mBART_tokenizer_custom

This model is a fine-tuned version of facebook/mbart-large-50 on the 한국어 방언 발화(경상도) dataset from AI Hub. It achieves the following results on the evaluation set:

• Loss: 0.0045
• Bleu: 78.9009
• Gen Len: 20.5578

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
0.048	1.0	12500	0.0259	93.2721	22.2035
0.0251	2.0	25000	0.0106	87.1169	21.7826
0.0105	3.0	37500	0.0045	78.9009	20.5578

Framework versions

• Transformers 4.22.2
• Pytorch 1.12.1+cu113
• Datasets 2.5.1
• Tokenizers 0.12.1