023d2b2e3bf369246547928ce4a970c8

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [de-nl] dataset. It achieves the following results on the evaluation set:

• Loss: 7.0125
• Data Size: 1.0
• Epoch Runtime: 101.7615
• Bleu: 0.2310

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	10.3040	0	9.3202	0.1035
No log	1	390	4.1664	0.0078	10.3157	2.5086
No log	2	780	3.9159	0.0156	11.4039	3.8475
No log	3	1170	3.5317	0.0312	14.1013	4.7239
No log	4	1560	3.0899	0.0625	17.2870	5.1470
0.2026	5	1950	2.7887	0.125	23.1371	5.2966
0.4366	6	2340	2.5578	0.25	34.1756	6.2572
4.468	7	2730	2.5387	0.5	56.7524	9.7045
2.269	8.0	3120	2.2165	1.0	102.4147	9.8848
1.8526	9.0	3510	2.1746	1.0	102.7382	13.4595
1.575	10.0	3900	2.1921	1.0	102.7911	11.4555
1.3235	11.0	4290	2.2693	1.0	103.3988	13.6100
1.0706	12.0	4680	2.3782	1.0	101.7039	11.3952
6.5681	13.0	5070	7.0125	1.0	101.7615	0.2310

Framework versions

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