swinv2-tiny-patch4-window8-256-dmae-humeda-DAV22

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: 1.0386
• Accuracy: 0.6346

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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: 0.3
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.1723	1.0	22	1.6164	0.1923
2.8734	2.0	44	1.4347	0.3846
2.1553	3.0	66	1.1520	0.4808
1.6674	4.0	88	1.0921	0.4808
1.1204	5.0	110	0.9091	0.5962
1.0373	6.0	132	0.8185	0.6923
0.9181	7.0	154	0.9377	0.6731
0.7475	8.0	176	0.8407	0.6731
0.6679	9.0	198	0.9488	0.6923
0.4914	10.0	220	0.8699	0.7115
0.4421	11.0	242	1.1132	0.6538
0.3759	12.0	264	0.9250	0.7115
0.4317	13.0	286	0.9220	0.6731
0.4137	14.0	308	1.0225	0.7115
0.3451	15.0	330	1.0872	0.6538
0.3482	16.0	352	1.0129	0.6731
0.3346	17.0	374	1.0314	0.6538
0.3105	18.0	396	1.0348	0.6538
0.252	19.0	418	1.0386	0.6346
0.317	19.0930	420	1.0386	0.6346

Framework versions

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