roberta-base-legal-multi-downstream-indian-ner

This model is a fine-tuned version of MHGanainy/roberta-base-legal-multi on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2526
• Precision: 0.6406
• Recall: 0.8244
• F1: 0.7210
• Accuracy: 0.9663

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: 3e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 1
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	172	0.2310	0.1233	0.4904	0.1971	0.8307
No log	2.0	344	0.1929	0.1983	0.5393	0.2900	0.8765
0.4324	3.0	516	0.1667	0.1773	0.4897	0.2604	0.8738
0.4324	4.0	688	0.1836	0.2957	0.6059	0.3975	0.9081
0.4324	5.0	860	0.2005	0.2855	0.5623	0.3787	0.9137
0.1106	6.0	1032	0.2003	0.3858	0.6974	0.4968	0.9323
0.1106	7.0	1204	0.2224	0.4182	0.6719	0.5155	0.9428
0.1106	8.0	1376	0.2221	0.3347	0.6147	0.4334	0.9312
0.0589	9.0	1548	0.1960	0.4067	0.7026	0.5152	0.9404
0.0589	10.0	1720	0.1904	0.5049	0.7410	0.6006	0.9524
0.0589	11.0	1892	0.2274	0.5337	0.7707	0.6307	0.9565
0.0359	12.0	2064	0.2471	0.5525	0.7696	0.6432	0.9575
0.0359	13.0	2236	0.2352	0.5649	0.7675	0.6508	0.9591
0.0359	14.0	2408	0.2297	0.5530	0.7661	0.6424	0.9586
0.0224	15.0	2580	0.2349	0.5702	0.7923	0.6632	0.9597
0.0224	16.0	2752	0.2465	0.6033	0.8052	0.6898	0.9624
0.0224	17.0	2924	0.2428	0.6100	0.8098	0.6959	0.9647
0.0143	18.0	3096	0.2543	0.6238	0.8154	0.7068	0.9646
0.0143	19.0	3268	0.2526	0.6305	0.8161	0.7114	0.9651
0.0143	20.0	3440	0.2526	0.6406	0.8244	0.7210	0.9663

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1