<!-- 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-small_rvl_cdip-NK1000_kd_NKD_t1.0_g1.5
This model is a fine-tuned version of WinKawaks/vit-small-patch16-224 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 3.6996
- Accuracy: 0.856
- Brier Loss: 0.2370
- Nll: 1.1806
- F1 Micro: 0.856
- F1 Macro: 0.8578
- Ece: 0.1015
- Aurc: 0.0302
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: 96
- eval_batch_size: 96
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|
| No log | 1.0 | 167 | 4.6372 | 0.632 | 0.5161 | 2.2084 | 0.632 | 0.6245 | 0.1480 | 0.1490 |
| No log | 2.0 | 334 | 4.2247 | 0.7143 | 0.3976 | 1.8912 | 0.7142 | 0.7110 | 0.0878 | 0.0942 |
| 4.8599 | 3.0 | 501 | 4.0290 | 0.7488 | 0.3551 | 1.7330 | 0.7488 | 0.7552 | 0.0711 | 0.0785 |
| 4.8599 | 4.0 | 668 | 3.8716 | 0.7903 | 0.2981 | 1.6409 | 0.7903 | 0.7898 | 0.0468 | 0.0593 |
| 4.8599 | 5.0 | 835 | 3.7535 | 0.8055 | 0.2829 | 1.5302 | 0.8055 | 0.8039 | 0.0465 | 0.0521 |
| 3.7258 | 6.0 | 1002 | 3.7365 | 0.8023 | 0.2787 | 1.5134 | 0.8023 | 0.8043 | 0.0352 | 0.0509 |
| 3.7258 | 7.0 | 1169 | 3.7092 | 0.811 | 0.2705 | 1.3930 | 0.811 | 0.8130 | 0.0472 | 0.0488 |
| 3.7258 | 8.0 | 1336 | 3.6799 | 0.8213 | 0.2643 | 1.4444 | 0.8213 | 0.8242 | 0.0484 | 0.0453 |
| 3.4329 | 9.0 | 1503 | 3.6148 | 0.8265 | 0.2522 | 1.3355 | 0.8265 | 0.8295 | 0.0522 | 0.0425 |
| 3.4329 | 10.0 | 1670 | 3.5723 | 0.826 | 0.2524 | 1.3332 | 0.826 | 0.8286 | 0.0637 | 0.0398 |
| 3.4329 | 11.0 | 1837 | 3.6298 | 0.8277 | 0.2565 | 1.3664 | 0.8277 | 0.8304 | 0.0720 | 0.0422 |
| 3.2987 | 12.0 | 2004 | 3.5604 | 0.8407 | 0.2376 | 1.3420 | 0.8407 | 0.8424 | 0.0609 | 0.0359 |
| 3.2987 | 13.0 | 2171 | 3.5885 | 0.8393 | 0.2446 | 1.3552 | 0.8393 | 0.8420 | 0.0712 | 0.0381 |
| 3.2987 | 14.0 | 2338 | 3.6191 | 0.8315 | 0.2518 | 1.3329 | 0.8315 | 0.8322 | 0.0772 | 0.0383 |
| 3.2268 | 15.0 | 2505 | 3.5920 | 0.837 | 0.2465 | 1.3397 | 0.8370 | 0.8399 | 0.0816 | 0.0372 |
| 3.2268 | 16.0 | 2672 | 3.5483 | 0.847 | 0.2324 | 1.2733 | 0.847 | 0.8489 | 0.0697 | 0.0337 |
| 3.2268 | 17.0 | 2839 | 3.5924 | 0.8438 | 0.2444 | 1.2686 | 0.8438 | 0.8450 | 0.0838 | 0.0355 |
| 3.174 | 18.0 | 3006 | 3.5909 | 0.8427 | 0.2419 | 1.2631 | 0.8427 | 0.8449 | 0.0856 | 0.0336 |
| 3.174 | 19.0 | 3173 | 3.5857 | 0.8452 | 0.2393 | 1.2979 | 0.8452 | 0.8474 | 0.0804 | 0.0338 |
| 3.174 | 20.0 | 3340 | 3.5700 | 0.8455 | 0.2373 | 1.2916 | 0.8455 | 0.8471 | 0.0824 | 0.0336 |
| 3.1369 | 21.0 | 3507 | 3.5578 | 0.8518 | 0.2298 | 1.2615 | 0.8518 | 0.8531 | 0.0779 | 0.0316 |
| 3.1369 | 22.0 | 3674 | 3.5659 | 0.8478 | 0.2349 | 1.2532 | 0.8478 | 0.8502 | 0.0848 | 0.0325 |
| 3.1369 | 23.0 | 3841 | 3.5506 | 0.8552 | 0.2302 | 1.2530 | 0.8552 | 0.8572 | 0.0817 | 0.0312 |
| 3.1077 | 24.0 | 4008 | 3.5551 | 0.857 | 0.2298 | 1.2669 | 0.857 | 0.8585 | 0.0817 | 0.0306 |
| 3.1077 | 25.0 | 4175 | 3.5563 | 0.8575 | 0.2259 | 1.2374 | 0.8575 | 0.8587 | 0.0820 | 0.0296 |
| 3.1077 | 26.0 | 4342 | 3.5642 | 0.8555 | 0.2312 | 1.2159 | 0.8555 | 0.8577 | 0.0855 | 0.0305 |
| 3.0885 | 27.0 | 4509 | 3.5739 | 0.856 | 0.2332 | 1.2143 | 0.856 | 0.8581 | 0.0854 | 0.0309 |
| 3.0885 | 28.0 | 4676 | 3.5544 | 0.855 | 0.2294 | 1.2305 | 0.855 | 0.8567 | 0.0860 | 0.0302 |
| 3.0885 | 29.0 | 4843 | 3.5574 | 0.8598 | 0.2262 | 1.2330 | 0.8598 | 0.8616 | 0.0839 | 0.0304 |
| 3.0716 | 30.0 | 5010 | 3.5673 | 0.8572 | 0.2291 | 1.2208 | 0.8572 | 0.8591 | 0.0888 | 0.0298 |
| 3.0716 | 31.0 | 5177 | 3.5818 | 0.853 | 0.2293 | 1.1947 | 0.853 | 0.8550 | 0.0917 | 0.0302 |
| 3.0716 | 32.0 | 5344 | 3.5792 | 0.858 | 0.2295 | 1.2086 | 0.858 | 0.8597 | 0.0881 | 0.0297 |
| 3.064 | 33.0 | 5511 | 3.5895 | 0.8575 | 0.2315 | 1.2009 | 0.8575 | 0.8590 | 0.0900 | 0.0296 |
| 3.064 | 34.0 | 5678 | 3.5923 | 0.8565 | 0.2293 | 1.1905 | 0.8565 | 0.8583 | 0.0901 | 0.0295 |
| 3.064 | 35.0 | 5845 | 3.5997 | 0.8562 | 0.2310 | 1.2128 | 0.8562 | 0.8577 | 0.0922 | 0.0297 |
| 3.0572 | 36.0 | 6012 | 3.6041 | 0.8572 | 0.2307 | 1.1932 | 0.8572 | 0.8589 | 0.0917 | 0.0296 |
| 3.0572 | 37.0 | 6179 | 3.6123 | 0.857 | 0.2319 | 1.1984 | 0.857 | 0.8587 | 0.0932 | 0.0294 |
| 3.0572 | 38.0 | 6346 | 3.6162 | 0.8585 | 0.2304 | 1.1909 | 0.8585 | 0.8600 | 0.0917 | 0.0293 |
| 3.0542 | 39.0 | 6513 | 3.6318 | 0.8575 | 0.2321 | 1.1982 | 0.8575 | 0.8590 | 0.0945 | 0.0298 |
| 3.0542 | 40.0 | 6680 | 3.6319 | 0.8572 | 0.2324 | 1.1905 | 0.8572 | 0.8588 | 0.0944 | 0.0296 |
| 3.0542 | 41.0 | 6847 | 3.6401 | 0.856 | 0.2327 | 1.1953 | 0.856 | 0.8577 | 0.0964 | 0.0295 |
| 3.0527 | 42.0 | 7014 | 3.6567 | 0.8572 | 0.2343 | 1.1821 | 0.8572 | 0.8588 | 0.0968 | 0.0299 |
| 3.0527 | 43.0 | 7181 | 3.6601 | 0.8562 | 0.2341 | 1.1885 | 0.8562 | 0.8580 | 0.0981 | 0.0299 |
| 3.0527 | 44.0 | 7348 | 3.6683 | 0.8572 | 0.2351 | 1.1854 | 0.8572 | 0.8588 | 0.0977 | 0.0299 |
| 3.0479 | 45.0 | 7515 | 3.6742 | 0.8568 | 0.2353 | 1.1894 | 0.8568 | 0.8584 | 0.0986 | 0.0299 |
| 3.0479 | 46.0 | 7682 | 3.6847 | 0.8565 | 0.2360 | 1.1813 | 0.8565 | 0.8582 | 0.1002 | 0.0301 |
| 3.0479 | 47.0 | 7849 | 3.6891 | 0.8562 | 0.2363 | 1.1814 | 0.8562 | 0.8581 | 0.1004 | 0.0302 |
| 3.0469 | 48.0 | 8016 | 3.6964 | 0.8558 | 0.2367 | 1.1806 | 0.8558 | 0.8575 | 0.1014 | 0.0302 |
| 3.0469 | 49.0 | 8183 | 3.6982 | 0.8565 | 0.2369 | 1.1808 | 0.8565 | 0.8583 | 0.1008 | 0.0302 |
| 3.0469 | 50.0 | 8350 | 3.6996 | 0.856 | 0.2370 | 1.1806 | 0.856 | 0.8578 | 0.1015 | 0.0302 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2
