This model is a fine-tuned version of google/vit-base-patch16-224 on the custom aircraft dataset (20 classes, ~100 imgs each, Wikimedia Commons) dataset. It achieves the following results on the evaluation set:

Loss: 1.4667
Accuracy: 0.675

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: 32
seed: 42
optimizer: Use OptimizerNames.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_steps: 0.1
num_epochs: 15
label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.7041	1.0	100	2.6548	0.185
2.0550	2.0	200	2.0971	0.41
1.7654	3.0	300	1.9024	0.435
1.5199	4.0	400	1.7441	0.52
1.4159	5.0	500	1.6400	0.58
1.2597	6.0	600	1.5958	0.605
1.1279	7.0	700	1.5441	0.645
1.0317	8.0	800	1.5333	0.63
1.0348	9.0	900	1.4501	0.685
0.9480	10.0	1000	1.4456	0.68
0.8717	11.0	1100	1.4176	0.68
0.8424	12.0	1200	1.4068	0.705
0.8521	13.0	1300	1.4088	0.69
0.8423	14.0	1400	1.4092	0.695
0.8189	15.0	1500	1.4049	0.695

Framework versions

Transformers 5.0.0
Pytorch 2.2.2
Datasets 4.8.4
Tokenizers 0.22.2

dubattim/aircraft-vit-fs26

aircraft-vit-fs26

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

dubattim/
aircraft-vit-fs26