jwt2706/google-vit-base-patch16-224-in21k-finetuned-food-classification-86M-v0.1

food-classification-86M-v0.1

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the food101 dataset. It achieves the following results on the evaluation set:

• Loss: 1.6079
• Accuracy: 0.892

Model description

Food image classification.

Intended uses & limitations

This was trained for fun and my own learning. But if you want to use it, go ahead.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.7263	0.99	62	2.5435	0.816
1.8437	2.0	125	1.7773	0.863
1.5811	2.98	186	1.6079	0.892

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2