<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
vit-base-patch16-224-best-finetuned-on-affectnet_short
This model is a fine-tuned version of google/vit-base-patch16-224 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.9712
- Accuracy: 0.6718
- Precision: 0.6698
- Recall: 0.6718
- F1: 0.6703
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: 16
- 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: 22
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|---|---|
| 1.9968 | 1.0 | 32 | 1.9113 | 0.2754 | 0.2518 | 0.2754 | 0.2280 |
| 1.4178 | 2.0 | 64 | 1.2704 | 0.5049 | 0.5149 | 0.5049 | 0.4900 |
| 1.1751 | 3.0 | 96 | 1.1116 | 0.5841 | 0.5891 | 0.5841 | 0.5787 |
| 1.0127 | 4.0 | 128 | 1.0237 | 0.6162 | 0.6335 | 0.6162 | 0.6141 |
| 0.9969 | 5.0 | 160 | 0.9890 | 0.6259 | 0.6294 | 0.6259 | 0.6150 |
| 0.9376 | 6.0 | 192 | 0.9768 | 0.6190 | 0.6335 | 0.6190 | 0.6183 |
| 0.8299 | 7.0 | 224 | 0.9579 | 0.6357 | 0.6339 | 0.6357 | 0.6282 |
| 0.7645 | 8.0 | 256 | 0.9366 | 0.6489 | 0.6559 | 0.6489 | 0.6474 |
| 0.7944 | 9.0 | 288 | 0.9303 | 0.6443 | 0.6494 | 0.6443 | 0.6447 |
| 0.7334 | 10.0 | 320 | 0.9510 | 0.6546 | 0.6634 | 0.6546 | 0.6523 |
| 0.6596 | 11.0 | 352 | 0.9369 | 0.6449 | 0.6528 | 0.6449 | 0.6428 |
| 0.6781 | 12.0 | 384 | 0.9717 | 0.6368 | 0.6513 | 0.6368 | 0.6360 |
| 0.5688 | 13.0 | 416 | 0.9509 | 0.6540 | 0.6531 | 0.6540 | 0.6495 |
| 0.5766 | 14.0 | 448 | 0.9485 | 0.6615 | 0.6655 | 0.6615 | 0.6601 |
| 0.5529 | 15.0 | 480 | 0.9590 | 0.6569 | 0.6561 | 0.6569 | 0.6538 |
| 0.4998 | 16.0 | 512 | 0.9677 | 0.6512 | 0.6514 | 0.6512 | 0.6488 |
| 0.4908 | 17.0 | 544 | 0.9670 | 0.6638 | 0.6645 | 0.6638 | 0.6616 |
| 0.4682 | 18.0 | 576 | 0.9635 | 0.6678 | 0.6707 | 0.6678 | 0.6684 |
| 0.4761 | 19.0 | 608 | 0.9680 | 0.6667 | 0.6674 | 0.6667 | 0.6658 |
| 0.4161 | 20.0 | 640 | 0.9701 | 0.6713 | 0.6719 | 0.6713 | 0.6701 |
| 0.4295 | 21.0 | 672 | 0.9712 | 0.6718 | 0.6698 | 0.6718 | 0.6703 |
| 0.434 | 22.0 | 704 | 0.9755 | 0.6707 | 0.6705 | 0.6707 | 0.6690 |
Framework versions
- Transformers 4.29.0
- Pytorch 2.0.0+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3
