BaViT_Large_Finetune_v0

This model is a fine-tuned version of VietAI/vit5-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4909
• Sacrebleu: 11.1985

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: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Sacrebleu
0.4969	1.0	2922	0.4594	6.2638
0.4322	2.0	5844	0.4101	8.4455
0.3585	3.0	8766	0.3905	9.4839
0.315	4.0	11688	0.3867	10.0488
0.2831	5.0	14610	0.3877	10.4073
0.2541	6.0	17532	0.3939	10.5730
0.2136	7.0	20454	0.4032	10.8551
0.1985	8.0	23376	0.4167	10.8040
0.1775	9.0	26298	0.4268	10.8393
0.1597	10.0	29220	0.4394	10.8971
0.143	11.0	32142	0.4533	10.9369
0.1396	12.0	35064	0.4673	11.0112
0.1258	13.0	37986	0.4771	11.1620
0.1178	14.0	40908	0.4849	11.1513
0.1109	15.0	43830	0.4909	11.1985

Framework versions

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