BARTBana_Translation_Replace_v0

This model is a fine-tuned version of IAmSkyDra/BARTBana_v4 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0029
• Sacrebleu: 20.9762

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: 100
• eval_batch_size: 100
• 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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Sacrebleu
0.0673	1.0	5881	0.0335	17.7815
0.0401	2.0	11762	0.0124	19.8597
0.0218	3.0	17643	0.0067	20.5145
0.0141	4.0	23524	0.0046	20.7775
0.0096	5.0	29405	0.0039	20.8673
0.007	6.0	35286	0.0035	20.9294
0.0051	7.0	41167	0.0032	20.9362
0.0043	8.0	47048	0.0031	20.9635
0.0032	9.0	52929	0.0030	20.9887
0.0028	10.0	58810	0.0029	20.9762

Framework versions

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