<!-- 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. -->
2_5e-3_5_0.5
This model is a fine-tuned version of bert-large-uncased on the super_glue dataset. It achieves the following results on the evaluation set:
- Loss: 1.0090
- Accuracy: 0.6991
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: 0.005
- train_batch_size: 16
- eval_batch_size: 8
- seed: 11
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 60.0
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy |
---|---|---|---|---|
2.0566 | 1.0 | 590 | 1.9336 | 0.6208 |
1.8329 | 2.0 | 1180 | 1.8941 | 0.6226 |
1.8027 | 3.0 | 1770 | 1.6503 | 0.6043 |
1.7269 | 4.0 | 2360 | 1.7276 | 0.5180 |
1.7224 | 5.0 | 2950 | 1.7866 | 0.6223 |
1.6611 | 6.0 | 3540 | 1.6363 | 0.5988 |
1.6862 | 7.0 | 4130 | 1.7201 | 0.5593 |
1.5648 | 8.0 | 4720 | 1.7083 | 0.6339 |
1.5735 | 9.0 | 5310 | 1.5898 | 0.5991 |
1.5494 | 10.0 | 5900 | 1.6325 | 0.6385 |
1.5284 | 11.0 | 6490 | 1.6925 | 0.6303 |
1.478 | 12.0 | 7080 | 1.7338 | 0.5355 |
1.5236 | 13.0 | 7670 | 1.5156 | 0.6394 |
1.46 | 14.0 | 8260 | 1.8612 | 0.6321 |
1.4214 | 15.0 | 8850 | 1.4616 | 0.6471 |
1.4158 | 16.0 | 9440 | 1.5174 | 0.6089 |
1.3776 | 17.0 | 10030 | 1.4633 | 0.6278 |
1.344 | 18.0 | 10620 | 1.4902 | 0.6135 |
1.3644 | 19.0 | 11210 | 1.3897 | 0.6615 |
1.3559 | 20.0 | 11800 | 1.3980 | 0.6670 |
1.3053 | 21.0 | 12390 | 1.4601 | 0.6651 |
1.3035 | 22.0 | 12980 | 1.3306 | 0.6700 |
1.3067 | 23.0 | 13570 | 1.3644 | 0.6700 |
1.2856 | 24.0 | 14160 | 1.2897 | 0.6691 |
1.2743 | 25.0 | 14750 | 1.3909 | 0.6691 |
1.2704 | 26.0 | 15340 | 1.2935 | 0.6642 |
1.2606 | 27.0 | 15930 | 1.2985 | 0.6425 |
1.2164 | 28.0 | 16520 | 1.3179 | 0.6761 |
1.2137 | 29.0 | 17110 | 1.2708 | 0.6768 |
1.2185 | 30.0 | 17700 | 1.2182 | 0.6862 |
1.1769 | 31.0 | 18290 | 1.2422 | 0.6682 |
1.1815 | 32.0 | 18880 | 1.3006 | 0.6777 |
1.1648 | 33.0 | 19470 | 1.2125 | 0.6862 |
1.1368 | 34.0 | 20060 | 1.1602 | 0.6661 |
1.1736 | 35.0 | 20650 | 1.1483 | 0.6835 |
1.1383 | 36.0 | 21240 | 1.1702 | 0.6896 |
1.1406 | 37.0 | 21830 | 1.1127 | 0.6835 |
1.1461 | 38.0 | 22420 | 1.1293 | 0.6875 |
1.1199 | 39.0 | 23010 | 1.1855 | 0.6881 |
1.0878 | 40.0 | 23600 | 1.1871 | 0.6902 |
1.0852 | 41.0 | 24190 | 1.0959 | 0.6936 |
1.0873 | 42.0 | 24780 | 1.1361 | 0.6942 |
1.0633 | 43.0 | 25370 | 1.0750 | 0.6911 |
1.0758 | 44.0 | 25960 | 1.1282 | 0.6645 |
1.0446 | 45.0 | 26550 | 1.0763 | 0.6832 |
1.0373 | 46.0 | 27140 | 1.0759 | 0.6817 |
1.0318 | 47.0 | 27730 | 1.0454 | 0.6908 |
1.0354 | 48.0 | 28320 | 1.0636 | 0.7031 |
1.0276 | 49.0 | 28910 | 1.0394 | 0.6927 |
1.0211 | 50.0 | 29500 | 1.0369 | 0.7015 |
1.0021 | 51.0 | 30090 | 1.0366 | 0.6865 |
0.983 | 52.0 | 30680 | 1.0274 | 0.6960 |
1.0137 | 53.0 | 31270 | 1.0278 | 0.7028 |
0.9825 | 54.0 | 31860 | 1.0339 | 0.6899 |
0.9792 | 55.0 | 32450 | 1.0142 | 0.6969 |
0.9937 | 56.0 | 33040 | 1.0140 | 0.7024 |
0.9755 | 57.0 | 33630 | 1.0173 | 0.6972 |
0.9517 | 58.0 | 34220 | 1.0078 | 0.7 |
0.988 | 59.0 | 34810 | 1.0116 | 0.7018 |
0.9702 | 60.0 | 35400 | 1.0090 | 0.6991 |
Framework versions
- Transformers 4.30.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3