gca_tab_ckpts

This model is a fine-tuned version of austin/mimic-pubmed-deberta-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0831
• F1 Adventitial inflammation: 0.4
• F1 Gca: 0.8889
• F1 Giant cells: 1.0
• F1 Intimal hyperplasia: 0.8
• Acc Adventitial inflammation: 0.9211
• Acc Gca: 0.9737
• Acc Giant cells: 1.0
• Acc Intimal hyperplasia: 0.9737
• Auc Adventitial inflammation: 0.9306
• Auc Gca: 0.9939
• Auc Giant cells: 1.0
• Auc Intimal hyperplasia: 0.9861

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Adventitial inflammation	F1 Gca	F1 Giant cells	F1 Intimal hyperplasia	Acc Adventitial inflammation	Acc Gca	Acc Giant cells	Acc Intimal hyperplasia	Auc Adventitial inflammation	Auc Gca	Auc Giant cells	Auc Intimal hyperplasia
0.5304	1.0	11	0.2758	0.0	0.0	0.0	0.0	0.9474	0.8684	0.9211	0.9474	0.0417	0.4182	0.0952	0.9444
0.4395	2.0	22	0.2832	0.0	0.0	0.0	0.0	0.9474	0.8684	0.9211	0.9474	0.9167	0.8909	0.9333	0.9167
0.4243	3.0	33	0.2308	0.0	0.0	0.0	0.0	0.9474	0.8684	0.9211	0.9474	0.9167	0.9394	0.9810	0.9722
0.2879	4.0	44	0.1196	0.4	0.8889	0.8571	0.6667	0.9211	0.9737	0.9737	0.9474	0.9444	0.9939	1.0	0.9861
0.1956	5.0	55	0.0871	0.6667	1.0	0.8571	0.5714	0.9474	1.0	0.9737	0.9211	0.9583	1.0	1.0	1.0
0.2323	6.0	66	0.0961	0.4	0.9091	0.8571	0.5714	0.9211	0.9737	0.9737	0.9211	0.9444	1.0	1.0	1.0
0.1629	7.0	77	0.1099	0.4	1.0	1.0	0.8	0.9211	1.0	1.0	0.9737	0.9444	1.0	1.0	0.9722
0.1599	8.0	88	0.0819	0.4	0.7500	1.0	0.8	0.9211	0.9474	1.0	0.9737	0.9583	1.0	1.0	1.0
0.1491	9.0	99	0.0835	0.4	0.7500	1.0	0.8	0.9211	0.9474	1.0	0.9737	0.9444	1.0	1.0	0.9861
0.1284	10.0	110	0.0775	0.4	0.8889	1.0	0.8	0.9211	0.9737	1.0	0.9737	0.9306	1.0	1.0	0.9722
0.1248	11.0	121	0.0782	0.4	0.8889	1.0	0.8	0.9211	0.9737	1.0	0.9737	0.9306	1.0	1.0	0.9861
0.1209	12.0	132	0.0786	0.4	0.8889	1.0	0.8	0.9211	0.9737	1.0	0.9737	0.9306	1.0	1.0	0.9861
0.1164	13.0	143	0.0810	0.4	0.8889	1.0	0.8	0.9211	0.9737	1.0	0.9737	0.9167	0.9939	1.0	0.9861
0.1167	14.0	154	0.0813	0.4	0.8889	1.0	0.8	0.9211	0.9737	1.0	0.9737	0.9306	0.9939	1.0	0.9722
0.1216	15.0	165	0.0831	0.4	0.8889	1.0	0.8	0.9211	0.9737	1.0	0.9737	0.9306	0.9939	1.0	0.9861

Framework versions

• Transformers 4.36.0
• Pytorch 2.1.0+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0