swin-tiny-patch4-window7-224-finetuned-eurosat

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.2235
• Accuracy: 0.9167

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: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 24

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.86	3	0.5572	0.7917
No log	2.0	7	0.4659	0.75
0.4694	2.86	10	0.4492	0.7917
0.4694	4.0	14	0.2875	0.875
0.4694	4.86	17	0.2463	0.875
0.3403	6.0	21	0.2235	0.9167
0.3403	6.86	24	0.2371	0.9167
0.3403	8.0	28	0.1865	0.9167
0.2581	8.86	31	0.3179	0.8333
0.2581	10.0	35	0.2050	0.8333
0.2581	10.86	38	0.2885	0.8333
0.192	12.0	42	0.2371	0.7917
0.192	12.86	45	0.1783	0.875
0.192	14.0	49	0.1164	0.9167
0.1479	14.86	52	0.1250	0.9167
0.1479	16.0	56	0.1491	0.875
0.1479	16.86	59	0.1409	0.875
0.1348	18.0	63	0.1192	0.9167
0.1348	18.86	66	0.1168	0.9167
0.1461	20.0	70	0.1106	0.9167
0.1461	20.57	72	0.1101	0.9167

Framework versions

• Transformers 4.27.4
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3