BARTBana_Translation_Finetune_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: 0.4295
• Sacrebleu: 7.5050

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: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Sacrebleu
0.6831	1.0	468	0.5818	3.2556
0.566	2.0	936	0.5188	4.6548
0.5127	3.0	1404	0.4878	5.3508
0.4804	4.0	1872	0.4683	5.8657
0.4558	5.0	2340	0.4551	6.2975
0.433	6.0	2808	0.4450	6.4311
0.4146	7.0	3276	0.4420	6.7296
0.3969	8.0	3744	0.4365	6.9791
0.3911	9.0	4212	0.4332	7.1487
0.3742	10.0	4680	0.4302	7.2803
0.3686	11.0	5148	0.4292	7.3851
0.3568	12.0	5616	0.4296	7.4003
0.3505	13.0	6084	0.4292	7.4202
0.3503	14.0	6552	0.4289	7.4984
0.3453	15.0	7020	0.4295	7.5050

Framework versions

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