new-legal-ner
This model is a fine-tuned version of law-ai/InLegalBERT on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.2370
- Precision: 0.6151
- Recall: 0.8766
- F1: 0.7230
- Accuracy: 0.9444
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: 3e-05
- train_batch_size: 16
- 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: linear
- lr_scheduler_warmup_steps: 0.1
- num_epochs: 6
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Accuracy | F1 | Validation Loss | Precision | Recall |
|---|---|---|---|---|---|---|---|
| 0.3107 | 1.0 | 688 | 0.8167 | 0.3887 | 0.2439 | 0.2676 | 0.7100 |
| 0.2381 | 2.0 | 1376 | 0.2173 | 0.3925 | 0.7724 | 0.5205 | 0.8837 |
| 0.1615 | 3.0 | 2064 | 0.2370 | 0.5189 | 0.8445 | 0.6428 | 0.9209 |
| 0.1052 | 4.0 | 2752 | 0.2241 | 0.5128 | 0.8382 | 0.6363 | 0.9201 |
| 0.0700 | 5.0 | 3440 | 0.2402 | 0.5829 | 0.8705 | 0.6982 | 0.9358 |
| 0.0546 | 6.0 | 4128 | 0.2447 | 0.6195 | 0.8819 | 0.7278 | 0.9432 |
Framework versions
- Transformers 5.12.1
- Pytorch 2.11.0+cu128
- Datasets 4.0.0
- Tokenizers 0.22.2