afro-xlmr-base-afr-finetuned-augmentation-LUNAR

This model is a fine-tuned version of Davlan/afro-xlmr-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2755
• F1: 0.5012
• Roc Auc: 0.6931
• Accuracy: 0.6908

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.4289	1.0	76	0.3283	0.0	0.5	0.3520
0.3509	2.0	152	0.3160	0.1005	0.5277	0.4836
0.2864	3.0	228	0.2631	0.2532	0.5927	0.5855
0.2383	4.0	304	0.2411	0.2210	0.5799	0.5559
0.1902	5.0	380	0.2632	0.2619	0.5909	0.5921
0.1545	6.0	456	0.2466	0.3297	0.6281	0.6612
0.1408	7.0	532	0.2407	0.3469	0.6253	0.6447
0.1087	8.0	608	0.2463	0.3989	0.6645	0.6809
0.0743	9.0	684	0.2497	0.4086	0.6561	0.6711
0.0832	10.0	760	0.2581	0.4063	0.6516	0.6908
0.0599	11.0	836	0.2583	0.4992	0.6911	0.7007
0.0553	12.0	912	0.2755	0.5012	0.6931	0.6908
0.0497	13.0	988	0.2741	0.4253	0.6688	0.6809
0.029	14.0	1064	0.2864	0.4033	0.6529	0.6776
0.0349	15.0	1140	0.3034	0.4742	0.6808	0.6776
0.0249	16.0	1216	0.2889	0.4089	0.6608	0.6678

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0