README.md

metadata

library_name: transformers
base_model: dmis-lab/biobert-base-cased-v1.1
tags:
  - generated_from_trainer
datasets:
  - ncbi_disease
model-index:
  - name: biobert-ner-model
    results: []

biobert-ner-model

This model is a fine-tuned version of dmis-lab/biobert-base-cased-v1.1 on the ncbi_disease dataset. It achieves the following results on the evaluation set:

• Loss: 0.0557
• Compositemention: {'precision': 0.2765957446808511, 'recall': 0.37142857142857144, 'f1': 0.3170731707317073, 'number': 35}
• Diseaseclass: {'precision': 0.5, 'recall': 0.6031746031746031, 'f1': 0.5467625899280575, 'number': 126}
• Modifier: {'precision': 0.7665198237885462, 'recall': 0.8169014084507042, 'f1': 0.7909090909090909, 'number': 213}
• Specificdisease: {'precision': 0.75625, 'recall': 0.8810679611650486, 'f1': 0.81390134529148, 'number': 412}
• Overall Precision: 0.6909
• Overall Recall: 0.7964
• Overall F1: 0.7400
• Overall Accuracy: 0.9854

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Compositemention	Diseaseclass	Modifier	Specificdisease	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.0688	1.0	75	0.0687	{'precision': 0.075, 'recall': 0.08571428571428572, 'f1': 0.08, 'number': 35}	{'precision': 0.3684210526315789, 'recall': 0.4444444444444444, 'f1': 0.4028776978417266, 'number': 126}	{'precision': 0.5581395348837209, 'recall': 0.676056338028169, 'f1': 0.6114649681528661, 'number': 213}	{'precision': 0.6083788706739527, 'recall': 0.8106796116504854, 'f1': 0.6951092611862643, 'number': 412}	0.5375	0.6832	0.6017	0.9797
0.0416	2.0	150	0.0621	{'precision': 0.038461538461538464, 'recall': 0.02857142857142857, 'f1': 0.03278688524590164, 'number': 35}	{'precision': 0.3706896551724138, 'recall': 0.3412698412698413, 'f1': 0.3553719008264463, 'number': 126}	{'precision': 0.6653225806451613, 'recall': 0.7746478873239436, 'f1': 0.7158351409978307, 'number': 213}	{'precision': 0.6208695652173913, 'recall': 0.866504854368932, 'f1': 0.7234042553191489, 'number': 412}	0.5865	0.7201	0.6465	0.9817
0.0399	3.0	225	0.0537	{'precision': 0.26666666666666666, 'recall': 0.34285714285714286, 'f1': 0.3, 'number': 35}	{'precision': 0.45390070921985815, 'recall': 0.5079365079365079, 'f1': 0.4794007490636704, 'number': 126}	{'precision': 0.7522123893805309, 'recall': 0.7981220657276995, 'f1': 0.7744874715261958, 'number': 213}	{'precision': 0.6927592954990215, 'recall': 0.8592233009708737, 'f1': 0.7670639219934994, 'number': 412}	0.6501	0.7634	0.7022	0.9848
0.0263	4.0	300	0.0548	{'precision': 0.3125, 'recall': 0.42857142857142855, 'f1': 0.3614457831325301, 'number': 35}	{'precision': 0.47096774193548385, 'recall': 0.5793650793650794, 'f1': 0.5195729537366549, 'number': 126}	{'precision': 0.776255707762557, 'recall': 0.7981220657276995, 'f1': 0.7870370370370371, 'number': 213}	{'precision': 0.7433264887063655, 'recall': 0.8786407766990292, 'f1': 0.8053392658509455, 'number': 412}	0.6821	0.7888	0.7316	0.9852
0.0214	5.0	375	0.0557	{'precision': 0.2765957446808511, 'recall': 0.37142857142857144, 'f1': 0.3170731707317073, 'number': 35}	{'precision': 0.5, 'recall': 0.6031746031746031, 'f1': 0.5467625899280575, 'number': 126}	{'precision': 0.7665198237885462, 'recall': 0.8169014084507042, 'f1': 0.7909090909090909, 'number': 213}	{'precision': 0.75625, 'recall': 0.8810679611650486, 'f1': 0.81390134529148, 'number': 412}	0.6909	0.7964	0.7400	0.9854

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1