swinv2-tiny-patch4-window8-256-dmae-humeda-DAV9

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: 1.2286
• Accuracy: 0.5192

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: 3e-05
• train_batch_size: 64
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• 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: 0.2
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	3	1.5944	0.2308
No log	2.0	6	1.5511	0.2692
No log	3.0	9	1.4915	0.3462
6.2974	4.0	12	1.4388	0.4615
6.2974	5.0	15	1.3927	0.4615
6.2974	6.0	18	1.3394	0.4423
5.3611	7.0	21	1.3108	0.4423
5.3611	8.0	24	1.3680	0.3462
5.3611	9.0	27	1.2718	0.4038
3.7205	10.0	30	1.2679	0.4231
3.7205	11.0	33	1.3010	0.4038
3.7205	12.0	36	1.2598	0.4231
3.7205	13.0	39	1.2016	0.4231
2.8178	14.0	42	1.1934	0.4423
2.8178	15.0	45	1.1842	0.4808
2.8178	16.0	48	1.1539	0.5
2.4001	17.0	51	1.1308	0.4808
2.4001	18.0	54	1.2173	0.4615
2.4001	19.0	57	1.1670	0.5
2.081	20.0	60	1.1792	0.5
2.081	21.0	63	1.2286	0.5192
2.081	22.0	66	1.2633	0.5
2.081	23.0	69	1.2380	0.5
1.8588	24.0	72	1.2498	0.4808
1.8588	25.0	75	1.2591	0.5
1.8588	26.0	78	1.2653	0.5
1.7634	27.0	81	1.2599	0.5
1.7634	28.0	84	1.2549	0.5
1.7634	29.0	87	1.2545	0.5192
1.8177	30.0	90	1.2547	0.5192

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0