STL-afroxlmr-base-stance

This model is a fine-tuned version of Davlan/afro-xlmr-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.9352
• Stance Precision: 0.7596
• Stance Recall: 0.7595
• F1: 0.7592
• Stance Acc: 0.7506

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 300
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Stance Precision	Stance Recall	F1	Stance Acc
0.8876	1.0	377	0.7709	0.7071	0.7115	0.7080	0.6995
0.6480	2.0	754	0.6859	0.7443	0.7520	0.7469	0.7406
0.5776	3.0	1131	0.6810	0.7553	0.7518	0.7518	0.7444
0.4660	4.0	1508	0.7406	0.7652	0.7461	0.7515	0.7419
0.3414	5.0	1885	0.7764	0.7514	0.7669	0.7560	0.7506
0.3206	6.0	2262	0.8319	0.7585	0.7625	0.7599	0.7519
0.2374	7.0	2639	0.8981	0.7601	0.7527	0.7549	0.7456
0.2484	8.0	3016	0.9228	0.7512	0.7522	0.7512	0.7431
0.1907	9.0	3393	0.9352	0.7596	0.7595	0.7592	0.7506

Framework versions

• Transformers 5.15.1
• Pytorch 2.11.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2