vit-invitrace-food

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: 0.1286
• Accuracy: 0.9684

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: 0.0002
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6128	0.2132	100	0.4694	0.9044
0.3905	0.4264	200	0.5236	0.8484
0.4315	0.6397	300	0.3988	0.8884
0.4028	0.8529	400	0.2213	0.9432
0.1097	1.0661	500	0.2963	0.92
0.1883	1.2793	600	0.2047	0.9448
0.137	1.4925	700	0.1695	0.9548
0.2309	1.7058	800	0.2159	0.9384
0.094	1.9190	900	0.1987	0.9452
0.0282	2.1322	1000	0.1861	0.9528
0.0231	2.3454	1100	0.1944	0.9476
0.0409	2.5586	1200	0.1625	0.96
0.0386	2.7719	1300	0.1486	0.9616
0.0249	2.9851	1400	0.1736	0.9572
0.012	3.1983	1500	0.1469	0.9624
0.0304	3.4115	1600	0.1405	0.9644
0.0052	3.6247	1700	0.1498	0.9636
0.0247	3.8380	1800	0.1286	0.9684

Framework versions

• Transformers 4.40.0
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1