vit-base_rvl_cdip-N1K_aAURC_2
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: 0.3354
- Accuracy: 0.8788
- Brier Loss: 0.2243
- Nll: 1.0945
- F1 Micro: 0.8788
- F1 Macro: 0.8793
- Ece: 0.1094
- Aurc: 0.0303
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: 2
- eval_batch_size: 2
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|
| 0.1996 | 1.0 | 8000 | 0.2395 | 0.8413 | 0.2593 | 1.3668 | 0.8413 | 0.8446 | 0.1016 | 0.0414 |
| 0.1357 | 2.0 | 16000 | 0.2446 | 0.8545 | 0.2435 | 1.2587 | 0.8545 | 0.8550 | 0.1066 | 0.0364 |
| 0.1083 | 3.0 | 24000 | 0.2702 | 0.8515 | 0.2584 | 1.2968 | 0.8515 | 0.8528 | 0.1126 | 0.0379 |
| 0.0578 | 4.0 | 32000 | 0.3397 | 0.8327 | 0.2931 | 1.3827 | 0.8327 | 0.8322 | 0.1391 | 0.0494 |
| 0.0294 | 5.0 | 40000 | 0.3407 | 0.8538 | 0.2662 | 1.3685 | 0.8537 | 0.8536 | 0.1280 | 0.0378 |
| 0.0099 | 6.0 | 48000 | 0.3489 | 0.8585 | 0.2602 | 1.2438 | 0.8585 | 0.8599 | 0.1254 | 0.0350 |
| 0.0058 | 7.0 | 56000 | 0.3433 | 0.868 | 0.2418 | 1.2328 | 0.868 | 0.8666 | 0.1172 | 0.0362 |
| 0.0026 | 8.0 | 64000 | 0.3414 | 0.87 | 0.2401 | 1.1910 | 0.87 | 0.8704 | 0.1162 | 0.0304 |
| 0.0026 | 9.0 | 72000 | 0.3460 | 0.8728 | 0.2357 | 1.1567 | 0.8728 | 0.8729 | 0.1143 | 0.0308 |
| 0.0024 | 10.0 | 80000 | 0.3354 | 0.8788 | 0.2243 | 1.0945 | 0.8788 | 0.8793 | 0.1094 | 0.0303 |
Framework versions
- Transformers 4.33.3
- Pytorch 2.2.0.dev20231002
- Datasets 2.7.1
- Tokenizers 0.13.3