<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
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
