convnext-tiny-224-finetuned-brs

This model is a fine-tuned version of facebook/convnext-tiny-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.8667
• Accuracy: 0.8235
• F1: 0.7273
• Precision (ppv): 0.8
• Recall (sensitivity): 0.6667
• Specificity: 0.9091
• Npv: 0.8333
• Auc: 0.7879

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: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision (ppv)	Recall (sensitivity)	Specificity	Npv	Auc
0.6766	6.25	100	0.7002	0.4706	0.5263	0.3846	0.8333	0.2727	0.75	0.5530
0.6408	12.49	200	0.6770	0.6471	0.5714	0.5	0.6667	0.6364	0.7778	0.6515
0.464	18.74	300	0.6624	0.5882	0.5882	0.4545	0.8333	0.4545	0.8333	0.6439
0.4295	24.98	400	0.6938	0.5294	0.5	0.4	0.6667	0.4545	0.7143	0.5606
0.3952	31.25	500	0.5974	0.7059	0.6154	0.5714	0.6667	0.7273	0.8	0.6970
0.1082	37.49	600	0.6163	0.6471	0.5	0.5	0.5	0.7273	0.7273	0.6136
0.1997	43.74	700	0.6155	0.7059	0.6154	0.5714	0.6667	0.7273	0.8	0.6970
0.1267	49.98	800	0.9063	0.6471	0.5714	0.5	0.6667	0.6364	0.7778	0.6515
0.1178	56.25	900	0.8672	0.7059	0.6667	0.5556	0.8333	0.6364	0.875	0.7348
0.2008	62.49	1000	0.7049	0.8235	0.7692	0.7143	0.8333	0.8182	0.9	0.8258
0.0996	68.74	1100	0.4510	0.8235	0.7692	0.7143	0.8333	0.8182	0.9	0.8258
0.0115	74.98	1200	0.7561	0.8235	0.7692	0.7143	0.8333	0.8182	0.9	0.8258
0.0177	81.25	1300	1.0400	0.7059	0.6667	0.5556	0.8333	0.6364	0.875	0.7348
0.0261	87.49	1400	0.9139	0.8235	0.7692	0.7143	0.8333	0.8182	0.9	0.8258
0.028	93.74	1500	0.7367	0.7647	0.7143	0.625	0.8333	0.7273	0.8889	0.7803
0.0056	99.98	1600	0.8667	0.8235	0.7273	0.8	0.6667	0.9091	0.8333	0.7879

Framework versions

• Transformers 4.23.1
• Pytorch 1.12.1+cu113
• Datasets 2.6.1
• Tokenizers 0.13.1