afro-xlmr-base-ptmz-MICRO

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: 0.4081
• F1: 0.5444
• Roc Auc: 0.7464
• Accuracy: 0.6143

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
0.2142	1.0	404	0.2462	0.2719	0.5799	0.5465
0.1698	2.0	808	0.2159	0.4521	0.6602	0.6085
0.0961	3.0	1212	0.2625	0.5019	0.7228	0.5833
0.0722	4.0	1616	0.2703	0.5072	0.7095	0.6182
0.0321	5.0	2020	0.2979	0.5040	0.7103	0.6143
0.016	6.0	2424	0.3259	0.5389	0.7293	0.6434
0.0231	7.0	2828	0.3676	0.4939	0.7046	0.6163
0.0102	8.0	3232	0.3594	0.5349	0.7327	0.6240
0.0041	9.0	3636	0.4071	0.5146	0.7212	0.6085
0.0032	10.0	4040	0.4081	0.5444	0.7464	0.6143
0.002	11.0	4444	0.4213	0.5142	0.7146	0.6202
0.0016	12.0	4848	0.4291	0.5246	0.7177	0.6357
0.0013	13.0	5252	0.4394	0.5325	0.7176	0.6376
0.0026	14.0	5656	0.4440	0.5347	0.7286	0.6357

Framework versions

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