fd934aa48c8b39ff63635ebfbbd95a47

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

• Loss: 2.1613
• Data Size: 1.0
• Epoch Runtime: 221.0547
• Bleu: 12.6073

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	5.4238	0	18.6349	0.8695
No log	1	872	4.0393	0.0078	20.2859	1.7117
No log	2	1744	3.3450	0.0156	23.2945	2.4262
0.0599	3	2616	2.9601	0.0312	27.6422	3.4170
0.1882	4	3488	2.6471	0.0625	33.9976	4.3086
2.5511	5	4360	2.3492	0.125	47.2096	5.4485
2.1813	6	5232	2.0884	0.25	72.9175	6.5252
1.7976	7	6104	1.8851	0.5	123.6483	7.8993
1.5706	8.0	6976	1.7057	1.0	222.0246	16.2860
1.26	9.0	7848	1.6795	1.0	219.2998	11.9915
1.0386	10.0	8720	1.7369	1.0	220.4786	11.3533
0.8437	11.0	9592	1.8282	1.0	220.6066	8.9669
0.6384	12.0	10464	1.9799	1.0	221.6479	10.4202
0.5059	13.0	11336	2.1613	1.0	221.0547	12.6073

Framework versions

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