legal-bert-base-uncased_FineTune_20251006_184619
This model is a fine-tuned version of nlpaueb/legal-bert-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0121
- Macro F1: 0.3502
- Micro F1: 0.8170
- Macro Precision: 0.3794
- Macro Recall: 0.3518
- Sample F1: 0.8225
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: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 35
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Macro F1 | Micro F1 | Macro Precision | Macro Recall | Sample F1 |
|---|---|---|---|---|---|---|---|---|
| 0.0142 | 1.0 | 426 | 0.0124 | 0.0137 | 0.5524 | 0.0147 | 0.0136 | 0.5342 |
| 0.0077 | 2.0 | 852 | 0.0076 | 0.0358 | 0.7366 | 0.0417 | 0.0348 | 0.7250 |
| 0.0058 | 3.0 | 1278 | 0.0059 | 0.0828 | 0.7708 | 0.1284 | 0.0730 | 0.7661 |
| 0.0048 | 4.0 | 1704 | 0.0051 | 0.1320 | 0.7920 | 0.1940 | 0.1189 | 0.7931 |
| 0.0039 | 5.0 | 2130 | 0.0049 | 0.2103 | 0.8026 | 0.2662 | 0.1961 | 0.8047 |
| 0.0037 | 6.0 | 2556 | 0.0050 | 0.2170 | 0.8143 | 0.2658 | 0.2042 | 0.8157 |
| 0.0029 | 7.0 | 2982 | 0.0051 | 0.2780 | 0.8017 | 0.3243 | 0.2710 | 0.7965 |
| 0.0026 | 8.0 | 3408 | 0.0055 | 0.2576 | 0.8077 | 0.3115 | 0.2396 | 0.8088 |
| 0.0021 | 9.0 | 3834 | 0.0059 | 0.3102 | 0.8077 | 0.3414 | 0.3109 | 0.8123 |
| 0.0012 | 10.0 | 4260 | 0.0062 | 0.3112 | 0.8133 | 0.3430 | 0.3126 | 0.8154 |
| 0.0013 | 11.0 | 4686 | 0.0072 | 0.2843 | 0.8051 | 0.3465 | 0.2772 | 0.8065 |
| 0.0009 | 12.0 | 5112 | 0.0078 | 0.3197 | 0.7968 | 0.3579 | 0.3198 | 0.7961 |
| 0.0007 | 13.0 | 5538 | 0.0077 | 0.3291 | 0.8070 | 0.3707 | 0.3222 | 0.8111 |
| 0.0005 | 14.0 | 5964 | 0.0084 | 0.3356 | 0.8015 | 0.3664 | 0.3327 | 0.8041 |
| 0.0005 | 15.0 | 6390 | 0.0091 | 0.3275 | 0.8064 | 0.3653 | 0.3272 | 0.8100 |
| 0.0005 | 16.0 | 6816 | 0.0089 | 0.3488 | 0.8046 | 0.3884 | 0.3435 | 0.8097 |
| 0.0004 | 17.0 | 7242 | 0.0097 | 0.2924 | 0.8017 | 0.3188 | 0.2963 | 0.8058 |
| 0.0003 | 18.0 | 7668 | 0.0093 | 0.3418 | 0.8101 | 0.3730 | 0.3395 | 0.8111 |
| 0.0002 | 19.0 | 8094 | 0.0099 | 0.3422 | 0.8152 | 0.3842 | 0.3313 | 0.8185 |
| 0.0001 | 20.0 | 8520 | 0.0102 | 0.3287 | 0.8065 | 0.3618 | 0.3239 | 0.8120 |
| 0.0001 | 21.0 | 8946 | 0.0102 | 0.3217 | 0.8132 | 0.3495 | 0.3236 | 0.8172 |
| 0.0001 | 22.0 | 9372 | 0.0106 | 0.3481 | 0.8146 | 0.3769 | 0.3497 | 0.8191 |
| 0.0001 | 23.0 | 9798 | 0.0107 | 0.3327 | 0.8137 | 0.3666 | 0.3335 | 0.8183 |
| 0.0 | 24.0 | 10224 | 0.0112 | 0.3628 | 0.8131 | 0.3865 | 0.3689 | 0.8187 |
| 0.0 | 25.0 | 10650 | 0.0113 | 0.3593 | 0.8145 | 0.3926 | 0.3575 | 0.8190 |
| 0.0 | 26.0 | 11076 | 0.0115 | 0.3469 | 0.8162 | 0.3764 | 0.3465 | 0.8210 |
| 0.0 | 27.0 | 11502 | 0.0117 | 0.3501 | 0.8163 | 0.3773 | 0.3511 | 0.8208 |
| 0.0 | 28.0 | 11928 | 0.0119 | 0.3438 | 0.8162 | 0.3771 | 0.3444 | 0.8219 |
| 0.0 | 29.0 | 12354 | 0.0119 | 0.3476 | 0.8163 | 0.3806 | 0.3471 | 0.8218 |
| 0.0 | 30.0 | 12780 | 0.0120 | 0.3466 | 0.8173 | 0.3802 | 0.3450 | 0.8229 |
| 0.0 | 31.0 | 13206 | 0.0120 | 0.3505 | 0.8171 | 0.3805 | 0.3507 | 0.8225 |
| 0.0 | 32.0 | 13632 | 0.0121 | 0.3502 | 0.8170 | 0.3794 | 0.3518 | 0.8225 |
Framework versions
- Transformers 4.56.2
- Pytorch 2.8.0+cu126
- Datasets 4.0.0
- Tokenizers 0.22.1