ultrasound_plane_quality-swin-base-patch4-window7-224_class_weight

This model is a fine-tuned version of microsoft/swin-base-patch4-window7-224 on the HASH Ultrasound Plane Quality Dataset dataset. It achieves the following results on the evaluation set:

Loss: 2.8436
Accuracy: 0.6923
Precision Macro: 0.5501
Recall Macro: 0.5104
F1 Macro: 0.5246
Sensitivity Good: 0.8291
Sensitivity Moderate: 0.3421
Sensitivity Bad: 0.36
Specificity Good: 0.4921
Specificity Moderate: 0.8470
Specificity Bad: 0.9592

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: 4
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 0.1
num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	Sensitivity Good	Sensitivity Moderate	Sensitivity Bad	Specificity Good	Specificity Moderate	Specificity Bad
3.9763	1.0	97	1.1896	0.5747	0.3037	0.3445	0.3196	0.7310	0.3026	0.0	0.3016	0.7268	1.0
4.3623	2.0	194	1.1449	0.6900	0.3011	0.3498	0.3167	0.9494	0.0	0.1	0.0952	1.0	0.9413
4.0362	3.0	291	1.1086	0.6923	0.5769	0.4185	0.4198	0.8861	0.2895	0.08	0.3810	0.8470	0.9949
3.7231	4.0	388	1.1964	0.5995	0.3698	0.4232	0.3588	0.7563	0.0132	0.5	0.3889	0.9836	0.7602
3.8047	5.0	485	1.0374	0.5543	0.4471	0.4966	0.4441	0.6076	0.3421	0.54	0.6746	0.8279	0.7628
3.7615	6.0	582	1.1068	0.6244	0.4489	0.4607	0.4519	0.7405	0.3816	0.26	0.5714	0.8005	0.9005
3.5785	7.0	679	1.1072	0.6199	0.4855	0.4875	0.4749	0.7089	0.4737	0.28	0.6111	0.7432	0.9362
2.7876	8.0	776	1.2497	0.5747	0.4527	0.4980	0.4601	0.6329	0.4211	0.44	0.6111	0.8005	0.8316
2.2637	9.0	873	1.6972	0.6290	0.4766	0.5089	0.4865	0.7247	0.3421	0.46	0.5397	0.8579	0.8622
1.8722	10.0	970	1.7193	0.7059	0.5355	0.4450	0.4586	0.8924	0.3026	0.14	0.3492	0.8907	0.9796
1.6991	11.0	1067	1.9214	0.6041	0.4627	0.4915	0.4664	0.6867	0.4079	0.38	0.6190	0.8497	0.8163
2.0717	12.0	1164	2.2192	0.5701	0.4500	0.4882	0.4561	0.6361	0.3684	0.46	0.6111	0.7568	0.8673
1.6707	13.0	1261	3.1380	0.7036	0.4912	0.4252	0.4372	0.9114	0.1842	0.18	0.2857	0.9372	0.9541
0.9623	14.0	1358	3.2447	0.6946	0.5214	0.4867	0.4990	0.8449	0.3553	0.26	0.4365	0.8798	0.9490
1.7668	15.0	1455	3.3081	0.6471	0.4783	0.4870	0.4807	0.7658	0.3553	0.34	0.5873	0.8087	0.9133
1.3514	16.0	1552	4.2638	0.7014	0.4996	0.4422	0.4545	0.9019	0.1447	0.28	0.3016	0.9290	0.9541
0.7454	17.0	1649	4.9444	0.7127	0.5934	0.4683	0.4920	0.8892	0.3158	0.2	0.3492	0.8934	0.9847
0.7730	18.0	1746	4.9574	0.7195	0.5622	0.4727	0.4959	0.9019	0.2763	0.24	0.3730	0.9071	0.9719
0.7580	19.0	1843	5.7157	0.6923	0.5526	0.4899	0.4934	0.8291	0.4605	0.18	0.5159	0.8197	0.9770

Framework versions

Transformers 5.0.0
Pytorch 2.10.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2

Beijuka/ultrasound_plane_quality-swin-base-patch4-window7-224_class_weight

ultrasound_plane_quality-swin-base-patch4-window7-224_class_weight

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions