legal_bert_small_summarized

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: 2.0708
• Accuracy: 0.815
• Precision: 0.5
• Recall: 0.1622
• F1: 0.2449
• D-index: 1.5040

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: 4
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1600
• 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	200	0.4799	0.815	0.0	0.0	0.0	1.4449
No log	2.0	400	0.5646	0.815	0.0	0.0	0.0	1.4449
0.5383	3.0	600	0.5505	0.815	0.0	0.0	0.0	1.4449
0.5383	4.0	800	0.4502	0.815	0.5	0.2162	0.3019	1.5231
0.5116	5.0	1000	0.6932	0.805	0.4444	0.2162	0.2909	1.5096
0.5116	6.0	1200	1.0173	0.795	0.4231	0.2973	0.3492	1.5244
0.5116	7.0	1400	1.2308	0.82	0.5714	0.1081	0.1818	1.4914
0.1778	8.0	1600	1.4035	0.815	0.5	0.2432	0.3273	1.5326
0.1778	9.0	1800	1.6336	0.815	0.5	0.1622	0.2449	1.5040
0.0255	10.0	2000	1.7291	0.82	0.5385	0.1892	0.28	1.5204
0.0255	11.0	2200	1.7801	0.825	0.5714	0.2162	0.3137	1.5367
0.0255	12.0	2400	1.8364	0.825	0.5714	0.2162	0.3137	1.5367
0.0	13.0	2600	1.8688	0.825	0.5714	0.2162	0.3137	1.5367
0.0	14.0	2800	1.9549	0.815	0.5	0.1622	0.2449	1.5040
0.0	15.0	3000	2.0022	0.815	0.5	0.1622	0.2449	1.5040
0.0	16.0	3200	1.9795	0.82	0.5385	0.1892	0.28	1.5204
0.0	17.0	3400	2.0438	0.815	0.5	0.1622	0.2449	1.5040
0.0	18.0	3600	2.0603	0.815	0.5	0.1622	0.2449	1.5040
0.0	19.0	3800	2.0722	0.815	0.5	0.1622	0.2449	1.5040
0.0014	20.0	4000	2.0708	0.815	0.5	0.1622	0.2449	1.5040

Framework versions

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