vit-base-patch16-224-ve-b-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: 0.6432
• Accuracy: 0.8431

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	0.96	6	1.3827	0.3137
1.378	1.92	12	1.3335	0.5490
1.378	2.88	18	1.2577	0.5882
1.2725	4.0	25	1.1886	0.4706
1.1073	4.96	31	1.1040	0.6275
1.1073	5.92	37	1.0658	0.6078
0.9657	6.88	43	1.0155	0.6667
0.8361	8.0	50	0.9330	0.7451
0.8361	8.96	56	0.9690	0.6667
0.7181	9.92	62	0.8910	0.7255
0.7181	10.88	68	0.8953	0.6863
0.6126	12.0	75	0.8343	0.7451
0.5096	12.96	81	0.8048	0.7059
0.5096	13.92	87	0.7977	0.7059
0.4348	14.88	93	0.7250	0.7451
0.4011	16.0	100	0.6432	0.8431
0.4011	16.96	106	0.7317	0.7255
0.3292	17.92	112	0.7015	0.7451
0.3292	18.88	118	0.6248	0.7647
0.309	20.0	125	0.6990	0.7451
0.2744	20.96	131	0.6591	0.7843
0.2744	21.92	137	0.6452	0.7647
0.2864	22.88	143	0.6290	0.7843
0.2864	23.04	144	0.6285	0.7843

Framework versions

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