minang_food_classification

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: 0.7593
• Accuracy: 0.9389

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: 1e-06
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3787	1.0	45	1.3959	0.75
1.283	2.0	90	1.2962	0.7889
1.1978	3.0	135	1.2320	0.8056
1.1274	4.0	180	1.1618	0.8333
1.0807	5.0	225	1.0641	0.8556
1.0178	6.0	270	1.0442	0.8333
0.9573	7.0	315	0.9657	0.8889
0.9072	8.0	360	0.9505	0.8833
0.872	9.0	405	0.9192	0.8944
0.8334	10.0	450	0.9023	0.8722
0.814	11.0	495	0.8054	0.9278
0.7668	12.0	540	0.8127	0.9444
0.7569	13.0	585	0.7832	0.9167
0.7483	14.0	630	0.8084	0.8833
0.7548	15.0	675	0.7762	0.8833
0.7258	16.0	720	0.7698	0.8889
0.7149	17.0	765	0.7466	0.9111
0.7001	18.0	810	0.7655	0.9222
0.7159	19.0	855	0.7529	0.9
0.7073	20.0	900	0.8048	0.8889

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2