scotus-v10

This model is a fine-tuned version of nlpaueb/legal-bert-small-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7971
• Accuracy: 0.8842

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: 2e-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
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.67	1.0	1219	1.3781	0.5718
1.1539	2.0	2438	0.9619	0.7077
0.6736	3.0	3657	0.7837	0.7668
0.4265	4.0	4876	0.5882	0.8309
0.2477	5.0	6095	0.5624	0.8446
0.1636	6.0	7314	0.5701	0.8542
0.0938	7.0	8533	0.6322	0.8603
0.0523	8.0	9752	0.6717	0.8606
0.0397	9.0	10971	0.6626	0.8738
0.0236	10.0	12190	0.6729	0.8811
0.0129	11.0	13409	0.7164	0.876
0.0098	12.0	14628	0.8041	0.8723
0.0066	13.0	15847	0.7792	0.8794
0.0055	14.0	17066	0.7704	0.8792
0.0032	15.0	18285	0.8369	0.8768
0.0023	16.0	19504	0.8685	0.8737
0.0012	17.0	20723	0.7631	0.888
0.0026	18.0	21942	0.8220	0.8808
0.0011	19.0	23161	0.7897	0.8845
0.0003	20.0	24380	0.7971	0.8842

Framework versions

• Transformers 4.33.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3