orchid219_ft_vit-large-patch16-224-in21k-finetuned-eurosat

This model is a fine-tuned version of gary109/orchid219_ft_vit-large-patch16-224-in21k on the image_folder dataset. It achieves the following results on the evaluation set:

• Loss: 0.9545
• Accuracy: 0.9231

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: 5e-05
• train_batch_size: 10
• eval_batch_size: 10
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 40
• 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
3.5728	0.96	17	2.1936	0.8718
1.6005	1.96	34	1.2044	0.9359
0.9764	2.96	51	0.9545	0.9231

Framework versions

• Transformers 4.19.2
• Pytorch 1.11.0+cu113
• Datasets 2.2.2
• Tokenizers 0.12.1