vit-base-riego

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.2998
• F1: 0.3729

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: 0.0002
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 16
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1
0.1696	0.79	100	1.1385	0.352
0.08	1.59	200	0.9071	0.3774
0.0928	2.38	300	1.1181	0.3454
0.0189	3.17	400	0.8262	0.3425
0.0728	3.97	500	0.9647	0.3747
0.0756	4.76	600	0.6097	0.4776
0.0018	5.56	700	1.3900	0.3652
0.002	6.35	800	0.7498	0.4606
0.0304	7.14	900	1.4367	0.3666
0.0024	7.94	1000	1.5714	0.3041
0.0463	8.73	1100	0.8038	0.4016
0.0014	9.52	1200	0.7175	0.4795
0.0015	10.32	1300	1.0347	0.3959
0.0009	11.11	1400	1.3881	0.3670
0.0131	11.9	1500	1.0780	0.4044
0.0007	12.7	1600	0.9834	0.4255
0.0011	13.49	1700	1.0753	0.4033
0.0007	14.29	1800	1.1514	0.3989
0.0007	15.08	1900	1.2373	0.3769
0.0007	15.87	2000	1.2998	0.3729

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2