smids_3x_deit_tiny_sgd_0001_fold4
This model is a fine-tuned version of facebook/deit-tiny-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.6133
- Accuracy: 0.745
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: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 1.2288 | 1.0 | 225 | 1.2017 | 0.3733 |
| 1.1101 | 2.0 | 450 | 1.1316 | 0.3717 |
| 1.1013 | 3.0 | 675 | 1.0936 | 0.395 |
| 1.0586 | 4.0 | 900 | 1.0632 | 0.42 |
| 1.0359 | 5.0 | 1125 | 1.0361 | 0.4367 |
| 0.9983 | 6.0 | 1350 | 1.0106 | 0.4633 |
| 1.019 | 7.0 | 1575 | 0.9868 | 0.475 |
| 0.9201 | 8.0 | 1800 | 0.9635 | 0.4933 |
| 0.9378 | 9.0 | 2025 | 0.9411 | 0.5117 |
| 0.9004 | 10.0 | 2250 | 0.9202 | 0.535 |
| 0.8802 | 11.0 | 2475 | 0.9003 | 0.5517 |
| 0.9236 | 12.0 | 2700 | 0.8818 | 0.565 |
| 0.8457 | 13.0 | 2925 | 0.8636 | 0.5817 |
| 0.7849 | 14.0 | 3150 | 0.8469 | 0.5917 |
| 0.8073 | 15.0 | 3375 | 0.8311 | 0.6017 |
| 0.8252 | 16.0 | 3600 | 0.8159 | 0.61 |
| 0.7763 | 17.0 | 3825 | 0.8016 | 0.6267 |
| 0.785 | 18.0 | 4050 | 0.7881 | 0.6333 |
| 0.7828 | 19.0 | 4275 | 0.7746 | 0.64 |
| 0.7255 | 20.0 | 4500 | 0.7617 | 0.6483 |
| 0.7412 | 21.0 | 4725 | 0.7495 | 0.6567 |
| 0.7373 | 22.0 | 4950 | 0.7382 | 0.665 |
| 0.7527 | 23.0 | 5175 | 0.7270 | 0.6667 |
| 0.6946 | 24.0 | 5400 | 0.7170 | 0.67 |
| 0.6784 | 25.0 | 5625 | 0.7075 | 0.6817 |
| 0.6623 | 26.0 | 5850 | 0.6988 | 0.6933 |
| 0.6734 | 27.0 | 6075 | 0.6906 | 0.6983 |
| 0.6676 | 28.0 | 6300 | 0.6828 | 0.7083 |
| 0.6739 | 29.0 | 6525 | 0.6758 | 0.7117 |
| 0.6897 | 30.0 | 6750 | 0.6692 | 0.715 |
| 0.6092 | 31.0 | 6975 | 0.6631 | 0.7167 |
| 0.5885 | 32.0 | 7200 | 0.6574 | 0.7217 |
| 0.618 | 33.0 | 7425 | 0.6519 | 0.7217 |
| 0.5886 | 34.0 | 7650 | 0.6471 | 0.7233 |
| 0.587 | 35.0 | 7875 | 0.6427 | 0.7233 |
| 0.6289 | 36.0 | 8100 | 0.6385 | 0.7233 |
| 0.6176 | 37.0 | 8325 | 0.6348 | 0.7267 |
| 0.6111 | 38.0 | 8550 | 0.6313 | 0.7283 |
| 0.5626 | 39.0 | 8775 | 0.6283 | 0.7317 |
| 0.5867 | 40.0 | 9000 | 0.6256 | 0.735 |
| 0.6103 | 41.0 | 9225 | 0.6232 | 0.7367 |
| 0.6145 | 42.0 | 9450 | 0.6210 | 0.7417 |
| 0.5742 | 43.0 | 9675 | 0.6191 | 0.7417 |
| 0.6152 | 44.0 | 9900 | 0.6175 | 0.745 |
| 0.6028 | 45.0 | 10125 | 0.6162 | 0.745 |
| 0.5964 | 46.0 | 10350 | 0.6151 | 0.745 |
| 0.5533 | 47.0 | 10575 | 0.6143 | 0.745 |
| 0.597 | 48.0 | 10800 | 0.6137 | 0.745 |
| 0.617 | 49.0 | 11025 | 0.6134 | 0.745 |
| 0.5745 | 50.0 | 11250 | 0.6133 | 0.745 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.1+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2