vit-large-patch16-224-finetuned-eurosat

This model is a fine-tuned version of google/vit-large-patch16-224 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4947
• Accuracy: 0.875
• Precision: 0.8796
• Recall: 0.8760
• F1: 0.8715

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
2.0089	1.0	10	1.4560	0.6188	0.6505	0.6232	0.6086
0.8976	2.0	20	0.4947	0.875	0.8796	0.8760	0.8715
0.3555	3.0	30	0.4390	0.8313	0.8272	0.8339	0.8274
0.1709	4.0	40	0.4177	0.85	0.8506	0.8559	0.8484
0.0882	5.0	50	0.4561	0.8625	0.8630	0.8610	0.8592
0.1136	6.0	60	0.5954	0.8375	0.8494	0.8346	0.8231
0.0863	7.0	70	0.5323	0.8375	0.8353	0.8373	0.8349
0.0637	8.0	80	0.4966	0.8438	0.8579	0.8470	0.8476
0.0536	9.0	90	0.6577	0.8375	0.8598	0.8423	0.8383
0.1069	10.0	100	0.5586	0.8438	0.8417	0.8429	0.8387
0.0446	11.0	110	0.5196	0.85	0.8703	0.8520	0.8520
0.0401	12.0	120	0.4480	0.8625	0.8651	0.8653	0.8632
0.0333	13.0	130	0.4955	0.8375	0.8441	0.8422	0.8393
0.036	14.0	140	0.5074	0.8438	0.8515	0.8477	0.8469
0.0288	15.0	150	0.5466	0.8438	0.8465	0.8441	0.8426
0.0141	16.0	160	0.6208	0.8313	0.8400	0.8324	0.8299
0.0145	17.0	170	0.5696	0.8438	0.8547	0.8458	0.8449
0.0194	18.0	180	0.5469	0.8562	0.8606	0.8575	0.8566
0.0139	19.0	190	0.5600	0.8625	0.8658	0.8670	0.8634
0.0184	20.0	200	0.5611	0.8625	0.8658	0.8675	0.8638

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1