AnimeCharacterClassifierMark1
This model is a fine-tuned version of google/vit-base-patch16-224 on the image_folder dataset. It achieves the following results on the evaluation set:
- Loss: 0.6720
- Accuracy: 0.8655
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: 128
- eval_batch_size: 128
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 512
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 42
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 5.0145 | 0.99 | 17 | 4.9303 | 0.0092 |
| 4.8416 | 1.97 | 34 | 4.7487 | 0.0287 |
| 4.4383 | 2.96 | 51 | 4.3597 | 0.1170 |
| 4.0762 | 4.0 | 69 | 3.6419 | 0.3224 |
| 3.108 | 4.99 | 86 | 2.8574 | 0.5246 |
| 2.1571 | 5.97 | 103 | 2.2129 | 0.6653 |
| 1.4685 | 6.96 | 120 | 1.7290 | 0.7495 |
| 1.1649 | 8.0 | 138 | 1.3862 | 0.7977 |
| 0.7905 | 8.99 | 155 | 1.1589 | 0.8214 |
| 0.5549 | 9.97 | 172 | 1.0263 | 0.8296 |
| 0.4577 | 10.96 | 189 | 0.8994 | 0.8368 |
| 0.2964 | 12.0 | 207 | 0.8086 | 0.8552 |
| 0.194 | 12.99 | 224 | 0.7446 | 0.8583 |
| 0.1358 | 13.97 | 241 | 0.7064 | 0.8573 |
| 0.1116 | 14.96 | 258 | 0.6720 | 0.8655 |
| 0.0811 | 16.0 | 276 | 0.6515 | 0.8645 |
Framework versions
- Transformers 4.33.0
- Pytorch 2.0.0
- Datasets 2.1.0
- Tokenizers 0.13.3