swinv2-tiny-patch4-window8-256-dmae-humeda-DAV44

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: 0.7302
• Accuracy: 0.75

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: 42

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	4	1.4782	0.4659
No log	2.0	8	1.3298	0.4432
4.6468	3.0	12	1.2295	0.5341
4.6468	4.0	16	1.1639	0.6477
4.6468	5.0	20	1.0070	0.6477
3.6233	6.0	24	0.9560	0.6477
3.6233	7.0	28	0.8686	0.6932
3.6233	8.0	32	0.8405	0.7045
2.7642	9.0	36	0.8296	0.7045
2.7642	10.0	40	0.8147	0.7159
2.7642	11.0	44	0.8032	0.7386
2.2276	12.0	48	0.7302	0.75
2.2276	13.0	52	0.7815	0.75
2.2276	14.0	56	0.7365	0.7273
2.0873	15.0	60	0.7417	0.75
2.0873	16.0	64	0.7103	0.75
2.0873	17.0	68	0.7166	0.75
1.7268	18.0	72	0.7360	0.7386
1.7268	19.0	76	0.7432	0.7159
1.7268	20.0	80	0.7206	0.7273
1.602	21.0	84	0.7302	0.75
1.602	22.0	88	0.7332	0.7159
1.602	23.0	92	0.7401	0.7045
1.4229	24.0	96	0.7472	0.7273
1.4229	25.0	100	0.7525	0.7273
1.4229	26.0	104	0.7436	0.7273
1.3233	27.0	108	0.7411	0.7273
1.3233	28.0	112	0.7398	0.7273
1.3233	29.0	116	0.7398	0.7159
1.2076	30.0	120	0.7407	0.7273
1.2076	31.0	124	0.7412	0.7273
1.2076	31.6154	126	0.7412	0.7273

Framework versions

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