vit-base-patch16-224-ve-U12-b-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: 0.6456
• 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: 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
No log	0.92	6	1.3806	0.4130
1.379	2.0	13	1.3103	0.5435
1.379	2.92	19	1.2269	0.4130
1.2758	4.0	26	1.1412	0.4565
1.121	4.92	32	1.0650	0.4783
1.121	6.0	39	1.0084	0.5217
0.9871	6.92	45	0.9395	0.6522
0.8612	8.0	52	0.8798	0.7174
0.8612	8.92	58	0.8219	0.7391
0.7653	10.0	65	0.7712	0.7826
0.6674	10.92	71	0.7328	0.7609
0.6674	12.0	78	0.6968	0.7391
0.568	12.92	84	0.6456	0.8478
0.4723	14.0	91	0.6528	0.8043
0.4723	14.92	97	0.7107	0.6739
0.4256	16.0	104	0.6335	0.7609
0.3524	16.92	110	0.5953	0.8261
0.3524	18.0	117	0.5824	0.8261
0.3282	18.92	123	0.6329	0.7174
0.3074	20.0	130	0.5775	0.8043
0.3074	20.92	136	0.5770	0.8043
0.3076	22.0	143	0.5749	0.8261
0.3076	22.15	144	0.5747	0.8261

Framework versions

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