BARTBana_Translation_v50

This model is a fine-tuned version of IAmSkyDra/BARTBana_v5 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4529
• Sacrebleu: 11.3538

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: 64
• eval_batch_size: 64
• 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
0.7129	1.0	742	0.6236	4.9074
0.6174	2.0	1484	0.5494	7.5076
0.5369	3.0	2226	0.5121	8.8742
0.5094	4.0	2968	0.4908	9.6042
0.4722	5.0	3710	0.4776	10.1784
0.4548	6.0	4452	0.4667	10.4266
0.4253	7.0	5194	0.4624	10.7054
0.408	8.0	5936	0.4566	10.8654
0.3979	9.0	6678	0.4518	11.0410
0.381	10.0	7420	0.4508	11.1446
0.3689	11.0	8162	0.4508	11.1509
0.3595	12.0	8904	0.4524	11.2866
0.3511	13.0	9646	0.4513	11.3042
0.3451	14.0	10388	0.4519	11.3478
0.3367	15.0	11130	0.4529	11.3538

Framework versions

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