afro-xlmr-base-arq-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: 0.5136
• F1: 0.5725
• Roc Auc: 0.6987
• Accuracy: 0.2957

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
0.6493	1.0	58	0.5823	0.0	0.4993	0.1087
0.5943	2.0	116	0.5618	0.1307	0.5245	0.1130
0.5728	3.0	174	0.5463	0.2063	0.5497	0.1348
0.5326	4.0	232	0.5335	0.2462	0.5581	0.1348
0.4913	5.0	290	0.5259	0.3849	0.6084	0.1652
0.4505	6.0	348	0.5067	0.4337	0.6319	0.1826
0.3857	7.0	406	0.5034	0.5164	0.6635	0.2
0.3759	8.0	464	0.5013	0.4906	0.6597	0.2
0.3219	9.0	522	0.5048	0.5114	0.6624	0.2087
0.2938	10.0	580	0.5037	0.5247	0.6744	0.2478
0.2736	11.0	638	0.5054	0.5363	0.6788	0.2478
0.2491	12.0	696	0.5079	0.5447	0.6863	0.2478
0.2412	13.0	754	0.5149	0.5502	0.6857	0.2609
0.2071	14.0	812	0.5159	0.5617	0.6905	0.2739
0.2084	15.0	870	0.5196	0.5573	0.6893	0.2609
0.1965	16.0	928	0.5136	0.5725	0.6987	0.2957
0.185	17.0	986	0.5141	0.5663	0.6924	0.2957
0.188	18.0	1044	0.5156	0.5651	0.6913	0.2783
0.1932	19.0	1102	0.5163	0.5640	0.6917	0.2826
0.184	20.0	1160	0.5165	0.5655	0.6928	0.2826

Framework versions

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