Metrics

• loss: 1.0434
• accuracy: 0.8218
• precision: 0.8145
• recall: 0.8218
• precision_macro: 0.6907
• recall_macro: 0.6533
• macro_fpr: 0.0897
• weighted_fpr: 0.0674
• weighted_specificity: 0.8528
• macro_specificity: 0.9187
• weighted_sensitivity: 0.8218
• macro_sensitivity: 0.6533
• f1_micro: 0.8218
• f1_macro: 0.6690
• f1_weighted: 0.8159
• runtime: 198.6459
• samples_per_second: 2.2600
• steps_per_second: 0.2870

case-analysis-InLegalBERT

This model is a fine-tuned version of law-ai/InLegalBERT on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.0434
• Accuracy: 0.8218
• Precision: 0.8145
• Recall: 0.8218
• Precision Macro: 0.6439
• Recall Macro: 0.6295
• Macro Fpr: 0.0890
• Weighted Fpr: 0.0674
• Weighted Specificity: 0.8544
• Macro Specificity: 0.9191
• Weighted Sensitivity: 0.8218
• Macro Sensitivity: 0.6295
• F1 Micro: 0.8218
• F1 Macro: 0.6335
• F1 Weighted: 0.8106

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: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	Precision Macro	Recall Macro	Macro Fpr	Weighted Fpr	Weighted Specificity	Macro Specificity	Weighted Sensitivity	Macro Sensitivity	F1 Micro	F1 Macro	F1 Weighted
No log	1.0	224	0.6546	0.8018	0.7632	0.8018	0.5777	0.6106	0.0978	0.0761	0.8432	0.9112	0.8018	0.6106	0.8018	0.5936	0.7820
No log	2.0	448	0.6831	0.8129	0.7732	0.8129	0.5845	0.6154	0.0923	0.0712	0.8554	0.9171	0.8129	0.6154	0.8129	0.5996	0.7926
0.607	3.0	672	0.7626	0.8263	0.8060	0.8263	0.6773	0.6341	0.0885	0.0655	0.8464	0.9182	0.8263	0.6341	0.8263	0.6362	0.8105
0.607	4.0	896	0.7839	0.8085	0.7991	0.8085	0.6391	0.6306	0.0896	0.0732	0.8754	0.9210	0.8085	0.6306	0.8085	0.6314	0.8017
0.316	5.0	1120	0.9381	0.8263	0.8127	0.8263	0.6688	0.6573	0.0822	0.0655	0.8780	0.9261	0.8263	0.6573	0.8263	0.6514	0.8161
0.316	6.0	1344	1.0434	0.8218	0.8145	0.8218	0.6907	0.6533	0.0897	0.0674	0.8528	0.9187	0.8218	0.6533	0.8218	0.6690	0.8159
0.1513	7.0	1568	1.2182	0.8018	0.8066	0.8018	0.6382	0.6399	0.0916	0.0761	0.8802	0.9205	0.8018	0.6399	0.8018	0.6375	0.8030
0.1513	8.0	1792	1.3193	0.8285	0.8070	0.8285	0.6566	0.6280	0.0882	0.0645	0.8521	0.9202	0.8285	0.6280	0.8285	0.6376	0.8152
0.0491	9.0	2016	1.3169	0.8330	0.8180	0.8330	0.6950	0.6555	0.0828	0.0627	0.8653	0.9246	0.8330	0.6555	0.8330	0.6687	0.8235
0.0491	10.0	2240	1.4460	0.8307	0.8109	0.8307	0.6584	0.6291	0.0868	0.0636	0.8533	0.9210	0.8307	0.6291	0.8307	0.6398	0.8184
0.0491	11.0	2464	1.4100	0.8419	0.8166	0.8419	0.6718	0.6399	0.0806	0.0589	0.8642	0.9265	0.8419	0.6399	0.8419	0.6464	0.8263
0.0148	12.0	2688	1.5364	0.8218	0.8105	0.8218	0.6661	0.6340	0.0903	0.0674	0.8505	0.9181	0.8218	0.6340	0.8218	0.6469	0.8137
0.0148	13.0	2912	1.5380	0.8307	0.8118	0.8307	0.6596	0.6304	0.0870	0.0636	0.8512	0.9205	0.8307	0.6304	0.8307	0.6409	0.8185
0.0031	14.0	3136	1.6139	0.8218	0.8108	0.8218	0.6451	0.6353	0.0860	0.0674	0.8685	0.9226	0.8218	0.6353	0.8218	0.6396	0.8159
0.0031	15.0	3360	1.6356	0.8263	0.8117	0.8263	0.6626	0.6477	0.0842	0.0655	0.8700	0.9241	0.8263	0.6477	0.8263	0.6529	0.8183
0.0043	16.0	3584	1.6745	0.8241	0.7994	0.8241	0.6244	0.6229	0.0884	0.0664	0.8543	0.9196	0.8241	0.6229	0.8241	0.6231	0.8108
0.0043	17.0	3808	1.7867	0.8085	0.7946	0.8085	0.6221	0.6336	0.0906	0.0732	0.8678	0.9191	0.8085	0.6336	0.8085	0.6229	0.7996
0.0008	18.0	4032	1.7511	0.8151	0.7971	0.8151	0.6110	0.6216	0.0901	0.0703	0.8644	0.9199	0.8151	0.6216	0.8151	0.6145	0.8046
0.0008	19.0	4256	1.5909	0.8441	0.8079	0.8441	0.6260	0.6374	0.0792	0.0580	0.8670	0.9278	0.8441	0.6374	0.8441	0.6311	0.8249
0.0008	20.0	4480	1.5721	0.8463	0.8212	0.8463	0.6727	0.6546	0.0761	0.0571	0.8753	0.9304	0.8463	0.6546	0.8463	0.6547	0.8316
0.0039	21.0	4704	1.5819	0.8396	0.8054	0.8396	0.6337	0.6200	0.0843	0.0599	0.8527	0.9231	0.8396	0.6200	0.8396	0.6245	0.8199
0.0039	22.0	4928	1.5906	0.8486	0.8236	0.8486	0.6814	0.6512	0.0770	0.0562	0.8680	0.9291	0.8486	0.6512	0.8486	0.6570	0.8333
0.0005	23.0	5152	1.7133	0.8263	0.8047	0.8263	0.6403	0.6431	0.0831	0.0655	0.8745	0.9252	0.8263	0.6431	0.8263	0.6367	0.8143
0.0005	24.0	5376	1.7813	0.8241	0.8022	0.8241	0.6515	0.6290	0.0894	0.0664	0.8490	0.9183	0.8241	0.6290	0.8241	0.6348	0.8108
0.0033	25.0	5600	1.7983	0.8218	0.8001	0.8218	0.6485	0.6281	0.0902	0.0674	0.8486	0.9176	0.8218	0.6281	0.8218	0.6328	0.8088
0.0033	26.0	5824	1.8070	0.8218	0.8001	0.8218	0.6485	0.6281	0.0902	0.0674	0.8486	0.9176	0.8218	0.6281	0.8218	0.6328	0.8088
0.0	27.0	6048	1.8198	0.8218	0.8024	0.8218	0.6439	0.6295	0.0890	0.0674	0.8544	0.9191	0.8218	0.6295	0.8218	0.6335	0.8106
0.0	28.0	6272	1.8243	0.8218	0.8024	0.8218	0.6439	0.6295	0.0890	0.0674	0.8544	0.9191	0.8218	0.6295	0.8218	0.6335	0.8106
0.0	29.0	6496	1.8277	0.8218	0.8024	0.8218	0.6439	0.6295	0.0890	0.0674	0.8544	0.9191	0.8218	0.6295	0.8218	0.6335	0.8106
0.0003	30.0	6720	1.8292	0.8218	0.8024	0.8218	0.6439	0.6295	0.0890	0.0674	0.8544	0.9191	0.8218	0.6295	0.8218	0.6335	0.8106

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.15.2