Bio_ClinicalBERT_fold_2_ternary_v1

This model is a fine-tuned version of emilyalsentzer/Bio_ClinicalBERT on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.8186
• F1: 0.8038

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	F1
No log	1.0	294	0.5629	0.7645
0.5579	2.0	588	0.5078	0.8078
0.5579	3.0	882	0.6622	0.7754
0.2341	4.0	1176	0.8584	0.7943
0.2341	5.0	1470	1.1953	0.7821
0.0942	6.0	1764	1.3193	0.7876
0.0338	7.0	2058	1.3324	0.7903
0.0338	8.0	2352	1.5043	0.7930
0.0202	9.0	2646	1.5255	0.7889
0.0202	10.0	2940	1.5382	0.7916
0.0119	11.0	3234	1.6377	0.7903
0.0051	12.0	3528	1.7349	0.7835
0.0051	13.0	3822	1.7297	0.7835
0.0082	14.0	4116	1.7817	0.7808
0.0082	15.0	4410	1.7105	0.7970
0.0054	16.0	4704	1.7325	0.7984
0.0054	17.0	4998	1.7919	0.7943
0.0049	18.0	5292	1.8850	0.7876
0.0045	19.0	5586	1.8237	0.7916
0.0045	20.0	5880	1.8760	0.7970
0.0024	21.0	6174	1.8544	0.7984
0.0024	22.0	6468	1.7852	0.8011
0.0005	23.0	6762	1.7795	0.8065
0.0031	24.0	7056	1.7978	0.7997
0.0031	25.0	7350	1.8186	0.8038

Framework versions

• Transformers 4.21.1
• Pytorch 1.12.0+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1