thesis release

config.json

@@ -4,7 +4,7 @@
     "BertForTokenClassification"
   ],
   "attention_probs_dropout_prob": 0.1,
-  "gradient_checkpointing": false,
+  "classifier_dropout": null,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
   "hidden_size": 768,
@@ -35,7 +35,8 @@
   "num_hidden_layers": 12,
   "pad_token_id": 0,
   "position_embedding_type": "absolute",
-  "transformers_version": "4.8.2",
+  "torch_dtype": "float32",
+  "transformers_version": "4.17.0",
   "type_vocab_size": 2,
   "use_cache": true,
   "vocab_size": 28996

eval_results.txt

@@ -1,20 +1,20 @@
-eval_loss = 0.1652524173259735
-eval_precision = 0.8504558443325623
-eval_recall = 0.8611187655001378
-eval_f1 = 0.8557540905045526
-eval_accuracy = 0.975697565751413
-eval_matrix = [[3901, 4, 60, 102, 1, 10, 538], [0, 4949, 15, 0, 42, 1, 371], [56, 18, 3770, 1, 0, 207, 372], [101, 0, 2, 2779, 0, 26, 335], [0, 32, 0, 0, 1383, 0, 213], [6, 4, 113, 28, 4, 2247, 335], [354, 467, 574, 247, 369, 453, 200220]]
+eval_loss = 0.17402765154838562
+eval_precision = 0.8519104303671023
+eval_recall = 0.8661870503597122
+eval_f1 = 0.8589894242068155
+eval_accuracy = 0.9750561746748678
+eval_matrix = [[1841, 0, 31, 44, 0, 4, 219], [0, 4941, 10, 0, 36, 0, 360], [71, 17, 3565, 1, 0, 172, 420], [64, 0, 1, 1221, 0, 28, 102], [0, 25, 0, 0, 1476, 2, 245], [2, 3, 119, 35, 6, 2095, 463], [360, 282, 417, 227, 226, 404, 156701]]
 eval_classification_report =               precision    recall  f1-score   support
 
-     Anatomy       0.85      0.83      0.84      4649
-    Chemical       0.89      0.91      0.90      5402
-     Disease       0.80      0.84      0.82      4465
+     Anatomy       0.76      0.84      0.80      2162
+    Chemical       0.92      0.92      0.92      5371
+     Disease       0.82      0.82      0.82      4282
 
-   micro avg       0.85      0.86      0.86     14516
-   macro avg       0.85      0.86      0.85     14516
-weighted avg       0.85      0.86      0.86     14516
+   micro avg       0.85      0.87      0.86     11815
+   macro avg       0.83      0.86      0.84     11815
+weighted avg       0.85      0.87      0.86     11815
 
-eval_runtime = 16.1004
-eval_samples_per_second = 152.977
-eval_steps_per_second = 0.621
-epoch = 35.0
+eval_runtime = 54.806
+eval_samples_per_second = 36.273
+eval_steps_per_second = 9.068
+epoch = 15.0

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:89ff1d03f6a557db79ca98a49d86d41def96fa8d586d91c968f19e6e8c8e0ecc
-size 430983665
+oid sha256:8c0de00884e3be0723c9ae9e5d6880db22290f31d8eaddd7c82fedd8d93daa90
+size 430984561

test_results.txt

@@ -0,0 +1,20 @@
+eval_loss = 0.17808516323566437
+eval_precision = 0.8430268069222938
+eval_recall = 0.855745384403417
+eval_f1 = 0.8493384841543878
+eval_accuracy = 0.975430554937475
+eval_matrix = [[4007, 4, 58, 78, 0, 9, 460], [0, 4872, 9, 0, 41, 0, 456], [62, 14, 3706, 3, 0, 187, 452], [107, 0, 3, 2829, 0, 29, 275], [0, 23, 0, 0, 1398, 0, 207], [9, 3, 144, 28, 3, 2151, 399], [397, 414, 581, 253, 356, 457, 200226]]
+eval_classification_report =               precision    recall  f1-score   support
+
+     Anatomy       0.84      0.85      0.84      4649
+    Chemical       0.90      0.89      0.89      5402
+     Disease       0.79      0.82      0.80      4465
+
+   micro avg       0.84      0.86      0.85     14516
+   macro avg       0.84      0.85      0.85     14516
+weighted avg       0.84      0.86      0.85     14516
+
+eval_runtime = 69.3988
+eval_samples_per_second = 35.491
+eval_steps_per_second = 8.876
+epoch = 15.0

tokenizer_config.json

@@ -1 +1 @@
-{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}
+{"do_lower_case": false, "do_basic_tokenize": true, "never_split": null, "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": "dmis-lab/biobert-base-cased-v1.1", "tokenizer_class": "BertTokenizer"}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:80b5808a16224f2716b8ece4c7a86a0ef8a4b06c4cf6511871afb0dc3fb549ac
-size 2607
+oid sha256:289790c3ee17f28436f1993d68436e419d5800ae5288688b3ce9ca02d7b8f557
+size 3055