marcelomoreno26/vit-base-patch-16-384-geoguessr

test-2-geoguessr-55

This model is a fine-tuned version of google/vit-base-patch16-384 on marcelomoreno26/geoguessr dataset. It achieves the following results on the evaluation set:

• Loss: 2.4296
• Accuracy: 0.3881
• F1: 0.1440

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: 2e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
No log	1.0	394	3.3218	0.1603	0.0292
3.5132	2.0	788	3.1195	0.2243	0.0494
3.1361	3.0	1182	2.9820	0.2587	0.0657
2.9576	4.0	1576	2.8774	0.2906	0.0816
2.9576	5.0	1970	2.7940	0.3125	0.0950
2.8298	6.0	2364	2.7271	0.3250	0.1042
2.7378	7.0	2758	2.6721	0.3366	0.1112
2.6526	8.0	3152	2.6268	0.3466	0.1156
2.5971	9.0	3546	2.5882	0.3546	0.1220
2.5971	10.0	3940	2.5558	0.3630	0.1269
2.5468	11.0	4334	2.5286	0.3699	0.1321
2.519	12.0	4728	2.5057	0.3721	0.1337
2.4769	13.0	5122	2.4865	0.3760	0.1359
2.4528	14.0	5516	2.4706	0.3805	0.1387
2.4528	15.0	5910	2.4577	0.3820	0.1391
2.4307	16.0	6304	2.4473	0.3840	0.1410
2.4207	17.0	6698	2.4395	0.3863	0.1428
2.4114	18.0	7092	2.4340	0.3874	0.1437
2.4114	19.0	7486	2.4307	0.3883	0.1440
2.4	20.0	7880	2.4296	0.3881	0.1440

Framework versions

• Transformers 4.39.1
• Pytorch 2.2.1
• Datasets 2.18.0
• Tokenizers 0.15.2