BARTBana_Translation_v01

This model is a fine-tuned version of vinai/bartpho-syllable on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4506
• Sacrebleu: 11.0518

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: 64
• eval_batch_size: 64
• 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: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Sacrebleu
0.7912	1.0	742	0.6747	2.7391
0.6623	2.0	1484	0.5821	5.7331
0.5662	3.0	2226	0.5306	7.8025
0.5337	4.0	2968	0.5027	8.6825
0.4913	5.0	3710	0.4858	9.4253
0.4711	6.0	4452	0.4725	9.8687
0.4399	7.0	5194	0.4638	10.2200
0.4208	8.0	5936	0.4584	10.3954
0.4096	9.0	6678	0.4533	10.7019
0.3924	10.0	7420	0.4509	10.8096
0.3796	11.0	8162	0.4499	10.7237
0.3704	12.0	8904	0.4502	10.9647
0.3618	13.0	9646	0.4494	10.9840
0.3556	14.0	10388	0.4496	11.0433
0.3476	15.0	11130	0.4506	11.0518

Framework versions

• Transformers 4.48.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0