vi-correction-model

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: 0.0561
• Sacrebleu: 22.9297

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: 12
• eval_batch_size: 12
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 48
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Sacrebleu
No log	1.0	104	0.0988	17.3540
No log	2.0	208	0.0661	21.5130
No log	3.0	312	0.0593	22.4272
No log	4.0	416	0.0564	22.8821
No log	5.0	520	0.0561	22.9297

Framework versions

• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1