huner_ncbi_disease_dslim

This model is a fine-tuned version of dslim/distilbert-NER on the transformer_dataset_ner dataset. It achieves the following results on the evaluation set:

• Loss: 0.1484
• Precision: 0.8325
• Recall: 0.8653
• F1: 0.8486
• Accuracy: 0.9850

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.1243	1.0	667	0.0669	0.7013	0.8412	0.7649	0.9787
0.0512	2.0	1334	0.0656	0.7825	0.8412	0.8108	0.9818
0.0221	3.0	2001	0.0744	0.7908	0.8501	0.8194	0.9822
0.0107	4.0	2668	0.1022	0.7940	0.8475	0.8199	0.9808
0.008	5.0	3335	0.1055	0.7818	0.8602	0.8191	0.9816
0.0057	6.0	4002	0.1173	0.8067	0.8590	0.832	0.9830
0.0027	7.0	4669	0.1188	0.8188	0.8501	0.8342	0.9834
0.0022	8.0	5336	0.1229	0.8080	0.8450	0.8261	0.9826
0.0019	9.0	6003	0.1341	0.8007	0.8526	0.8258	0.9834
0.0019	10.0	6670	0.1360	0.8045	0.8628	0.8326	0.9822
0.0011	11.0	7337	0.1376	0.8163	0.8640	0.8395	0.9838
0.0008	12.0	8004	0.1447	0.8007	0.8577	0.8282	0.9833
0.0006	13.0	8671	0.1381	0.8139	0.8615	0.8370	0.9839
0.0005	14.0	9338	0.1398	0.8297	0.8666	0.8477	0.9843
0.0004	15.0	10005	0.1404	0.8232	0.8640	0.8431	0.9842
0.0003	16.0	10672	0.1486	0.8329	0.8551	0.8439	0.9838
0.0	17.0	11339	0.1469	0.8114	0.8691	0.8393	0.9837
0.0002	18.0	12006	0.1500	0.8297	0.8602	0.8447	0.9843
0.0001	19.0	12673	0.1489	0.8315	0.8653	0.8481	0.9849
0.0	20.0	13340	0.1484	0.8325	0.8653	0.8486	0.9850

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1