afro-xlmr-base-ary-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: 0.3039
• F1: 0.5359
• Roc Auc: 0.7304
• Accuracy: 0.5105

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
0.3909	1.0	108	0.4005	0.0	0.5	0.2471
0.3379	2.0	216	0.3598	0.0617	0.5153	0.2727
0.3195	3.0	324	0.3172	0.2951	0.6014	0.4336
0.2657	4.0	432	0.3034	0.3616	0.6395	0.4848
0.2547	5.0	540	0.2844	0.4671	0.6847	0.5035
0.2004	6.0	648	0.2903	0.4405	0.6738	0.5012
0.1695	7.0	756	0.2880	0.4687	0.6886	0.5221
0.1317	8.0	864	0.3039	0.5359	0.7304	0.5105
0.112	9.0	972	0.3076	0.4946	0.6997	0.5198
0.1025	10.0	1080	0.3090	0.4987	0.7032	0.5058
0.0946	11.0	1188	0.3187	0.5202	0.7145	0.5221
0.0848	12.0	1296	0.3326	0.4990	0.7012	0.5198

Framework versions

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