convnext-nano-1e-4-augment
This model is a fine-tuned version of facebook/convnextv2-nano-22k-384 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.2728
- Accuracy: 0.9276
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: 0.0001
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.8282 | 1.0 | 275 | 0.5148 | 0.8537 |
| 0.5209 | 2.0 | 550 | 0.4151 | 0.8839 |
| 0.3867 | 3.0 | 825 | 0.3643 | 0.9010 |
| 0.3183 | 4.0 | 1100 | 0.3241 | 0.9050 |
| 0.2679 | 5.0 | 1375 | 0.3290 | 0.9046 |
| 0.2364 | 6.0 | 1650 | 0.3088 | 0.9137 |
| 0.1981 | 7.0 | 1925 | 0.2982 | 0.9137 |
| 0.1704 | 8.0 | 2200 | 0.2899 | 0.9169 |
| 0.1572 | 9.0 | 2475 | 0.2868 | 0.9201 |
| 0.168 | 10.0 | 2750 | 0.2866 | 0.9205 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.15.2