Modelpublic

wiorz/legal_bert_small_defined_summarized

sourceHugging Facecc-by-sa-4.0updated 3y agoView on Hugging Face

0likes15downloads

Model Card

legalbertsmalldefinedsummarized

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: 1.4178
—Accuracy: 0.87
—Precision: 0.6
—Recall: 0.2143
—F1: 0.3158
—D-index: 1.5771

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
—trainbatchsize: 4
—evalbatchsize: 8
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 1600
—num_epochs: 20
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	D-index
No log	1.0	200	0.4008	0.86	0.0	0.0	0.0	1.4803
No log	2.0	400	0.3871	0.86	0.0	0.0	0.0	1.4803
0.5179	3.0	600	0.4293	0.87	0.625	0.1786	0.2778	1.5635
0.5179	4.0	800	0.6702	0.87	0.625	0.1786	0.2778	1.5635
0.3816	5.0	1000	0.7388	0.865	0.5455	0.2143	0.3077	1.5706
0.3816	6.0	1200	1.0422	0.86	0.5	0.1786	0.2632	1.5504
0.3816	7.0	1400	1.0804	0.875	0.7143	0.1786	0.2857	1.5700
0.0567	8.0	1600	1.1490	0.875	0.6364	0.25	0.3590	1.5970
0.0567	9.0	1800	1.3190	0.865	0.5556	0.1786	0.2703	1.5570
0.0125	10.0	2000	1.4220	0.835	0.3913	0.3214	0.3529	1.5718
0.0125	11.0	2200	1.3567	0.855	0.4706	0.2857	0.3556	1.5845
0.0125	12.0	2400	1.3349	0.875	0.7143	0.1786	0.2857	1.5700
0.0021	13.0	2600	1.3494	0.87	0.5714	0.2857	0.3810	1.6038
0.0021	14.0	2800	1.3747	0.87	0.6	0.2143	0.3158	1.5771
0.0	15.0	3000	1.3890	0.87	0.6	0.2143	0.3158	1.5771
0.0	16.0	3200	1.4069	0.875	0.6667	0.2143	0.3243	1.5835
0.0	17.0	3400	1.4185	0.875	0.6667	0.2143	0.3243	1.5835
0.0	18.0	3600	1.3945	0.865	0.5385	0.25	0.3415	1.5840
0.0	19.0	3800	1.3921	0.87	0.6	0.2143	0.3158	1.5771
0.0037	20.0	4000	1.4178	0.87	0.6	0.2143	0.3158	1.5771

Framework versions

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