<!-- 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_rvl_cdip-N1K_ce_64
This model is a fine-tuned version of jordyvl/vit-base_rvl-cdip on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5145
- Accuracy: 0.8908
- Brier Loss: 0.1847
- Nll: 0.9466
- F1 Micro: 0.8907
- F1 Macro: 0.8910
- Ece: 0.0829
- Aurc: 0.0191
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: 2e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Brier Loss | Nll | F1 Micro | F1 Macro | Ece | Aurc |
|---|---|---|---|---|---|---|---|---|---|---|
| No log | 1.0 | 250 | 0.4009 | 0.8892 | 0.1695 | 1.1791 | 0.8892 | 0.8896 | 0.0538 | 0.0185 |
| 0.1472 | 2.0 | 500 | 0.4214 | 0.8938 | 0.1688 | 1.1365 | 0.8938 | 0.8948 | 0.0527 | 0.0199 |
| 0.1472 | 3.0 | 750 | 0.4245 | 0.8898 | 0.1722 | 1.0919 | 0.8898 | 0.8900 | 0.0633 | 0.0185 |
| 0.0462 | 4.0 | 1000 | 0.4571 | 0.891 | 0.1776 | 1.0386 | 0.891 | 0.8914 | 0.0699 | 0.0198 |
| 0.0462 | 5.0 | 1250 | 0.4775 | 0.8922 | 0.1797 | 1.0236 | 0.8922 | 0.8926 | 0.0745 | 0.0196 |
| 0.0118 | 6.0 | 1500 | 0.4953 | 0.8878 | 0.1845 | 0.9920 | 0.8878 | 0.8882 | 0.0823 | 0.0190 |
| 0.0118 | 7.0 | 1750 | 0.5052 | 0.89 | 0.1847 | 0.9631 | 0.89 | 0.8903 | 0.0820 | 0.0193 |
| 0.0065 | 8.0 | 2000 | 0.5068 | 0.8905 | 0.1832 | 0.9653 | 0.8905 | 0.8910 | 0.0816 | 0.0190 |
| 0.0065 | 9.0 | 2250 | 0.5143 | 0.8905 | 0.1850 | 0.9551 | 0.8905 | 0.8908 | 0.0833 | 0.0191 |
| 0.0053 | 10.0 | 2500 | 0.5145 | 0.8908 | 0.1847 | 0.9466 | 0.8907 | 0.8910 | 0.0829 | 0.0191 |
Framework versions
- Transformers 4.33.3
- Pytorch 2.2.0.dev20231002
- Datasets 2.7.1
- Tokenizers 0.13.3
