afro-xlmr-base-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: 0.3494
• F1: 0.7257
• Roc Auc: 0.8196
• Accuracy: 0.5305

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: 30

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.4381	1.0	144	0.4182	0.1435	0.5480	0.2164
0.3725	2.0	288	0.3353	0.5562	0.7123	0.4258
0.304	3.0	432	0.3059	0.6420	0.7561	0.4834
0.2326	4.0	576	0.3015	0.6706	0.7808	0.5131
0.2242	5.0	720	0.3015	0.6902	0.7942	0.5009
0.1752	6.0	864	0.3019	0.7017	0.8020	0.5166
0.1378	7.0	1008	0.2968	0.7029	0.8012	0.5323
0.1141	8.0	1152	0.3105	0.7012	0.8031	0.5218
0.0951	9.0	1296	0.3298	0.7112	0.8125	0.5201
0.085	10.0	1440	0.3288	0.6955	0.7956	0.5323
0.0693	11.0	1584	0.3553	0.7029	0.8108	0.5201
0.0579	12.0	1728	0.3494	0.7257	0.8196	0.5305
0.045	13.0	1872	0.3637	0.7165	0.8155	0.5288
0.0395	14.0	2016	0.3993	0.7177	0.8205	0.5201
0.0383	15.0	2160	0.4184	0.7099	0.8129	0.5113
0.0275	16.0	2304	0.4123	0.7151	0.8170	0.5288

Framework versions

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