BaViT5_v01

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

• Loss: 0.4623
• Sacrebleu: 14.3803

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
0.6515	1.0	2966	0.5899	7.7723
0.576	2.0	5932	0.5257	10.4904
0.4939	3.0	8898	0.4969	11.8064
0.4842	4.0	11864	0.4793	12.5193
0.4459	5.0	14830	0.4704	12.9876
0.4222	6.0	17796	0.4632	13.2632
0.4005	7.0	20762	0.4612	13.5868
0.3869	8.0	23728	0.4580	13.8162
0.381	9.0	26694	0.4556	13.9756
0.3594	10.0	29660	0.4561	14.0827
0.363	11.0	32626	0.4578	14.1701
0.3427	12.0	35592	0.4591	14.2903
0.3425	13.0	38558	0.4603	14.3091
0.3377	14.0	41524	0.4611	14.3649
0.314	15.0	44490	0.4623	14.3803

Framework versions

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