swin-tiny-patch4-window7-224-finetuned-sealv1
This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.4069
- Accuracy: 0.9470
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: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| No log | 0.95 | 9 | 1.1667 | 0.6023 |
| 1.2777 | 2.0 | 19 | 0.9542 | 0.8864 |
| 0.8743 | 2.95 | 28 | 0.5694 | 0.9015 |
| 0.5282 | 4.0 | 38 | 0.3682 | 0.9129 |
| 0.2988 | 4.95 | 47 | 0.2135 | 0.9545 |
| 0.1832 | 6.0 | 57 | 0.2820 | 0.9167 |
| 0.1867 | 6.95 | 66 | 0.1944 | 0.9432 |
| 0.1077 | 8.0 | 76 | 0.2345 | 0.9432 |
| 0.0571 | 8.95 | 85 | 0.2389 | 0.9470 |
| 0.0379 | 10.0 | 95 | 0.2260 | 0.9432 |
| 0.0233 | 10.95 | 104 | 0.2329 | 0.9432 |
| 0.0163 | 12.0 | 114 | 0.2610 | 0.9356 |
| 0.019 | 12.95 | 123 | 0.3660 | 0.9508 |
| 0.0113 | 14.0 | 133 | 0.2777 | 0.9470 |
| 0.0084 | 14.95 | 142 | 0.3123 | 0.9508 |
| 0.008 | 16.0 | 152 | 0.3222 | 0.9470 |
| 0.0048 | 16.95 | 161 | 0.3232 | 0.9470 |
| 0.0075 | 18.0 | 171 | 0.3476 | 0.9508 |
| 0.0048 | 18.95 | 180 | 0.3304 | 0.9470 |
| 0.0143 | 20.0 | 190 | 0.4560 | 0.9432 |
| 0.0143 | 20.95 | 199 | 0.3720 | 0.9432 |
| 0.0019 | 22.0 | 209 | 0.3579 | 0.9394 |
| 0.0063 | 22.95 | 218 | 0.4064 | 0.9432 |
| 0.0023 | 24.0 | 228 | 0.4741 | 0.9394 |
| 0.0015 | 24.95 | 237 | 0.4111 | 0.9470 |
| 0.0022 | 26.0 | 247 | 0.3914 | 0.9432 |
| 0.0008 | 26.95 | 256 | 0.3945 | 0.9432 |
| 0.0024 | 28.0 | 266 | 0.4053 | 0.9470 |
| 0.0026 | 28.42 | 270 | 0.4069 | 0.9470 |
Framework versions
- Transformers 4.38.2
- Pytorch 1.10.2+cu113
- Datasets 2.18.0
- Tokenizers 0.15.2