vit-base-patch16-224-ve-U10-24

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

• Loss: 0.7874
• Accuracy: 0.7647

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: 5.5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 24

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3749	0.95	15	1.2899	0.4706
1.1573	1.97	31	1.0844	0.5686
0.985	2.98	47	0.9140	0.6078
0.7024	4.0	63	0.8578	0.6863
0.5699	4.95	78	0.6802	0.7451
0.3784	5.97	94	0.8856	0.7059
0.2631	6.98	110	0.7526	0.7451
0.2201	8.0	126	0.7924	0.7255
0.1933	8.95	141	0.7874	0.7647
0.1592	9.97	157	0.9583	0.6863
0.154	10.98	173	0.9961	0.7059
0.1531	12.0	189	0.8916	0.7451
0.1153	12.95	204	0.9174	0.7451
0.1154	13.97	220	1.0267	0.7059
0.0922	14.98	236	0.9766	0.7255
0.0901	16.0	252	1.0410	0.7255
0.074	16.95	267	1.1869	0.6863
0.0743	17.97	283	1.1094	0.7255
0.084	18.98	299	1.0520	0.7255
0.0713	20.0	315	1.1213	0.7059
0.061	20.95	330	1.0927	0.7451
0.0669	21.97	346	1.0806	0.7255
0.0654	22.86	360	1.0647	0.7255

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0