0105fe24732a50b97855fbe14e22ea2c

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

• Loss: 2.1048
• Data Size: 1.0
• Epoch Runtime: 220.8327
• Bleu: 8.4820

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	5.0778	0	18.4198	0.7817
No log	1	872	2.8099	0.0078	20.0522	2.9635
No log	2	1744	2.5155	0.0156	23.2213	3.9756
0.0443	3	2616	2.3422	0.0312	28.0736	4.7776
0.1489	4	3488	2.1943	0.0625	33.4731	5.8386
2.1531	5	4360	2.0525	0.125	45.6730	6.8558
1.931	6	5232	1.9039	0.25	72.8835	9.1409
1.664	7	6104	1.7619	0.5	121.2179	9.7207
1.4828	8.0	6976	1.6508	1.0	221.1274	8.6549
1.2051	9.0	7848	1.6374	1.0	219.9815	7.8702
0.991	10.0	8720	1.7239	1.0	220.9367	8.6925
0.801	11.0	9592	1.8398	1.0	219.3326	10.2962
0.6091	12.0	10464	1.9613	1.0	220.6576	8.4806
0.4868	13.0	11336	2.1048	1.0	220.8327	8.4820

Framework versions

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