swinv2-tiny-patch4-window8-256-dmae-humeda-DAV40

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: 0.7428
• Accuracy: 0.7614

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: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	4	1.5349	0.3409
No log	2.0	8	1.3213	0.4432
4.7629	3.0	12	1.2541	0.4432
4.7629	4.0	16	1.2072	0.6023
4.7629	5.0	20	1.1313	0.6364
3.7987	6.0	24	1.0712	0.6477
3.7987	7.0	28	0.9677	0.6591
3.7987	8.0	32	0.8655	0.7159
3.0437	9.0	36	0.8564	0.6818
3.0437	10.0	40	0.8003	0.6818
3.0437	11.0	44	0.7987	0.7386
2.4867	12.0	48	0.7619	0.7159
2.4867	13.0	52	0.7426	0.7386
2.4867	14.0	56	0.7492	0.6932
2.147	15.0	60	0.7827	0.7159
2.147	16.0	64	0.7509	0.7045
2.147	17.0	68	0.7364	0.7386
1.8443	18.0	72	0.7705	0.7159
1.8443	19.0	76	0.7515	0.7273
1.8443	20.0	80	0.7470	0.7386
1.659	21.0	84	0.7495	0.75
1.659	22.0	88	0.7237	0.75
1.659	23.0	92	0.7440	0.75
1.5303	24.0	96	0.7367	0.75
1.5303	25.0	100	0.7428	0.7614
1.5303	26.0	104	0.7407	0.75
1.4305	27.0	108	0.7406	0.75
1.4305	28.0	112	0.7423	0.75
1.4305	29.0	116	0.7427	0.75
1.3529	30.0	120	0.7428	0.75

Framework versions

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