BARTBana_Translation_v51

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: 0.4505
• Sacrebleu: 11.7970

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
0.6917	1.0	742	0.5995	6.7000
0.6008	2.0	1484	0.5330	8.7480
0.5208	3.0	2226	0.4989	9.7564
0.4962	4.0	2968	0.4808	10.3233
0.46	5.0	3710	0.4711	10.7276
0.4428	6.0	4452	0.4615	10.9196
0.4125	7.0	5194	0.4566	11.2077
0.3955	8.0	5936	0.4515	11.3811
0.3856	9.0	6678	0.4496	11.5736
0.3687	10.0	7420	0.4482	11.5727
0.3553	11.0	8162	0.4496	11.6036
0.3467	12.0	8904	0.4508	11.6907
0.3386	13.0	9646	0.4505	11.7970
0.3324	14.0	10388	0.4510	11.7524
0.3245	15.0	11130	0.4528	11.7921

Framework versions

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