vit-base_rvl-cdip-tiny_rvl_cdip-NK1000_og_simkd
This model is a fine-tuned version of WinKawaks/vit-tiny-patch16-224 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 268.5710
- Accuracy: 0.832
- Brier Loss: 0.3051
- Nll: 1.8984
- F1 Micro: 0.832
- F1 Macro: 0.8328
- Ece: 0.1458
- Aurc: 0.0566
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: 0.0001
- train_batch_size: 32
- eval_batch_size: 32
- 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: 50
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Brier Loss | Nll | F1 Micro | F1 Macro | Ece | Aurc |
|---|---|---|---|---|---|---|---|---|---|---|
| 286.5314 | 1.0 | 500 | 283.8548 | 0.4555 | 0.8016 | 2.8438 | 0.4555 | 0.3694 | 0.3223 | 0.2744 |
| 283.1635 | 2.0 | 1000 | 282.0722 | 0.5577 | 0.6555 | 2.5676 | 0.5577 | 0.5116 | 0.2428 | 0.1873 |
| 282.0305 | 3.0 | 1500 | 281.5496 | 0.6355 | 0.5598 | 2.4665 | 0.6355 | 0.6232 | 0.2150 | 0.1424 |
| 281.2235 | 4.0 | 2000 | 280.5680 | 0.7065 | 0.4447 | 2.2146 | 0.7065 | 0.7028 | 0.1680 | 0.1010 |
| 280.4245 | 5.0 | 2500 | 279.8773 | 0.7245 | 0.4675 | 2.2237 | 0.7245 | 0.7253 | 0.1993 | 0.1058 |
| 279.6686 | 6.0 | 3000 | 279.1086 | 0.7748 | 0.3740 | 2.0556 | 0.7748 | 0.7729 | 0.1601 | 0.0700 |
| 278.8635 | 7.0 | 3500 | 278.2839 | 0.7675 | 0.3894 | 1.9970 | 0.7675 | 0.7689 | 0.1713 | 0.0791 |
| 278.1131 | 8.0 | 4000 | 277.6409 | 0.7977 | 0.3416 | 1.9272 | 0.7977 | 0.7969 | 0.1478 | 0.0618 |
| 277.4522 | 9.0 | 4500 | 277.2180 | 0.8083 | 0.3280 | 1.9560 | 0.8083 | 0.8122 | 0.1474 | 0.0559 |
| 276.8156 | 10.0 | 5000 | 276.5454 | 0.8145 | 0.3158 | 1.8932 | 0.8145 | 0.8149 | 0.1404 | 0.0531 |
| 276.2433 | 11.0 | 5500 | 275.8951 | 0.8117 | 0.3233 | 1.8813 | 0.8117 | 0.8122 | 0.1460 | 0.0551 |
| 275.7038 | 12.0 | 6000 | 275.6540 | 0.8217 | 0.3033 | 1.8605 | 0.8217 | 0.8228 | 0.1355 | 0.0521 |
| 275.1847 | 13.0 | 6500 | 275.1825 | 0.8263 | 0.3063 | 1.8834 | 0.8263 | 0.8282 | 0.1384 | 0.0513 |
| 274.7168 | 14.0 | 7000 | 274.8503 | 0.8203 | 0.3133 | 1.8742 | 0.8203 | 0.8222 | 0.1448 | 0.0520 |
| 274.2753 | 15.0 | 7500 | 274.2773 | 0.8273 | 0.3015 | 1.8833 | 0.8273 | 0.8282 | 0.1392 | 0.0497 |
| 273.8617 | 16.0 | 8000 | 273.9056 | 0.825 | 0.3018 | 1.8546 | 0.825 | 0.8273 | 0.1391 | 0.0527 |
| 273.4509 | 17.0 | 8500 | 273.4976 | 0.827 | 0.3028 | 1.8656 | 0.827 | 0.8270 | 0.1415 | 0.0500 |
| 273.0643 | 18.0 | 9000 | 273.0985 | 0.8315 | 0.2977 | 1.8671 | 0.8315 | 0.8326 | 0.1382 | 0.0491 |
| 272.7104 | 19.0 | 9500 | 272.9490 | 0.8273 | 0.3035 | 1.8686 | 0.8273 | 0.8285 | 0.1427 | 0.0525 |
| 272.3669 | 20.0 | 10000 | 272.6702 | 0.8253 | 0.3052 | 1.8809 | 0.8253 | 0.8258 | 0.1441 | 0.0499 |
| 272.0331 | 21.0 | 10500 | 272.2651 | 0.833 | 0.2966 | 1.8759 | 0.833 | 0.8340 | 0.1397 | 0.0479 |
| 271.7213 | 22.0 | 11000 | 272.2740 | 0.829 | 0.2999 | 1.8507 | 0.8290 | 0.8295 | 0.1419 | 0.0493 |
| 271.4253 | 23.0 | 11500 | 271.7973 | 0.8327 | 0.2962 | 1.8837 | 0.8327 | 0.8326 | 0.1404 | 0.0490 |
| 271.1327 | 24.0 | 12000 | 271.5110 | 0.8355 | 0.2930 | 1.8580 | 0.8355 | 0.8365 | 0.1393 | 0.0502 |
| 270.858 | 25.0 | 12500 | 271.1653 | 0.828 | 0.3035 | 1.8529 | 0.828 | 0.8287 | 0.1455 | 0.0520 |
| 270.5978 | 26.0 | 13000 | 270.9584 | 0.8283 | 0.3056 | 1.8615 | 0.8283 | 0.8282 | 0.1442 | 0.0535 |
| 270.3636 | 27.0 | 13500 | 270.7707 | 0.832 | 0.3030 | 1.8635 | 0.832 | 0.8331 | 0.1442 | 0.0507 |
| 270.1365 | 28.0 | 14000 | 270.3265 | 0.8287 | 0.3096 | 1.8857 | 0.8287 | 0.8295 | 0.1458 | 0.0551 |
| 269.9005 | 29.0 | 14500 | 270.4089 | 0.8307 | 0.3017 | 1.8722 | 0.8308 | 0.8306 | 0.1441 | 0.0528 |
| 269.6876 | 30.0 | 15000 | 270.2905 | 0.8303 | 0.3043 | 1.8792 | 0.8303 | 0.8308 | 0.1446 | 0.0518 |
| 269.5015 | 31.0 | 15500 | 269.9496 | 0.834 | 0.2997 | 1.8925 | 0.834 | 0.8344 | 0.1413 | 0.0519 |
| 269.3106 | 32.0 | 16000 | 269.8872 | 0.8333 | 0.2990 | 1.8822 | 0.8333 | 0.8333 | 0.1419 | 0.0524 |
| 269.1204 | 33.0 | 16500 | 269.7998 | 0.8303 | 0.3057 | 1.9016 | 0.8303 | 0.8310 | 0.1463 | 0.0541 |
| 268.9658 | 34.0 | 17000 | 269.3946 | 0.8347 | 0.3003 | 1.8922 | 0.8347 | 0.8356 | 0.1423 | 0.0535 |
| 268.8073 | 35.0 | 17500 | 269.5928 | 0.8327 | 0.3035 | 1.8508 | 0.8327 | 0.8332 | 0.1443 | 0.0546 |
| 268.6654 | 36.0 | 18000 | 269.2020 | 0.8307 | 0.3058 | 1.8891 | 0.8308 | 0.8317 | 0.1456 | 0.0543 |
| 268.5213 | 37.0 | 18500 | 269.3784 | 0.8295 | 0.3095 | 1.8732 | 0.8295 | 0.8299 | 0.1478 | 0.0549 |
| 268.3883 | 38.0 | 19000 | 269.0580 | 0.8303 | 0.3060 | 1.8621 | 0.8303 | 0.8303 | 0.1466 | 0.0559 |
| 268.2752 | 39.0 | 19500 | 269.0785 | 0.8317 | 0.3038 | 1.8956 | 0.8317 | 0.8320 | 0.1449 | 0.0534 |
| 268.1814 | 40.0 | 20000 | 268.8612 | 0.8357 | 0.3029 | 1.9057 | 0.8357 | 0.8367 | 0.1416 | 0.0557 |
| 268.0695 | 41.0 | 20500 | 268.8330 | 0.8303 | 0.3047 | 1.8963 | 0.8303 | 0.8309 | 0.1475 | 0.0565 |
| 267.9566 | 42.0 | 21000 | 268.7392 | 0.8313 | 0.3055 | 1.9059 | 0.8313 | 0.8319 | 0.1462 | 0.0550 |
| 267.9045 | 43.0 | 21500 | 268.6012 | 0.8307 | 0.3063 | 1.8974 | 0.8308 | 0.8317 | 0.1478 | 0.0565 |
| 267.8455 | 44.0 | 22000 | 268.8788 | 0.832 | 0.3042 | 1.8960 | 0.832 | 0.8325 | 0.1447 | 0.0553 |
| 267.776 | 45.0 | 22500 | 268.5588 | 0.829 | 0.3093 | 1.9087 | 0.8290 | 0.8296 | 0.1489 | 0.0556 |
| 267.7253 | 46.0 | 23000 | 268.4604 | 0.8303 | 0.3079 | 1.9259 | 0.8303 | 0.8307 | 0.1474 | 0.0573 |
| 267.6786 | 47.0 | 23500 | 268.5825 | 0.8317 | 0.3062 | 1.9043 | 0.8317 | 0.8322 | 0.1464 | 0.0561 |
| 267.6514 | 48.0 | 24000 | 268.4191 | 0.8315 | 0.3072 | 1.8933 | 0.8315 | 0.8320 | 0.1466 | 0.0576 |
| 267.6384 | 49.0 | 24500 | 268.4131 | 0.8303 | 0.3087 | 1.9256 | 0.8303 | 0.8306 | 0.1483 | 0.0576 |
| 267.6158 | 50.0 | 25000 | 268.5710 | 0.832 | 0.3051 | 1.8984 | 0.832 | 0.8328 | 0.1458 | 0.0566 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2