vit-base-patch16-224-ve-U11-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: 0.4436
• Accuracy: 0.9130

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	1.3798	0.5435
1.3792	2.0	13	1.3091	0.6522
1.3792	2.92	19	1.2227	0.5870
1.2783	4.0	26	1.1263	0.6087
1.1226	4.92	32	1.0466	0.6522
1.1226	6.0	39	0.9854	0.5870
0.9881	6.92	45	0.9303	0.6957
0.8707	8.0	52	0.8806	0.7826
0.8707	8.92	58	0.8234	0.7826
0.7604	10.0	65	0.7159	0.8261
0.6452	10.92	71	0.6929	0.8478
0.6452	12.0	78	0.6491	0.8696
0.5576	12.92	84	0.5924	0.8478
0.4708	14.0	91	0.5551	0.8478
0.4708	14.92	97	0.6354	0.8043
0.422	16.0	104	0.5130	0.8696
0.3546	16.92	110	0.5302	0.8696
0.3546	18.0	117	0.4436	0.9130
0.3353	18.92	123	0.5621	0.8261
0.3106	20.0	130	0.4912	0.8696
0.3106	20.92	136	0.4747	0.8913
0.312	22.0	143	0.4603	0.8913
0.312	22.15	144	0.4598	0.8913

Framework versions

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