legal_long_legal_ver2_test_sm

This model is a fine-tuned version of kiddothe2b/legal-longformer-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.7379
• Accuracy: 0.5519
• Precision: 0.5139
• Recall: 0.5663
• F1: 0.5388
• D-index: 1.2690

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: 16
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	D-index
No log	0.98	26	0.6957	0.4599	0.44	0.6173	0.5138	1.1026
No log	2.0	53	0.6960	0.4528	0.4262	0.5306	0.4727	1.0831
No log	2.98	79	0.6978	0.4481	0.4312	0.6071	0.5042	1.0794
No log	4.0	106	0.7232	0.4788	0.4622	0.7806	0.5806	1.1493
No log	4.98	132	0.7340	0.5189	0.4828	0.5714	0.5234	1.2095
No log	6.0	159	0.8623	0.5425	0.5049	0.5255	0.515	1.2493
No log	6.98	185	1.2325	0.5448	0.5116	0.3367	0.4062	1.2412
No log	8.0	212	1.4773	0.5165	0.4717	0.3827	0.4225	1.1925
No log	8.98	238	1.6199	0.5330	0.4941	0.4286	0.4590	1.2258
No log	10.0	265	1.8976	0.5259	0.4900	0.6276	0.5503	1.2261
No log	10.98	291	2.1687	0.4953	0.4622	0.5612	0.5069	1.1653
No log	12.0	318	2.3087	0.4882	0.4578	0.5816	0.5124	1.1535
No log	12.98	344	2.5168	0.4953	0.4667	0.6429	0.5408	1.1708
No log	14.0	371	2.5389	0.5142	0.4788	0.5765	0.5231	1.2012
No log	14.98	397	2.4224	0.5330	0.4957	0.5918	0.5395	1.2366
No log	16.0	424	2.6391	0.5212	0.4852	0.5867	0.5312	1.2148
No log	16.98	450	2.7235	0.5307	0.4932	0.5510	0.5205	1.2297
No log	18.0	477	2.7272	0.5425	0.5045	0.5765	0.5381	1.2527
0.2333	18.98	503	2.7222	0.5495	0.5117	0.5561	0.5330	1.2641
0.2333	19.62	520	2.7379	0.5519	0.5139	0.5663	0.5388	1.2690

Framework versions

• Transformers 4.27.4
• Pytorch 1.13.1+cu116
• Datasets 2.11.0
• Tokenizers 0.13.2