afro-xlmr-large-76L-vmw-swap-delete-finetuned-30-epochs

This model is a fine-tuned version of Davlan/afro-xlmr-large-76L on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4054
• F1: 0.4092
• Roc Auc: 0.6439
• Accuracy: 0.5449

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: Use OptimizerNames.ADAMW_TORCH 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: 100
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.3177	1.0	89	0.3209	0.0	0.5	0.4354
0.2968	2.0	178	0.3099	0.0	0.5	0.4354
0.2886	3.0	267	0.3029	0.0145	0.5038	0.4438
0.2625	4.0	356	0.2947	0.0824	0.5227	0.4438
0.2367	5.0	445	0.2933	0.1385	0.5395	0.4719
0.1819	6.0	534	0.3031	0.1951	0.5597	0.4691
0.1498	7.0	623	0.3104	0.2790	0.5936	0.5028
0.1189	8.0	712	0.3306	0.2798	0.5981	0.5169
0.0821	9.0	801	0.3400	0.3019	0.6007	0.4944
0.0615	10.0	890	0.3459	0.3624	0.6310	0.5084
0.0455	11.0	979	0.3633	0.3599	0.6290	0.5169
0.0318	12.0	1068	0.3712	0.3783	0.6312	0.5449
0.0279	13.0	1157	0.3923	0.3502	0.6184	0.5281
0.0207	14.0	1246	0.3883	0.3753	0.6309	0.5365
0.0148	15.0	1335	0.4054	0.4092	0.6439	0.5449
0.0153	16.0	1424	0.4125	0.3827	0.6320	0.5281
0.0097	17.0	1513	0.4258	0.4011	0.6356	0.5365
0.0065	18.0	1602	0.4407	0.3792	0.6298	0.5197
0.0064	19.0	1691	0.4324	0.4054	0.6370	0.5534
0.0049	20.0	1780	0.4541	0.3897	0.6332	0.5449
0.0043	21.0	1869	0.4587	0.3539	0.6177	0.5337

Framework versions

• Transformers 4.48.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0