF": 21,
+    "L-AGE": 22,
+    "L-DATE": 23,
+    "L-EMAIL": 24,
+    "L-HOSP": 25,
+    "L-ID": 26,
+    "L-LOC": 27,
+    "L-OTHERPHI": 28,
+    "L-PATIENT": 29,
+    "L-PATORG": 30,
+    "L-PHONE": 31,
+    "L-STAFF": 32,
+    "O": 33,
+    "U-AGE": 34,
+    "U-DATE": 35,
+    "U-EMAIL": 36,
+    "U-HOSP": 37,
+    "U-ID": 38,
+    "U-LOC": 39,
+    "U-OTHERPHI": 40,
+    "U-PATIENT": 41,
+    "U-PATORG": 42,
+    "U-PHONE": 43,
+    "U-STAFF": 44
+  },
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "transformers_version": "4.6.1",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 28996
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:842d9f0d394c71528bb73f87748c580a9a3ea82973a18e0871fd579e3eb21c6b
+size 431100529

special_tokens_map.json

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

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": false, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "special_tokens_map_file": null, "name_or_path": "emilyalsentzer/Bio_ClinicalBERT", "do_basic_tokenize": true, "never_split": null}