swinv2-tiny-patch4-window8-256-finetuned-PE
This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.3083
- Accuracy: 0.8720
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: 0.00025
- train_batch_size: 256
- eval_batch_size: 256
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 1024
- 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 |
|---|---|---|---|---|
| No log | 0.92 | 9 | 0.6391 | 0.6690 |
| 0.6873 | 1.95 | 19 | 0.5293 | 0.7376 |
| 0.6233 | 2.97 | 29 | 0.6385 | 0.6853 |
| 0.5976 | 4.0 | 39 | 0.4447 | 0.7970 |
| 0.5552 | 4.92 | 48 | 0.4029 | 0.8266 |
| 0.552 | 5.95 | 58 | 0.3675 | 0.8429 |
| 0.5055 | 6.97 | 68 | 0.3409 | 0.8581 |
| 0.4816 | 8.0 | 78 | 0.3322 | 0.8615 |
| 0.455 | 8.92 | 87 | 0.3166 | 0.8639 |
| 0.4428 | 9.95 | 97 | 0.3100 | 0.8662 |
| 0.4398 | 10.97 | 107 | 0.3713 | 0.8365 |
| 0.4318 | 12.0 | 117 | 0.4019 | 0.8284 |
| 0.4431 | 12.92 | 126 | 0.3074 | 0.8714 |
| 0.4437 | 13.95 | 136 | 0.3156 | 0.8656 |
| 0.4482 | 14.97 | 146 | 0.3516 | 0.8476 |
| 0.4353 | 16.0 | 156 | 0.3162 | 0.8598 |
| 0.4218 | 16.92 | 165 | 0.3018 | 0.8685 |
| 0.4111 | 17.95 | 175 | 0.3143 | 0.8650 |
| 0.4224 | 18.97 | 185 | 0.3146 | 0.8592 |
| 0.4114 | 20.0 | 195 | 0.3097 | 0.8691 |
| 0.4103 | 20.92 | 204 | 0.3038 | 0.8703 |
| 0.3989 | 21.95 | 214 | 0.2893 | 0.8796 |
| 0.3908 | 22.97 | 224 | 0.2956 | 0.8755 |
| 0.3923 | 24.0 | 234 | 0.3041 | 0.8685 |
| 0.3842 | 24.92 | 243 | 0.2876 | 0.8749 |
| 0.3808 | 25.95 | 253 | 0.2907 | 0.8767 |
| 0.382 | 26.97 | 263 | 0.3018 | 0.8738 |
| 0.3816 | 28.0 | 273 | 0.2812 | 0.8825 |
| 0.379 | 28.92 | 282 | 0.2960 | 0.8633 |
| 0.3858 | 29.95 | 292 | 0.2960 | 0.8743 |
| 0.3546 | 30.97 | 302 | 0.2850 | 0.8807 |
| 0.3656 | 32.0 | 312 | 0.2905 | 0.8784 |
| 0.3707 | 32.92 | 321 | 0.2926 | 0.8743 |
| 0.3651 | 33.95 | 331 | 0.2941 | 0.8796 |
| 0.3584 | 34.97 | 341 | 0.3133 | 0.8615 |
| 0.36 | 36.0 | 351 | 0.3181 | 0.8679 |
| 0.3496 | 36.92 | 360 | 0.3036 | 0.8685 |
| 0.3458 | 37.95 | 370 | 0.2939 | 0.8732 |
| 0.3431 | 38.97 | 380 | 0.3062 | 0.8703 |
| 0.3512 | 40.0 | 390 | 0.2914 | 0.8755 |
| 0.3512 | 40.92 | 399 | 0.3164 | 0.8674 |
| 0.3403 | 41.95 | 409 | 0.3063 | 0.8679 |
| 0.3423 | 42.97 | 419 | 0.3018 | 0.8720 |
| 0.3312 | 44.0 | 429 | 0.3094 | 0.8697 |
| 0.3365 | 44.92 | 438 | 0.3062 | 0.8755 |
| 0.3319 | 45.95 | 448 | 0.3081 | 0.8720 |
| 0.3409 | 46.15 | 450 | 0.3083 | 0.8720 |
Framework versions
- Transformers 4.33.3
- Pytorch 2.0.1+cu117
- Datasets 2.14.5
- Tokenizers 0.13.3