hushem_40x_deit_tiny_sgd_0001_fold2
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: 1.2757
- Accuracy: 0.4667
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 |
|---|---|---|---|---|
| 1.3997 | 1.0 | 215 | 1.4647 | 0.2444 |
| 1.3365 | 2.0 | 430 | 1.4330 | 0.2222 |
| 1.2917 | 3.0 | 645 | 1.4094 | 0.2444 |
| 1.2427 | 4.0 | 860 | 1.3927 | 0.2667 |
| 1.2441 | 5.0 | 1075 | 1.3794 | 0.2667 |
| 1.1872 | 6.0 | 1290 | 1.3684 | 0.2667 |
| 1.1795 | 7.0 | 1505 | 1.3589 | 0.3111 |
| 1.1209 | 8.0 | 1720 | 1.3504 | 0.3333 |
| 1.1403 | 9.0 | 1935 | 1.3427 | 0.3778 |
| 1.0825 | 10.0 | 2150 | 1.3360 | 0.3778 |
| 1.0205 | 11.0 | 2365 | 1.3306 | 0.3778 |
| 1.0287 | 12.0 | 2580 | 1.3256 | 0.4222 |
| 1.0526 | 13.0 | 2795 | 1.3201 | 0.4444 |
| 0.979 | 14.0 | 3010 | 1.3158 | 0.4444 |
| 1.009 | 15.0 | 3225 | 1.3119 | 0.4444 |
| 1.0242 | 16.0 | 3440 | 1.3068 | 0.4444 |
| 0.9586 | 17.0 | 3655 | 1.3041 | 0.4222 |
| 0.9705 | 18.0 | 3870 | 1.3009 | 0.4222 |
| 0.9559 | 19.0 | 4085 | 1.2993 | 0.4222 |
| 0.95 | 20.0 | 4300 | 1.2983 | 0.4444 |
| 0.9501 | 21.0 | 4515 | 1.2955 | 0.4444 |
| 0.9287 | 22.0 | 4730 | 1.2949 | 0.4444 |
| 0.8978 | 23.0 | 4945 | 1.2936 | 0.4444 |
| 0.8221 | 24.0 | 5160 | 1.2913 | 0.4444 |
| 0.8642 | 25.0 | 5375 | 1.2902 | 0.4444 |
| 0.8893 | 26.0 | 5590 | 1.2888 | 0.4444 |
| 0.8888 | 27.0 | 5805 | 1.2875 | 0.4444 |
| 0.8399 | 28.0 | 6020 | 1.2872 | 0.4444 |
| 0.8384 | 29.0 | 6235 | 1.2862 | 0.4444 |
| 0.8557 | 30.0 | 6450 | 1.2852 | 0.4444 |
| 0.8264 | 31.0 | 6665 | 1.2846 | 0.4444 |
| 0.7947 | 32.0 | 6880 | 1.2839 | 0.4222 |
| 0.7889 | 33.0 | 7095 | 1.2827 | 0.4222 |
| 0.829 | 34.0 | 7310 | 1.2822 | 0.4444 |
| 0.754 | 35.0 | 7525 | 1.2813 | 0.4444 |
| 0.7758 | 36.0 | 7740 | 1.2807 | 0.4444 |
| 0.8928 | 37.0 | 7955 | 1.2794 | 0.4444 |
| 0.734 | 38.0 | 8170 | 1.2794 | 0.4444 |
| 0.7594 | 39.0 | 8385 | 1.2785 | 0.4444 |
| 0.775 | 40.0 | 8600 | 1.2779 | 0.4444 |
| 0.7835 | 41.0 | 8815 | 1.2773 | 0.4667 |
| 0.7569 | 42.0 | 9030 | 1.2769 | 0.4667 |
| 0.7974 | 43.0 | 9245 | 1.2769 | 0.4667 |
| 0.7959 | 44.0 | 9460 | 1.2766 | 0.4667 |
| 0.8113 | 45.0 | 9675 | 1.2762 | 0.4667 |
| 0.7344 | 46.0 | 9890 | 1.2759 | 0.4667 |
| 0.7955 | 47.0 | 10105 | 1.2758 | 0.4667 |
| 0.7831 | 48.0 | 10320 | 1.2757 | 0.4667 |
| 0.7467 | 49.0 | 10535 | 1.2757 | 0.4667 |
| 0.8192 | 50.0 | 10750 | 1.2757 | 0.4667 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.1+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2