redbioma/deit-cemede-redbioma-ct-fewer-classes

deit-CEMEDE

This model is a fine-tuned version of facebook/deit-base-distilled-patch16-224 on the cemede dataset. It achieves the following results on the evaluation set:

• Loss: 0.4633
• Accuracy: 0.8662
• F1: 0.8361

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: 0.0002
• 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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.6954	0.1805	100	0.7439	0.7835	0.6861
0.5041	0.3610	200	0.6603	0.7600	0.7444
0.4322	0.5415	300	1.3682	0.6844	0.6798
0.5982	0.7220	400	0.5710	0.8080	0.7293
0.4534	0.9025	500	0.4633	0.8662	0.8361
0.0807	1.0830	600	0.6804	0.8437	0.8159
0.272	1.2635	700	0.7165	0.7988	0.7810
0.1832	1.4440	800	0.7701	0.8090	0.7822
0.1927	1.6245	900	0.6627	0.8294	0.8095
0.0959	1.8051	1000	0.5265	0.8468	0.8123
0.1794	1.9856	1100	0.7515	0.8049	0.7926
0.0628	2.1661	1200	0.4937	0.8805	0.8760
0.0888	2.3466	1300	0.5553	0.8621	0.8473
0.0802	2.5271	1400	0.6491	0.8539	0.8409
0.0164	2.7076	1500	0.5064	0.8989	0.9043

Framework versions

• Transformers 4.52.4
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1