<!-- 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. -->
legal_bert_sm_gen1_large
This model is a fine-tuned version of nlpaueb/legal-bert-base-uncased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.3342
- Accuracy: 0.8342
- Precision: 0.6462
- Recall: 0.3993
- F1: 0.4936
- D-index: 1.6166
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: 32
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 96000
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 | D-index |
|---|---|---|---|---|---|---|---|---|
| 0.4153 | 1.0 | 1500 | 0.3934 | 0.8279 | 0.7013 | 0.2610 | 0.3804 | 1.5630 |
| 0.3682 | 2.0 | 3000 | 0.3595 | 0.8448 | 0.6802 | 0.4405 | 0.5347 | 1.6439 |
| 0.3509 | 3.0 | 4500 | 0.3559 | 0.847 | 0.7258 | 0.3923 | 0.5094 | 1.6314 |
| 0.3266 | 4.0 | 6000 | 0.3545 | 0.8484 | 0.7335 | 0.3944 | 0.5130 | 1.6339 |
| 0.2927 | 5.0 | 7500 | 0.3728 | 0.8519 | 0.7251 | 0.4323 | 0.5417 | 1.6506 |
| 0.265 | 6.0 | 9000 | 0.3836 | 0.8511 | 0.7019 | 0.4594 | 0.5554 | 1.6581 |
| 0.2284 | 7.0 | 10500 | 0.4332 | 0.8477 | 0.6611 | 0.5076 | 0.5743 | 1.6688 |
| 0.1903 | 8.0 | 12000 | 0.4834 | 0.8452 | 0.6970 | 0.4166 | 0.5215 | 1.6368 |
| 0.1527 | 9.0 | 13500 | 0.5702 | 0.8413 | 0.6809 | 0.4068 | 0.5093 | 1.6285 |
| 0.1296 | 10.0 | 15000 | 0.5942 | 0.8374 | 0.6585 | 0.4088 | 0.5044 | 1.6240 |
| 0.1158 | 11.0 | 16500 | 0.7754 | 0.8408 | 0.6680 | 0.4249 | 0.5194 | 1.6336 |
| 0.1054 | 12.0 | 18000 | 0.7936 | 0.8357 | 0.6062 | 0.5368 | 0.5694 | 1.6622 |
| 0.0879 | 13.0 | 19500 | 1.0568 | 0.8317 | 0.6971 | 0.2985 | 0.4180 | 1.5806 |
| 0.0834 | 14.0 | 21000 | 0.9730 | 0.8377 | 0.6393 | 0.4545 | 0.5313 | 1.6389 |
| 0.0744 | 15.0 | 22500 | 1.0385 | 0.8358 | 0.6390 | 0.4343 | 0.5172 | 1.6301 |
| 0.0675 | 16.0 | 24000 | 1.1625 | 0.8353 | 0.6305 | 0.4496 | 0.5249 | 1.6342 |
| 0.065 | 17.0 | 25500 | 1.2138 | 0.8325 | 0.6546 | 0.3652 | 0.4688 | 1.6034 |
| 0.0539 | 18.0 | 27000 | 1.2701 | 0.8334 | 0.6754 | 0.3409 | 0.4531 | 1.5967 |
| 0.0479 | 19.0 | 28500 | 1.2759 | 0.8367 | 0.6303 | 0.4681 | 0.5372 | 1.6420 |
| 0.0503 | 20.0 | 30000 | 1.3342 | 0.8342 | 0.6462 | 0.3993 | 0.4936 | 1.6166 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3
