legal-bert-base-uncased

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: 1.1536
• Accuracy: 0.8203
• Precision: 0.8212
• Recall: 0.8203
• Precision Macro: 0.7660
• Recall Macro: 0.7548
• Macro Fpr: 0.0156
• Weighted Fpr: 0.0150
• Weighted Specificity: 0.9766
• Macro Specificity: 0.9867
• Weighted Sensitivity: 0.8242
• Macro Sensitivity: 0.7548
• F1 Micro: 0.8242
• F1 Macro: 0.7566
• F1 Weighted: 0.8221

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	Precision Macro	Recall Macro	Macro Fpr	Weighted Fpr	Weighted Specificity	Macro Specificity	Weighted Sensitivity	Macro Sensitivity	F1 Micro	F1 Macro	F1 Weighted
1.1096	1.0	643	0.6748	0.7978	0.7855	0.7978	0.6239	0.6340	0.0188	0.0178	0.9702	0.9845	0.7978	0.6340	0.7978	0.6134	0.7840
0.6187	2.0	1286	0.6449	0.8110	0.8196	0.8110	0.7806	0.7327	0.0169	0.0164	0.9755	0.9858	0.8110	0.7327	0.8110	0.7268	0.8090
0.4747	3.0	1929	0.8151	0.8149	0.8192	0.8149	0.7659	0.7390	0.0166	0.0160	0.9761	0.9861	0.8149	0.7390	0.8149	0.7370	0.8125
0.2645	4.0	2572	0.9345	0.8218	0.8198	0.8218	0.7446	0.7413	0.0158	0.0152	0.9774	0.9866	0.8218	0.7413	0.8218	0.7385	0.8189
0.1901	5.0	3215	1.0929	0.8195	0.8242	0.8195	0.8264	0.7432	0.0161	0.0155	0.9750	0.9863	0.8195	0.7432	0.8195	0.7595	0.8166
0.1131	6.0	3858	1.1536	0.8203	0.8212	0.8203	0.7968	0.7786	0.0159	0.0154	0.9766	0.9865	0.8203	0.7786	0.8203	0.7840	0.8197
0.063	7.0	4501	1.3218	0.8118	0.8184	0.8118	0.7518	0.7526	0.0166	0.0163	0.9773	0.9859	0.8118	0.7526	0.8118	0.7495	0.8136
0.0264	8.0	5144	1.3863	0.8257	0.8262	0.8257	0.7784	0.7768	0.0155	0.0149	0.9768	0.9868	0.8257	0.7768	0.8257	0.7730	0.8247
0.03	9.0	5787	1.5542	0.8079	0.8167	0.8079	0.7639	0.7653	0.0172	0.0167	0.9744	0.9855	0.8079	0.7653	0.8079	0.7595	0.8096
0.0149	10.0	6430	1.5835	0.8141	0.8155	0.8141	0.7545	0.7361	0.0168	0.0160	0.9730	0.9858	0.8141	0.7361	0.8141	0.7412	0.8127
0.005	11.0	7073	1.5325	0.8242	0.8250	0.8242	0.7805	0.7812	0.0156	0.0150	0.9758	0.9867	0.8242	0.7812	0.8242	0.7681	0.8226
0.003	12.0	7716	1.5714	0.8288	0.8299	0.8288	0.7701	0.7679	0.0152	0.0145	0.9765	0.9870	0.8288	0.7679	0.8288	0.7626	0.8276
0.0033	13.0	8359	1.5511	0.8249	0.8219	0.8249	0.7676	0.7598	0.0156	0.0149	0.9760	0.9867	0.8249	0.7598	0.8249	0.7608	0.8225
0.0018	14.0	9002	1.5510	0.8249	0.8225	0.8249	0.7686	0.7554	0.0155	0.0149	0.9767	0.9868	0.8249	0.7554	0.8249	0.7572	0.8224
0.0008	15.0	9645	1.5469	0.8242	0.8220	0.8242	0.7660	0.7548	0.0156	0.0150	0.9766	0.9867	0.8242	0.7548	0.8242	0.7566	0.8221

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.2