afro-xlmr-base-amh-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: 0.3598
• F1: 0.7180
• Roc Auc: 0.8216
• Accuracy: 0.5711

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
0.3732	1.0	238	0.3390	0.3871	0.6361	0.3741
0.3078	2.0	476	0.2957	0.4993	0.6893	0.4573
0.255	3.0	714	0.3038	0.4991	0.6918	0.4489
0.2146	4.0	952	0.2743	0.5751	0.7355	0.4889
0.1812	5.0	1190	0.2834	0.5907	0.7533	0.4932
0.1566	6.0	1428	0.2816	0.6564	0.7837	0.5153
0.1454	7.0	1666	0.2748	0.6717	0.7939	0.5427
0.1151	8.0	1904	0.2930	0.6693	0.8009	0.5469
0.0807	9.0	2142	0.3085	0.6799	0.7997	0.5458
0.0643	10.0	2380	0.3011	0.6978	0.8078	0.5574
0.0626	11.0	2618	0.3296	0.6945	0.8138	0.5522
0.0461	12.0	2856	0.3366	0.6896	0.8001	0.5564
0.0342	13.0	3094	0.3503	0.6893	0.8178	0.5522
0.0301	14.0	3332	0.3453	0.7036	0.8136	0.5669
0.0209	15.0	3570	0.3575	0.7135	0.8176	0.5680
0.0171	16.0	3808	0.3632	0.7042	0.8158	0.5616
0.017	17.0	4046	0.3598	0.7180	0.8216	0.5711
0.0223	18.0	4284	0.3610	0.7065	0.8170	0.5701
0.0207	19.0	4522	0.3622	0.7153	0.8212	0.5680
0.0179	20.0	4760	0.3629	0.7117	0.8213	0.5669

Framework versions

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