Imene/vit-base-patch16-224-in21k-wwwwii

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

• Train Loss: 0.8024
• Train Accuracy: 0.9939
• Train Top-3-accuracy: 0.9997
• Validation Loss: 1.6739
• Validation Accuracy: 0.6267
• Validation Top-3-accuracy: 0.8349
• Epoch: 9

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:

• optimizer: {'inner_optimizer': {'class_name': 'AdamWeightDecay', 'config': {'name': 'AdamWeightDecay', 'learning_rate': {'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 4e-05, 'decay_steps': 1620, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}}, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False, 'weight_decay_rate': 0.001}}, 'dynamic': True, 'initial_scale': 32768.0, 'dynamic_growth_steps': 2000}
• training_precision: mixed_float16

Training results

Train Loss	Train Accuracy	Train Top-3-accuracy	Validation Loss	Validation Accuracy	Validation Top-3-accuracy	Epoch
3.6793	0.125	0.2805	3.4078	0.2151	0.4756	0
3.1763	0.3448	0.6265	3.0167	0.4209	0.6640	1
2.7546	0.5419	0.7852	2.6634	0.5326	0.7651	2
2.3537	0.6855	0.8843	2.3971	0.5547	0.7860	3
1.9989	0.7814	0.9279	2.2236	0.5837	0.7907	4
1.6670	0.875	0.9698	2.0757	0.5977	0.7907	5
1.3815	0.9352	0.9890	1.8921	0.6198	0.8174	6
1.1407	0.9651	0.9956	1.7976	0.6244	0.8174	7
0.9451	0.9866	0.9983	1.7227	0.6349	0.8267	8
0.8024	0.9939	0.9997	1.6739	0.6267	0.8349	9

Framework versions

• Transformers 4.21.2
• TensorFlow 2.8.2
• Datasets 2.4.0
• Tokenizers 0.12.1