legal_bert_sm_cv_summarized_defined_4

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: 1.6665
• Accuracy: 0.811
• Precision: 0.5357
• Recall: 0.2308
• F1: 0.3226
• D-index: 1.5269

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 8000
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	D-index
No log	1.0	250	0.4854	0.804	0.0	0.0	0.0	1.4356
0.5596	2.0	500	0.4545	0.805	0.0	0.0	0.0	1.4370
0.5596	3.0	750	0.4570	0.811	0.6667	0.0615	0.1127	1.4675
0.4293	4.0	1000	0.4673	0.815	0.6316	0.1231	0.2060	1.4949
0.4293	5.0	1250	0.4893	0.828	0.6949	0.2103	0.3228	1.5429
0.3311	6.0	1500	0.5062	0.828	0.6533	0.2513	0.3630	1.5569
0.3311	7.0	1750	0.5584	0.826	0.7059	0.1846	0.2927	1.5313
0.2126	8.0	2000	0.7423	0.821	0.6333	0.1949	0.2980	1.5281
0.2126	9.0	2250	0.8720	0.804	0.4933	0.1897	0.2741	1.5031
0.1327	10.0	2500	0.9116	0.811	0.5268	0.3026	0.3844	1.5513
0.1327	11.0	2750	0.9677	0.8	0.4809	0.3231	0.3865	1.5434
0.0803	12.0	3000	1.1951	0.795	0.4627	0.3179	0.3769	1.5349
0.0803	13.0	3250	1.3724	0.819	0.5946	0.2256	0.3271	1.5360
0.0584	14.0	3500	1.4260	0.806	0.5056	0.2308	0.3169	1.5201
0.0584	15.0	3750	1.4684	0.812	0.5327	0.2923	0.3775	1.5492
0.0437	16.0	4000	1.5562	0.815	0.5658	0.2205	0.3173	1.5288
0.0437	17.0	4250	1.5812	0.814	0.5763	0.1744	0.2677	1.5115
0.0357	18.0	4500	1.6058	0.805	0.5	0.2308	0.3158	1.5187
0.0357	19.0	4750	1.6784	0.813	0.5465	0.2410	0.3345	1.5331
0.0334	20.0	5000	1.6665	0.811	0.5357	0.2308	0.3226	1.5269

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3