HCMUS-vietnamese-correction

This model is a fine-tuned version of vinai/bartpho-syllable on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0225
• Bleu: 89.2375
• Cer: 0.0392
• Wer: 0.0749
• F1: 0.9744

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: 8
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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: linear
• num_epochs: 7
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Cer	Wer	F1
0.8639	1.0	573	0.0500	77.4679	0.0732	0.1532	0.9343
0.0666	2.0	1146	0.0325	84.8950	0.0513	0.1020	0.9600
0.0425	3.0	1719	0.0271	87.0135	0.0448	0.0893	0.9676
0.032	4.0	2292	0.0244	88.0650	0.0421	0.0825	0.9707
0.026	5.0	2865	0.0230	88.7480	0.0405	0.0781	0.9724
0.0217	6.0	3438	0.0224	89.0337	0.0394	0.0757	0.9746
0.0169	7.0	4011	0.0225	89.2375	0.0392	0.0749	0.9744

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1