vit-base-patch16-224-in21k-finetuned-mgasior-07-02-2024
This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.8842
- F1: 0.7717
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: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | F1 |
|---|---|---|---|---|
| 1.731 | 0.98 | 35 | 1.6748 | 0.3386 |
| 1.5196 | 1.99 | 71 | 1.4890 | 0.4173 |
| 1.3727 | 2.99 | 107 | 1.2938 | 0.5276 |
| 1.2194 | 4.0 | 143 | 1.1519 | 0.6457 |
| 1.1538 | 4.98 | 178 | 1.0544 | 0.6693 |
| 1.0379 | 5.99 | 214 | 0.9852 | 0.7165 |
| 1.0232 | 6.99 | 250 | 0.9439 | 0.7323 |
| 0.9586 | 8.0 | 286 | 0.9136 | 0.7480 |
| 0.9374 | 8.98 | 321 | 0.8946 | 0.7638 |
| 0.96 | 9.79 | 350 | 0.8842 | 0.7717 |
Framework versions
- Transformers 4.36.1
- Pytorch 2.1.2+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0