aesthetics_v2
This model is a fine-tuned version of facebook/dinov2-large on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 1.6501
- Accuracy: 0.5581
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: 64
- eval_batch_size: 64
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 256
- 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 |
|---|---|---|---|---|
| 1.1465 | 0.17 | 20 | 1.6860 | 0.5313 |
| 1.2703 | 0.34 | 40 | 1.8412 | 0.5014 |
| 1.3152 | 0.52 | 60 | 1.8200 | 0.5042 |
| 1.2313 | 0.69 | 80 | 1.7971 | 0.5112 |
| 1.3476 | 0.86 | 100 | 1.7649 | 0.5100 |
| 1.2597 | 1.03 | 120 | 1.7454 | 0.5175 |
| 1.0094 | 1.2 | 140 | 1.7356 | 0.5257 |
| 0.9743 | 1.37 | 160 | 1.7074 | 0.5352 |
| 1.0209 | 1.55 | 180 | 1.7331 | 0.5322 |
| 1.0692 | 1.72 | 200 | 1.7370 | 0.5331 |
| 1.0556 | 1.89 | 220 | 1.6788 | 0.5487 |
| 0.8634 | 2.06 | 240 | 1.6644 | 0.5536 |
| 0.79 | 2.23 | 260 | 1.6848 | 0.5531 |
| 0.7916 | 2.4 | 280 | 1.6761 | 0.5528 |
| 0.7454 | 2.58 | 300 | 1.6520 | 0.5534 |
| 0.7497 | 2.75 | 320 | 1.6337 | 0.5554 |
| 0.7537 | 2.92 | 340 | 1.6501 | 0.5581 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.2.0
- Datasets 2.17.1
- Tokenizers 0.15.2