huyhuyvu01/DeBERTa_large_chartering_email_cls

BBC_CLS_deberta_v3_large_v2

This model is a fine-tuned version of microsoft/deberta-v3-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0857
• Accuracy: 0.9866
• Precision: 0.9723
• Recall: 0.9780
• F1: 0.9751

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
1.235	1.0	66	0.6331	0.7964	0.4047	0.4873	0.4418
0.4336	2.0	132	0.2201	0.8971	0.6754	0.7091	0.6910
0.2133	3.0	198	0.0990	0.9776	0.9476	0.9786	0.9602
0.1083	4.0	264	0.1038	0.9821	0.9656	0.9651	0.9653
0.0848	5.0	330	0.0907	0.9866	0.9782	0.9714	0.9747
0.1087	6.0	396	0.1270	0.9799	0.9672	0.9689	0.9671
0.1011	7.0	462	0.1289	0.9754	0.9677	0.9660	0.9667
0.0827	8.0	528	0.0990	0.9799	0.9818	0.9479	0.9632
0.0621	9.0	594	0.0857	0.9866	0.9723	0.9780	0.9751
0.0444	10.0	660	0.1071	0.9843	0.9769	0.9663	0.9715

Framework versions

• Transformers 4.35.0.dev0
• Pytorch 1.13.1
• Datasets 2.13.0
• Tokenizers 0.14.1