resnet-50-finetuned-omars3
This model is a fine-tuned version of microsoft/resnet-50 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.7345
- Accuracy: 0.7436
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: 0.0005
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- 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 |
|---|---|---|---|---|
| 1.3928 | 1.0 | 11 | 1.3871 | 0.2308 |
| 1.3864 | 2.0 | 22 | 1.3820 | 0.3077 |
| 1.3694 | 3.0 | 33 | 1.3510 | 0.5385 |
| 1.3513 | 4.0 | 44 | 1.2942 | 0.4872 |
| 1.3067 | 5.0 | 55 | 1.1984 | 0.6154 |
| 1.2184 | 6.0 | 66 | 0.9974 | 0.6923 |
| 1.0967 | 7.0 | 77 | 0.7869 | 0.6667 |
| 0.9731 | 8.0 | 88 | 0.7923 | 0.7436 |
| 0.9506 | 9.0 | 99 | 0.7161 | 0.6667 |
| 0.7783 | 10.0 | 110 | 0.6736 | 0.6923 |
| 0.7072 | 11.0 | 121 | 0.6693 | 0.7436 |
| 0.6669 | 12.0 | 132 | 0.7203 | 0.6923 |
| 0.6579 | 13.0 | 143 | 0.6195 | 0.7949 |
| 0.6695 | 14.0 | 154 | 0.6395 | 0.7692 |
| 0.678 | 15.0 | 165 | 0.6870 | 0.7692 |
| 0.5919 | 16.0 | 176 | 0.6681 | 0.7692 |
| 0.5459 | 17.0 | 187 | 0.6895 | 0.7692 |
| 0.5635 | 18.0 | 198 | 0.6617 | 0.7692 |
| 0.5378 | 19.0 | 209 | 0.6401 | 0.7949 |
| 0.5105 | 20.0 | 220 | 0.7108 | 0.7692 |
| 0.4656 | 21.0 | 231 | 0.7267 | 0.7692 |
| 0.5338 | 22.0 | 242 | 0.7531 | 0.7436 |
| 0.4846 | 23.0 | 253 | 0.7103 | 0.7179 |
| 0.4212 | 24.0 | 264 | 0.7809 | 0.7436 |
| 0.4677 | 25.0 | 275 | 0.7825 | 0.7692 |
| 0.4496 | 26.0 | 286 | 0.8240 | 0.6923 |
| 0.3784 | 27.0 | 297 | 0.7563 | 0.7179 |
| 0.4949 | 28.0 | 308 | 0.6823 | 0.7692 |
| 0.4612 | 29.0 | 319 | 0.7542 | 0.6667 |
| 0.4491 | 30.0 | 330 | 0.7345 | 0.7436 |
Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.0
- Tokenizers 0.13.3