afro-xlmr-base-arq-noaug

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: 0.5799
• F1: 0.5159
• Roc Auc: 0.6575
• Accuracy: 0.26

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.6724	1.0	29	0.6041	0.0	0.5	0.12
0.6216	2.0	58	0.5785	0.0	0.5	0.12
0.5853	3.0	87	0.5688	0.0081	0.5021	0.12
0.5778	4.0	116	0.5585	0.1240	0.5286	0.14
0.5443	5.0	145	0.5399	0.2358	0.5557	0.15
0.4872	6.0	174	0.5496	0.3667	0.5942	0.16
0.4641	7.0	203	0.5601	0.3801	0.5922	0.18
0.45	8.0	232	0.5493	0.3595	0.5927	0.2
0.4081	9.0	261	0.5552	0.4353	0.6152	0.21
0.3877	10.0	290	0.5468	0.4512	0.6210	0.21
0.3661	11.0	319	0.5692	0.4819	0.6364	0.21
0.336	12.0	348	0.5657	0.4960	0.6484	0.24
0.3116	13.0	377	0.5773	0.4949	0.6444	0.24
0.3004	14.0	406	0.5799	0.5159	0.6575	0.26
0.2778	15.0	435	0.5897	0.4973	0.6478	0.24
0.2807	16.0	464	0.5928	0.4976	0.6460	0.22
0.2715	17.0	493	0.5933	0.4951	0.6437	0.23
0.2692	18.0	522	0.5952	0.4905	0.6404	0.23

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0