vit-base-patch16-224-in21k-finetuned

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.0051
• Accuracy: 1.0

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: 2e-05
• train_batch_size: 10
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	90	0.1809	0.9911
No log	2.0	180	0.0815	0.9911
No log	3.0	270	0.0542	0.9911
No log	4.0	360	0.0298	1.0
No log	5.0	450	0.0312	0.9955
0.1429	6.0	540	0.0235	1.0
0.1429	7.0	630	0.0196	1.0
0.1429	8.0	720	0.0154	1.0
0.1429	9.0	810	0.0145	1.0
0.1429	10.0	900	0.0125	1.0
0.1429	11.0	990	0.0115	1.0
0.0196	12.0	1080	0.0167	0.9955
0.0196	13.0	1170	0.0102	1.0
0.0196	14.0	1260	0.0093	1.0
0.0196	15.0	1350	0.0085	1.0
0.0196	16.0	1440	0.0079	1.0
0.0148	17.0	1530	0.0075	1.0
0.0148	18.0	1620	0.0074	1.0
0.0148	19.0	1710	0.0069	1.0
0.0148	20.0	1800	0.0065	1.0
0.0148	21.0	1890	0.0062	1.0
0.0148	22.0	1980	0.0062	1.0
0.0069	23.0	2070	0.0057	1.0
0.0069	24.0	2160	0.0055	1.0
0.0069	25.0	2250	0.0054	1.0
0.0069	26.0	2340	0.0053	1.0
0.0069	27.0	2430	0.0052	1.0
0.0055	28.0	2520	0.0051	1.0
0.0055	29.0	2610	0.0051	1.0
0.0055	30.0	2700	0.0051	1.0

Framework versions

• Transformers 4.38.1
• Pytorch 1.10.0+cu111
• Datasets 2.17.1
• Tokenizers 0.15.2