multilingual-xlm-roberta-base-lumasaba-ner-v1

This model is a fine-tuned version of xlm-roberta-base on the Beijuka/Multilingual_PII_NER_dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.4011
• Precision: 0.9757
• Recall: 0.9511
• F1: 0.9632
• Accuracy: 0.9546

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: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	398	0.8255	0.8041	0.7424	0.7720	0.7204
1.3835	2.0	796	0.4975	0.8921	0.8348	0.8625	0.8448
0.5267	3.0	1194	0.3900	0.8921	0.8739	0.8829	0.8842
0.3019	4.0	1592	0.3742	0.9178	0.8919	0.9047	0.9065
0.3019	5.0	1990	0.2945	0.9358	0.9358	0.9358	0.9314
0.1825	6.0	2388	0.3457	0.9406	0.9303	0.9354	0.9275
0.1222	7.0	2786	0.3336	0.9418	0.9507	0.9462	0.9421
0.0787	8.0	3184	0.4745	0.9610	0.9264	0.9434	0.9374
0.0563	9.0	3582	0.4344	0.9472	0.9421	0.9446	0.9340
0.0563	10.0	3980	0.3393	0.9594	0.9616	0.9605	0.9481
0.0528	11.0	4378	0.4272	0.9518	0.9287	0.9402	0.9305
0.0356	12.0	4776	0.3165	0.9708	0.9632	0.9670	0.9580
0.0283	13.0	5174	0.3633	0.9629	0.9546	0.9587	0.9533
0.0254	14.0	5572	0.3616	0.9490	0.9616	0.9553	0.9520
0.0254	15.0	5970	0.3528	0.9490	0.9624	0.9557	0.9498

Framework versions

• Transformers 4.55.4
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4