Noisy-deberta-v3-small-Label_B-768-epochs-9

This model is a fine-tuned version of microsoft/deberta-v3-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0653
• Accuracy: 0.9855
• F1: 0.9855
• Precision: 0.9857
• Recall: 0.9855

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: 5e-05
• train_batch_size: 24
• eval_batch_size: 24
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 96
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 9

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.1494	0.9995	533	0.1549	0.9532	0.9530	0.9557	0.9532
0.091	1.9991	1066	0.1355	0.9599	0.9598	0.9637	0.9599
0.0506	2.9986	1599	0.0560	0.9850	0.9850	0.9851	0.9850
0.0395	4.0	2133	0.0653	0.9855	0.9855	0.9857	0.9855
0.0485	4.9995	2666	0.1011	0.9828	0.9828	0.9832	0.9828
0.0087	5.9991	3199	0.1476	0.9793	0.9792	0.9797	0.9793
0.0122	6.9986	3732	0.1278	0.9826	0.9826	0.9829	0.9826
0.0014	8.0	4266	0.1800	0.9780	0.9780	0.9787	0.9780
0.0018	8.9958	4797	0.1545	0.9810	0.9810	0.9815	0.9810

Framework versions

• Transformers 4.46.3
• Pytorch 2.4.0
• Datasets 3.1.0
• Tokenizers 0.20.3