<!-- 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. -->
mobilebert_sa_GLUE_Experiment_logit_kd_data_aug_mrpc_128
This model is a fine-tuned version of google/mobilebert-uncased on the GLUE MRPC dataset. It achieves the following results on the evaluation set:
- Loss: 0.1262
- Accuracy: 0.9877
- F1: 0.9911
- Combined Score: 0.9894
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: 128
- eval_batch_size: 128
- seed: 10
- distributed_type: multi-GPU
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Combined Score |
|---|---|---|---|---|---|---|
| 0.3021 | 1.0 | 1959 | 0.2407 | 0.9510 | 0.9636 | 0.9573 |
| 0.2326 | 2.0 | 3918 | 0.2009 | 0.9779 | 0.9841 | 0.9810 |
| 0.224 | 3.0 | 5877 | 0.1790 | 0.9730 | 0.9807 | 0.9769 |
| 0.219 | 4.0 | 7836 | 0.1789 | 0.9804 | 0.9859 | 0.9831 |
| 0.2153 | 5.0 | 9795 | 0.1804 | 0.9804 | 0.9859 | 0.9831 |
| 0.2121 | 6.0 | 11754 | 0.1754 | 0.9755 | 0.9824 | 0.9789 |
| 0.2088 | 7.0 | 13713 | 0.1661 | 0.9804 | 0.9859 | 0.9831 |
| 0.2056 | 8.0 | 15672 | 0.1654 | 0.9779 | 0.9841 | 0.9810 |
| 0.2031 | 9.0 | 17631 | 0.1714 | 0.9828 | 0.9876 | 0.9852 |
| 0.2012 | 10.0 | 19590 | 0.1617 | 0.9828 | 0.9876 | 0.9852 |
| 0.1993 | 11.0 | 21549 | 0.1610 | 0.9828 | 0.9876 | 0.9852 |
| 0.1978 | 12.0 | 23508 | 0.1507 | 0.9853 | 0.9894 | 0.9873 |
| 0.1964 | 13.0 | 25467 | 0.1496 | 0.9902 | 0.9929 | 0.9915 |
| 0.1953 | 14.0 | 27426 | 0.1569 | 0.9828 | 0.9876 | 0.9852 |
| 0.1943 | 15.0 | 29385 | 0.1524 | 0.9877 | 0.9911 | 0.9894 |
| 0.1934 | 16.0 | 31344 | 0.1492 | 0.9877 | 0.9911 | 0.9894 |
| 0.1926 | 17.0 | 33303 | 0.1480 | 0.9902 | 0.9929 | 0.9915 |
| 0.1918 | 18.0 | 35262 | 0.1416 | 0.9828 | 0.9876 | 0.9852 |
| 0.1912 | 19.0 | 37221 | 0.1420 | 0.9853 | 0.9894 | 0.9873 |
| 0.1905 | 20.0 | 39180 | 0.1396 | 0.9853 | 0.9894 | 0.9873 |
| 0.1899 | 21.0 | 41139 | 0.1458 | 0.9853 | 0.9894 | 0.9873 |
| 0.1893 | 22.0 | 43098 | 0.1484 | 0.9877 | 0.9911 | 0.9894 |
| 0.1888 | 23.0 | 45057 | 0.1407 | 0.9877 | 0.9911 | 0.9894 |
| 0.1883 | 24.0 | 47016 | 0.1372 | 0.9804 | 0.9858 | 0.9831 |
| 0.1878 | 25.0 | 48975 | 0.1354 | 0.9877 | 0.9911 | 0.9894 |
| 0.1873 | 26.0 | 50934 | 0.1368 | 0.9902 | 0.9929 | 0.9915 |
| 0.1869 | 27.0 | 52893 | 0.1378 | 0.9926 | 0.9947 | 0.9936 |
| 0.1865 | 28.0 | 54852 | 0.1381 | 0.9853 | 0.9894 | 0.9873 |
| 0.1861 | 29.0 | 56811 | 0.1329 | 0.9877 | 0.9911 | 0.9894 |
| 0.1857 | 30.0 | 58770 | 0.1342 | 0.9902 | 0.9929 | 0.9915 |
| 0.1854 | 31.0 | 60729 | 0.1346 | 0.9902 | 0.9929 | 0.9915 |
| 0.1849 | 32.0 | 62688 | 0.1323 | 0.9902 | 0.9929 | 0.9915 |
| 0.1846 | 33.0 | 64647 | 0.1317 | 0.9877 | 0.9911 | 0.9894 |
| 0.1843 | 34.0 | 66606 | 0.1318 | 0.9877 | 0.9911 | 0.9894 |
| 0.1839 | 35.0 | 68565 | 0.1311 | 0.9926 | 0.9947 | 0.9936 |
| 0.1837 | 36.0 | 70524 | 0.1291 | 0.9877 | 0.9911 | 0.9894 |
| 0.1834 | 37.0 | 72483 | 0.1313 | 0.9853 | 0.9893 | 0.9873 |
| 0.1831 | 38.0 | 74442 | 0.1262 | 0.9877 | 0.9911 | 0.9894 |
| 0.1828 | 39.0 | 76401 | 0.1288 | 0.9877 | 0.9911 | 0.9894 |
| 0.1825 | 40.0 | 78360 | 0.1295 | 0.9926 | 0.9947 | 0.9936 |
| 0.1823 | 41.0 | 80319 | 0.1277 | 0.9902 | 0.9929 | 0.9915 |
| 0.182 | 42.0 | 82278 | 0.1265 | 0.9902 | 0.9929 | 0.9915 |
| 0.1818 | 43.0 | 84237 | 0.1273 | 0.9902 | 0.9929 | 0.9915 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.9.0
- Tokenizers 0.13.2
