vit-base-patch16-224-U8-10b
This model is a fine-tuned version of google/vit-base-patch16-224 on the imagefolder dataset "dmae-ve-U8". It achieves the following results on the evaluation set:
- Loss: 0.5349
- Accuracy: 0.8627
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: 5.5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 1.2473 | 1.0 | 20 | 1.1671 | 0.5882 |
| 0.955 | 2.0 | 40 | 0.9392 | 0.6471 |
| 0.735 | 3.0 | 60 | 0.7247 | 0.6863 |
| 0.5341 | 4.0 | 80 | 0.5977 | 0.8235 |
| 0.3864 | 5.0 | 100 | 0.6556 | 0.7451 |
| 0.2837 | 6.0 | 120 | 0.6781 | 0.7255 |
| 0.2332 | 7.0 | 140 | 0.5419 | 0.8431 |
| 0.1974 | 8.0 | 160 | 0.5349 | 0.8627 |
| 0.1857 | 9.0 | 180 | 0.5606 | 0.8235 |
| 0.1907 | 10.0 | 200 | 0.4875 | 0.8431 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.2+cu118
- Datasets 2.16.1
- Tokenizers 0.15.0