BARTBana_Translation_v2

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

• Loss: 0.4520
• Sacrebleu: 11.7352

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
0.695	1.0	742	0.6021	6.3321
0.5976	2.0	1484	0.5291	8.6429
0.5171	3.0	2226	0.4958	9.7101
0.4919	4.0	2968	0.4781	10.3323
0.4556	5.0	3710	0.4680	10.7812
0.4387	6.0	4452	0.4577	10.8965
0.4095	7.0	5194	0.4538	11.1963
0.3924	8.0	5936	0.4499	11.2119
0.3815	9.0	6678	0.4486	11.4155
0.3647	10.0	7420	0.4468	11.4443
0.3525	11.0	8162	0.4479	11.5941
0.3435	12.0	8904	0.4489	11.5933
0.3349	13.0	9646	0.4500	11.7211
0.3289	14.0	10388	0.4508	11.7113
0.3202	15.0	11130	0.4520	11.7352

Framework versions

• Transformers 4.48.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0