clauseguard-legal-bert

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: 0.0238
• Micro F1: 0.7604
• Macro F1: 0.7794
• Precision: 0.7374
• Recall: 0.7849
• F1 Limitation of l: 0.625
• F1 Unilateral term: 0.8101
• F1 Unilateral chan: 0.7143
• F1 Content removal: 0.75
• F1 Contract by usi: 0.8
• F1 Choice of law: 0.8889
• F1 Jurisdiction: 0.9412
• F1 Arbitration: 0.7059

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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Micro F1	Macro F1	Precision	Recall	F1 Limitation of l	F1 Unilateral term	F1 Unilateral chan	F1 Content removal	F1 Contract by usi	F1 Choice of law	F1 Jurisdiction	F1 Arbitration
0.0879	1.0	346	0.0753	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
0.0390	2.0	692	0.0381	0.1864	0.0766	0.8667	0.1044	0.4565	0.1562	0.0	0.0	0.0	0.0	0.0	0.0
0.0313	3.0	1038	0.0273	0.7069	0.6811	0.7328	0.6827	0.7188	0.7788	0.6792	0.5116	0.72	0.7586	0.8205	0.4615
0.0271	4.0	1384	0.0284	0.6819	0.6798	0.6062	0.7791	0.6625	0.7903	0.6301	0.6154	0.5075	0.8387	0.8718	0.5217
0.0132	5.0	1730	0.0264	0.7059	0.7106	0.6897	0.7229	0.6726	0.7717	0.6076	0.6207	0.7317	0.8	0.8718	0.6087
0.0115	6.0	2076	0.0265	0.7081	0.6938	0.7308	0.6867	0.7107	0.7826	0.6364	0.6429	0.7234	0.7333	0.6897	0.6316
0.0067	7.0	2422	0.0254	0.7495	0.7520	0.7153	0.7871	0.7183	0.7769	0.7368	0.7097	0.7391	0.8485	0.85	0.6364
0.0037	8.0	2768	0.0285	0.7250	0.7358	0.6633	0.7992	0.6950	0.7376	0.7273	0.6667	0.68	0.8485	0.8947	0.6364
0.0052	9.0	3114	0.0278	0.7337	0.7267	0.6840	0.7912	0.7338	0.8167	0.6234	0.6207	0.7391	0.8485	0.8718	0.56
0.0035	10.0	3460	0.0298	0.7338	0.7314	0.6645	0.8193	0.7143	0.8	0.6575	0.6667	0.7083	0.8235	0.8718	0.6087

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2