kinyarwanda-ner-afroxlmr-large

This model is a fine-tuned version of Davlan/afro-xlmr-large-76L on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1999
• Precision: 0.8547
• Recall: 0.8725
• F1: 0.8635
• Accuracy: 0.9727
• F1 Per: 0.9380
• F1 Org: 0.7889
• F1 Loc: 0.9089
• F1 Date: 0.7702

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy	F1 Per	F1 Org	F1 Loc	F1 Date
0.1774	1.0	245	0.0964	0.825	0.8902	0.8563	0.9728	0.9487	0.8138	0.8719	0.7425
0.1203	2.0	490	0.1068	0.8136	0.8642	0.8381	0.9709	0.9243	0.8266	0.8669	0.6811
0.0667	3.0	735	0.1071	0.8330	0.8699	0.8511	0.9735	0.9444	0.8041	0.8520	0.7449
0.0609	4.0	980	0.1133	0.8514	0.8969	0.8736	0.9744	0.9419	0.8549	0.8989	0.7497
0.0694	5.0	1225	0.1306	0.8487	0.8728	0.8606	0.9693	0.9447	0.8412	0.8878	0.7093
0.0174	6.0	1470	0.1511	0.8409	0.8935	0.8664	0.9718	0.9443	0.8676	0.8807	0.7179
0.0287	7.0	1715	0.1359	0.8626	0.8887	0.8754	0.9738	0.9421	0.8688	0.8994	0.7378
0.0379	8.0	1960	0.1502	0.8446	0.8926	0.8679	0.9728	0.9535	0.8440	0.8853	0.7367
0.0118	9.0	2205	0.1579	0.8543	0.8950	0.8741	0.9732	0.9575	0.8532	0.8831	0.7506
0.0264	10.0	2450	0.1537	0.8704	0.9061	0.8879	0.9762	0.9635	0.8798	0.8999	0.7556
0.0143	11.0	2695	0.1554	0.8767	0.8940	0.8853	0.9750	0.9486	0.8665	0.9015	0.7795
0.0130	12.0	2940	0.1976	0.8458	0.8825	0.8637	0.9725	0.9402	0.8297	0.8833	0.7490
0.0047	13.0	3185	0.2175	0.8642	0.8825	0.8732	0.9724	0.9491	0.8510	0.8918	0.7455
0.0031	14.0	3430	0.1827	0.8616	0.8935	0.8773	0.9744	0.9434	0.8515	0.8966	0.7739
0.0079	15.0	3675	0.1963	0.8507	0.8863	0.8681	0.9722	0.9448	0.8315	0.8850	0.7610
0.0115	16.0	3920	0.1838	0.8578	0.8921	0.8746	0.9733	0.9585	0.8492	0.8775	0.7608
0.0187	17.0	4165	0.1709	0.8743	0.8979	0.8859	0.9759	0.9568	0.8604	0.9056	0.7744
0.0015	18.0	4410	0.2055	0.8577	0.8911	0.8741	0.9727	0.9465	0.8470	0.8906	0.7640
0.0043	19.0	4655	0.2113	0.8655	0.8955	0.8802	0.9754	0.9623	0.8566	0.8842	0.7663
0.0062	20.0	4900	0.2348	0.8490	0.8858	0.8670	0.9717	0.9520	0.8401	0.8786	0.7465

Framework versions

• Transformers 5.5.4
• Pytorch 2.11.0+cu130
• Datasets 2.21.0
• Tokenizers 0.22.2