vit-base-patch16-224-ve-U13b-80RX1

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: 0.7770
• Accuracy: 0.8478

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3157	0.99	51	1.2968	0.3478
1.0334	2.0	103	1.0060	0.5217
0.691	2.99	154	0.7506	0.7609
0.5005	4.0	206	0.6433	0.7826
0.3478	4.99	257	0.5674	0.7609
0.3339	6.0	309	0.6623	0.7609
0.2533	6.99	360	0.6905	0.7391
0.138	8.0	412	0.7251	0.7826
0.1289	8.99	463	0.7467	0.7391
0.152	10.0	515	0.9011	0.7174
0.2609	10.99	566	1.0150	0.7174
0.2202	12.0	618	0.9713	0.7826
0.1083	12.99	669	1.1106	0.6739
0.07	14.0	721	1.1211	0.7174
0.0791	14.99	772	1.1830	0.7609
0.0427	16.0	824	0.7770	0.8478
0.1219	16.99	875	1.0962	0.7391
0.0739	18.0	927	0.9447	0.7609
0.1989	18.99	978	1.1543	0.7391
0.1097	20.0	1030	1.1795	0.7609
0.1204	20.99	1081	1.2679	0.6739
0.0514	22.0	1133	1.0646	0.7174
0.0612	22.99	1184	1.1413	0.6957
0.0207	24.0	1236	0.8928	0.7826
0.1063	24.99	1287	1.1186	0.7609
0.1076	26.0	1339	1.1741	0.7609
0.0714	26.99	1390	1.0977	0.8043
0.062	28.0	1442	1.3965	0.7174
0.0617	28.99	1493	1.1849	0.7609
0.0536	30.0	1545	1.0865	0.7826
0.0707	30.99	1596	1.2081	0.7609
0.0967	32.0	1648	1.3300	0.7391
0.0564	32.99	1699	1.2240	0.7826
0.0435	34.0	1751	1.2391	0.7609
0.043	34.99	1802	1.1813	0.7609
0.0218	36.0	1854	1.2496	0.7826
0.0043	36.99	1905	1.2797	0.7174
0.0051	38.0	1957	1.2493	0.7391
0.0123	38.99	2008	1.2538	0.7391
0.0546	39.61	2040	1.2530	0.7609

Framework versions

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