5d74adf4697a2c3be40e651a99176d87

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

• Loss: 1.9312
• Data Size: 1.0
• Epoch Runtime: 219.8573
• Bleu: 9.3037

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	4.9898	0	18.6756	1.5284
No log	1	872	2.2949	0.0078	20.4799	5.2908
No log	2	1744	2.1446	0.0156	23.5651	5.8491
0.0373	3	2616	2.0362	0.0312	28.0025	6.4647
0.1298	4	3488	1.9498	0.0625	33.8992	7.1144
1.926	5	4360	1.8543	0.125	47.0578	10.1709
1.7713	6	5232	1.7587	0.25	71.8204	8.8269
1.5418	7	6104	1.6585	0.5	121.6652	11.8798
1.3874	8.0	6976	1.5874	1.0	220.7199	10.2576
1.11	9.0	7848	1.5874	1.0	219.4134	9.5841
0.9061	10.0	8720	1.6921	1.0	219.6200	9.9177
0.7159	11.0	9592	1.7954	1.0	223.0220	9.5128
0.5291	12.0	10464	1.9312	1.0	219.8573	9.3037

Framework versions

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