alenatz/relation-extraction-biobert-biocause

README.md

@@ -1,5 +1,5 @@
 ---
-base_model: dmis-lab/biobert-v1.1
+base_model: dmis-lab/biobert-base-cased-v1.2
 tags:
 - generated_from_trainer
 metrics:
@@ -17,15 +17,16 @@ should probably proofread and complete it, then remove this comment. -->
 
 # relation-biobert-biocause
 
-This model is a fine-tuned version of [dmis-lab/biobert-v1.1](https://huggingface.co/dmis-lab/biobert-v1.1) on the None dataset.
+This model is a fine-tuned version of [dmis-lab/biobert-base-cased-v1.2](https://huggingface.co/dmis-lab/biobert-base-cased-v1.2) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0398
-- Precision: 0.3308
-- Recall: 0.3333
-- F1: 0.3321
-- Accuracy: 0.9925
-- Balanced Accuracy: 0.3333
-- Kappa: 0.0
+- Loss: 0.2103
+- Precision: 0.1164
+- Recall: 0.625
+- F1: 0.1963
+- Accuracy: 0.9448
+- Relation P: 0.1164
+- Relation R: 0.625
+- Relation F1: 0.1963
 
 ## Model description
 
@@ -44,26 +45,30 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 2e-05
-- train_batch_size: 8
-- eval_batch_size: 8
+- learning_rate: 4e-05
+- train_batch_size: 16
+- eval_batch_size: 16
 - seed: 42
-- gradient_accumulation_steps: 2
-- total_train_batch_size: 16
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - num_epochs: 1
 
 ### Training results
 
-| Training Loss | Epoch  | Step | Validation Loss | Precision | Recall | F1     | Accuracy | Balanced Accuracy | Kappa |
-|:-------------:|:------:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|:-----------------:|:-----:|
-| 0.0983        | 0.5764 | 100  | 0.0462          | 0.3308    | 0.3333 | 0.3321 | 0.9925   | 0.3333            | 0.0   |
+| Training Loss | Epoch  | Step | Validation Loss | Precision | Recall | F1     | Accuracy | Relation P | Relation R | Relation F1 |
+|:-------------:|:------:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|:----------:|:----------:|:-----------:|
+| 0.6563        | 0.1282 | 20   | 0.2984          | 0.0211    | 0.2105 | 0.0384 | 0.8265   | 0.0211     | 0.2105     | 0.0384      |
+| 0.6563        | 0.2564 | 40   | 0.2302          | 0.0763    | 0.4605 | 0.1308 | 0.9266   | 0.0763     | 0.4605     | 0.1308      |
+| 0.6563        | 0.3846 | 60   | 0.4003          | 0.1406    | 0.5921 | 0.2273 | 0.9617   | 0.1406     | 0.5921     | 0.2273      |
+| 0.6563        | 0.5128 | 80   | 0.2185          | 0.0554    | 0.4671 | 0.0990 | 0.8933   | 0.0554     | 0.4671     | 0.0990      |
+| 0.6563        | 0.6410 | 100  | 0.2261          | 0.1345    | 0.7105 | 0.2262 | 0.9510   | 0.1345     | 0.7105     | 0.2262      |
+| 0.6563        | 0.7692 | 120  | 0.2315          | 0.1259    | 0.6579 | 0.2114 | 0.9502   | 0.1259     | 0.6579     | 0.2114      |
+| 0.6563        | 0.8974 | 140  | 0.2324          | 0.1417    | 0.6711 | 0.2339 | 0.9547   | 0.1417     | 0.6711     | 0.2339      |
 
 
 ### Framework versions
 
-- Transformers 4.41.2
+- Transformers 4.42.4
 - Pytorch 2.3.0+cu121
 - Datasets 2.20.0
 - Tokenizers 0.19.1

config.json

@@ -1,24 +1,21 @@
 {
-  "_name_or_path": "dmis-lab/biobert-v1.1",
+  "_name_or_path": "dmis-lab/biobert-base-cased-v1.2",
   "architectures": [
-    "BertForTokenClassification"
+    "BertForUnbalancedTokenClassification"
   ],
   "attention_probs_dropout_prob": 0.1,
   "classifier_dropout": null,
-  "gradient_checkpointing": false,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
   "hidden_size": 768,
   "id2label": {
     "0": "O",
-    "1": "B-RELATION",
-    "2": "I-RELATION"
+    "1": "I-REL"
   },
   "initializer_range": 0.02,
   "intermediate_size": 3072,
   "label2id": {
-    "B-RELATION": 1,
-    "I-RELATION": 2,
+    "I-REL": 1,
     "O": 0
   },
   "layer_norm_eps": 1e-12,
@@ -29,7 +26,7 @@
   "pad_token_id": 0,
   "position_embedding_type": "absolute",
   "torch_dtype": "float32",
-  "transformers_version": "4.41.2",
+  "transformers_version": "4.42.4",
   "type_vocab_size": 2,
   "use_cache": true,
   "vocab_size": 28996

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7915356e386023376eb43a853a5fa069dbcc93a74b99a9536ab30a42379e0f78
-size 430911284
+oid sha256:234c6a38c4fd4ed65f28c898d49ddf69a5a77cbb2459f7807e060ee20ae9ea8b
+size 433270744

runs/Jul13_20-11-13_a518198a24e3/events.out.tfevents.1720901476.a518198a24e3.2562.14

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a63b8cc941fa6f43d26b1fac05a8b0a21ca140bd2b2da92dc7a0394f46a138b9
+size 9867

runs/Jul13_20-11-13_a518198a24e3/events.out.tfevents.1720901584.a518198a24e3.2562.15

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c08c1b10d53a64130a354a73f610568c43fe07947721eb85e212797e9b84d2c
+size 723

tokenizer.json

@@ -1,11 +1,6 @@
 {
   "version": "1.0",
-  "truncation": {
-    "direction": "Right",
-    "max_length": 512,
-    "strategy": "LongestFirst",
-    "stride": 0
-  },
+  "truncation": null,
   "padding": null,
   "added_tokens": [
     {
@@ -59,7 +54,7 @@
     "clean_text": true,
     "handle_chinese_chars": true,
     "strip_accents": null,
-    "lowercase": false
+    "lowercase": true
   },
   "pre_tokenizer": {
     "type": "BertPreTokenizer"

tokenizer_config.json

@@ -44,9 +44,9 @@
   "clean_up_tokenization_spaces": true,
   "cls_token": "[CLS]",
   "do_basic_tokenize": true,
-  "do_lower_case": false,
+  "do_lower_case": true,
   "mask_token": "[MASK]",
-  "model_max_length": 512,
+  "model_max_length": 1000000000000000019884624838656,
   "never_split": null,
   "pad_token": "[PAD]",
   "sep_token": "[SEP]",

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:91044e2900a7f2a0b7e49f54311f5995df446ecc659b5083369e8efb6d06d627
-size 5048
+oid sha256:44604e4feb03bd233f157d2b3797dea1025e6fc6ce9089c21c0bd797adb0f5ca
+size 5112