vit-base-patch16-224-ve-b-U10-40

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: 0.5211
• 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: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.96	6	1.3845	0.2549
1.3817	1.92	12	1.3529	0.4706
1.3817	2.88	18	1.2772	0.5882
1.2986	4.0	25	1.2121	0.3922
1.1298	4.96	31	1.1164	0.5882
1.1298	5.92	37	1.0879	0.5882
0.9842	6.88	43	0.9898	0.6863
0.8402	8.0	50	0.9233	0.7843
0.8402	8.96	56	0.9650	0.6471
0.7084	9.92	62	0.8243	0.7451
0.7084	10.88	68	0.7988	0.7647
0.5914	12.0	75	0.8114	0.7451
0.461	12.96	81	0.7652	0.7451
0.461	13.92	87	0.7406	0.7451
0.3769	14.88	93	0.6916	0.7451
0.3376	16.0	100	0.6182	0.7843
0.3376	16.96	106	0.8395	0.6863
0.2606	17.92	112	0.6941	0.7255
0.2606	18.88	118	0.7345	0.7255
0.2314	20.0	125	0.7374	0.7059
0.1907	20.96	131	0.7490	0.7647
0.1907	21.92	137	0.7292	0.7255
0.1804	22.88	143	0.7301	0.7451
0.1447	24.0	150	0.7224	0.7647
0.1447	24.96	156	0.7415	0.7255
0.1537	25.92	162	0.6668	0.7843
0.1537	26.88	168	0.7188	0.7451
0.1471	28.0	175	0.7291	0.7451
0.1241	28.96	181	0.5919	0.8039
0.1241	29.92	187	0.5211	0.8431
0.1058	30.88	193	0.6107	0.7843
0.1032	32.0	200	0.6863	0.7647
0.1032	32.96	206	0.6295	0.7647
0.1116	33.92	212	0.6061	0.7843
0.1116	34.88	218	0.6610	0.7843
0.0871	36.0	225	0.6109	0.8039
0.1037	36.96	231	0.6116	0.7843
0.1037	37.92	237	0.6176	0.8039
0.0802	38.4	240	0.6169	0.8039

Framework versions

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