This model is a fine-tuned version of google/vit-base-patch16-224 on 71 Dog Breeds-Image Data Set (Kaggle). It achieves the following results on the evaluation set:

Loss: 0.0763
Accuracy: 0.9743

Model description

This Model is a Transfer Learning-based model and trained with the size of 224x224 pixels. This model can predict dog with 71 classes of breeds.

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: 32
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4379	1.0	249	0.2430	0.93
0.1998	2.0	498	0.1380	0.9514
0.0739	3.0	747	0.1008	0.9614
0.0135	4.0	996	0.0834	0.9671
0.0036	5.0	1245	0.0763	0.9743

Framework versions

Transformers 4.44.2
Pytorch 2.4.1+cu121
Datasets 3.0.1
Tokenizers 0.19.1

raffaelsiregar/dog-breeds-classification

Dog Breeds Classification