finetuned-mango-types

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:

• Loss: 0.5751
• Accuracy: 0.9292

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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.9926	1.0	22	1.9526	0.3833
1.7976	2.0	44	1.7500	0.6083
1.5678	3.0	66	1.5025	0.7583
1.3907	4.0	88	1.2804	0.9
1.0873	5.0	110	1.1005	0.9042
0.9511	6.0	132	1.0130	0.8875
0.8476	7.0	154	0.9424	0.8833
0.7511	8.0	176	0.8325	0.9042
0.6985	9.0	198	0.7894	0.9083
0.6515	10.0	220	0.8052	0.8792
0.5775	11.0	242	0.7600	0.8792
0.5458	12.0	264	0.6684	0.925
0.5331	13.0	286	0.7148	0.8917
0.4823	14.0	308	0.6849	0.9125
0.4579	15.0	330	0.6414	0.9167
0.4435	16.0	352	0.6557	0.8833
0.4411	17.0	374	0.5968	0.9083
0.453	18.0	396	0.5751	0.9292
0.445	19.0	418	0.6035	0.9083
0.4357	20.0	440	0.6010	0.9083

Framework versions

• Transformers 4.38.1
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.2