swinv2-small-patch4-window8-256-finetuned-eurosat

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

• Loss: 1.2717
• Accuracy: 0.6875

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	4	0.6579	0.6339
No log	2.0	8	0.7129	0.5
0.6364	3.0	12	0.6774	0.5982
0.6364	4.0	16	0.6584	0.6786
0.3486	5.0	20	0.6864	0.6786
0.3486	6.0	24	0.8473	0.6429
0.3486	7.0	28	0.9735	0.6339
0.1224	8.0	32	0.8121	0.6964
0.1224	9.0	36	1.2379	0.6429
0.0424	10.0	40	1.1585	0.6875
0.0424	11.0	44	1.5274	0.6161
0.0424	12.0	48	1.1415	0.6607
0.0353	13.0	52	1.4422	0.6518
0.0353	14.0	56	1.6677	0.625
0.0141	15.0	60	1.1960	0.6696
0.0141	16.0	64	1.5515	0.625
0.0141	17.0	68	1.7990	0.6161
0.0135	18.0	72	1.4437	0.6607
0.0135	19.0	76	1.2816	0.7054
0.0073	20.0	80	1.2717	0.6875

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.2+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2