7f3d02cc1ca68d7a3fab0eaf07a2fe3c

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

• Loss: 3.8499
• Data Size: 1.0
• Epoch Runtime: 13.6951
• Bleu: 7.1182

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	14.1474	0	1.5749	0.0854
No log	1	27	9.1364	0.0078	2.1336	0.7866
No log	2	54	8.3005	0.0156	3.0551	0.3817
No log	3	81	7.6918	0.0312	4.1676	0.3472
No log	4	108	7.1488	0.0625	5.2954	0.6329
No log	5	135	6.5539	0.125	6.7583	0.8528
No log	6	162	5.9818	0.25	7.9090	1.2757
No log	7	189	5.1589	0.5	10.1384	2.0743
1.2979	8.0	216	4.0883	1.0	12.7502	3.6540
1.2979	9.0	243	6.3313	1.0	12.0525	2.9182
5.5738	10.0	270	14.3975	1.0	13.1673	0.0088
5.5738	11.0	297	4.3063	1.0	14.4976	3.4813
12.6142	12.0	324	3.5081	1.0	15.9867	4.6739
2.7996	13.0	351	3.3913	1.0	12.6326	5.5771
2.7996	14.0	378	3.4230	1.0	12.8004	6.0400
1.796	15.0	405	3.9428	1.0	13.0838	5.9026
1.796	16.0	432	3.8249	1.0	13.8742	6.2388
1.6499	17.0	459	3.8499	1.0	13.6951	7.1182

Framework versions

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