tmvar_5e-05_ES12

This model is a fine-tuned version of microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0154
• Precision: 0.8549
• Recall: 0.8919
• F1: 0.8730
• Accuracy: 0.9967

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.2914	1.47	25	0.1017	0.0	0.0	0.0	0.9843
0.0755	2.94	50	0.0449	0.2811	0.2811	0.2811	0.9857
0.0322	4.41	75	0.0272	0.3946	0.4757	0.4314	0.9906
0.0218	5.88	100	0.0227	0.5191	0.6595	0.5810	0.9924
0.0081	7.35	125	0.0144	0.7949	0.8378	0.8158	0.9963
0.0038	8.82	150	0.0129	0.8639	0.8919	0.8777	0.9971
0.0021	10.29	175	0.0129	0.865	0.9351	0.8987	0.9976
0.0014	11.76	200	0.0122	0.8923	0.9405	0.9158	0.9980
0.001	13.24	225	0.0121	0.8677	0.8865	0.8770	0.9976
0.001	14.71	250	0.0118	0.8934	0.9514	0.9215	0.9981
0.001	16.18	275	0.0162	0.8901	0.8757	0.8828	0.9967
0.0005	17.65	300	0.0117	0.8838	0.9459	0.9138	0.9982
0.0008	19.12	325	0.0119	0.8788	0.9405	0.9086	0.9981
0.0007	20.59	350	0.0139	0.8958	0.9297	0.9125	0.9978
0.0009	22.06	375	0.0141	0.8673	0.9189	0.8924	0.9975
0.0008	23.53	400	0.0136	0.8964	0.9351	0.9153	0.9977
0.0005	25.0	425	0.0140	0.8953	0.9243	0.9096	0.9976
0.0005	26.47	450	0.0132	0.8744	0.9405	0.9062	0.9981
0.0005	27.94	475	0.0132	0.8788	0.9405	0.9086	0.9978
0.0014	29.41	500	0.0170	0.8610	0.8703	0.8656	0.9968
0.0023	30.88	525	0.0258	0.7845	0.7676	0.7760	0.9955
0.0007	32.35	550	0.0168	0.8135	0.8486	0.8307	0.9962
0.0006	33.82	575	0.0193	0.8804	0.8757	0.8780	0.9969
0.0004	35.29	600	0.0154	0.8549	0.8919	0.8730	0.9967

Framework versions

• Transformers 4.27.4
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.2