95acc3bd640ec07904f0a56a90b37add

This model is a fine-tuned version of facebook/mbart-large-50 on the Helsinki-NLP/opus_books [de-nl] dataset. It achieves the following results on the evaluation set:

• Loss: 2.5353
• Data Size: 1.0
• Epoch Runtime: 98.4929
• Bleu: 9.7760

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Bleu
No log	0	0	8.0922	0	8.4855	0.5464
No log	1	390	5.2927	0.0078	9.9542	3.2771
No log	2	780	4.9121	0.0156	10.5833	4.5891
No log	3	1170	3.8158	0.0312	13.6072	6.2612
No log	4	1560	3.1217	0.0625	16.5894	5.9088
0.2094	5	1950	2.6141	0.125	22.2090	6.0787
1.5489	6	2340	6.5837	0.25	33.0776	0.0145
2.4607	7	2730	2.3101	0.5	54.5691	15.9157
2.0185	8.0	3120	2.1011	1.0	99.2213	18.2214
1.6148	9.0	3510	2.1404	1.0	98.1079	15.1043
1.2872	10.0	3900	2.2005	1.0	97.4431	13.7137
0.9797	11.0	4290	2.3714	1.0	97.4517	9.9202
0.7148	12.0	4680	2.5353	1.0	98.4929	9.7760

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1