Boya1_Adamax_1-e4_20Epoch_Deit-tiny-patch16_fold4

This model is a fine-tuned version of facebook/deit-tiny-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 3.5379
• Accuracy: 0.5623

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.4184	1.0	923	1.5125	0.4894
1.1581	2.0	1846	1.3440	0.5398
1.0675	3.0	2769	1.2921	0.5683
0.9984	4.0	3692	1.3169	0.5756
0.6437	5.0	4615	1.3971	0.5629
0.491	6.0	5538	1.5307	0.5547
0.3697	7.0	6461	1.6679	0.5615
0.2372	8.0	7384	1.9476	0.5461
0.0824	9.0	8307	2.1631	0.5531
0.0471	10.0	9230	2.4822	0.5485
0.0645	11.0	10153	2.7301	0.5523
0.0461	12.0	11076	2.8827	0.5588
0.0021	13.0	11999	3.1615	0.5575
0.0011	14.0	12922	3.1796	0.5612
0.0141	15.0	13845	3.2737	0.5566
0.0004	16.0	14768	3.3570	0.5593
0.0004	17.0	15691	3.4150	0.5621
0.0003	18.0	16614	3.4800	0.5615
0.0002	19.0	17537	3.5180	0.5615
0.0002	20.0	18460	3.5379	0.5623

Framework versions

• Transformers 4.40.1
• Pytorch 2.1.0
• Datasets 2.19.0
• Tokenizers 0.19.1