swinv2-tiny-patch4-window8-256-dmae-humeda-DAV37

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.8351
• Accuracy: 0.6538

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• 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: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.8	3	1.6284	0.1346
No log	1.8	6	1.5966	0.2404
No log	2.8	9	1.5076	0.3942
6.28	3.8	12	1.2912	0.4615
6.28	4.8	15	1.2137	0.5096
6.28	5.8	18	1.1917	0.5385
6.28	6.8	21	1.1498	0.5673
2.9539	7.8	24	1.2026	0.5865
2.9539	8.8	27	1.2711	0.5962
2.9539	9.8	30	1.3534	0.625
2.9539	10.8	33	1.3210	0.625
0.9643	11.8	36	1.3940	0.6346
0.9643	12.8	39	1.4859	0.6346
0.9643	13.8	42	1.4965	0.6346
0.9643	14.8	45	1.5463	0.625
0.3275	15.8	48	1.5885	0.6346
0.3275	16.8	51	1.6466	0.6442
0.3275	17.8	54	1.8351	0.6538
0.3275	18.8	57	1.8326	0.6442
0.1501	19.8	60	1.7521	0.6346
0.1501	20.8	63	1.7806	0.6538
0.1501	21.8	66	1.7669	0.6538
0.1501	22.8	69	1.8874	0.6346
0.09	23.8	72	1.8827	0.6538
0.09	24.8	75	1.8330	0.6538
0.09	25.8	78	1.8331	0.6538
0.09	26.8	81	1.8410	0.6538
0.0595	27.8	84	1.8441	0.6442
0.0595	28.8	87	1.8444	0.6538
0.0595	29.8	90	1.8446	0.6538

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0