kinyarwanda-topic-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: 0.7530
• F1: 0.8560
• Precision: 0.8622
• Recall: 0.8578
• Accuracy: 0.8578

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: 2e-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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Precision	Recall	Accuracy
3.6525	1.0	22	1.5583	0.5944	0.6642	0.6566	0.6566
2.3638	2.0	44	0.8003	0.7966	0.8184	0.7980	0.7980
1.2553	3.0	66	0.5707	0.8248	0.8414	0.8283	0.8283
0.7472	4.0	88	0.4984	0.8482	0.8565	0.8485	0.8485
0.2923	5.0	110	0.6913	0.7861	0.7907	0.7879	0.7879
0.2096	6.0	132	0.7531	0.7972	0.8264	0.8081	0.8081
0.1107	7.0	154	0.7435	0.8384	0.8570	0.8384	0.8384
0.0742	8.0	176	0.8058	0.8591	0.8671	0.8586	0.8586
0.0077	9.0	198	0.9892	0.8086	0.8165	0.8081	0.8081
0.0278	10.0	220	0.9835	0.8478	0.8644	0.8485	0.8485
0.0022	11.0	242	1.2416	0.8191	0.8364	0.8182	0.8182
0.0028	12.0	264	1.0595	0.8386	0.8552	0.8384	0.8384
0.0017	13.0	286	1.1092	0.8470	0.8596	0.8485	0.8485

Framework versions

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