03c4322dd20d14fd1fd1e6ae6947b0ed

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

• Loss: 2.3776
• Data Size: 1.0
• Epoch Runtime: 204.1863
• Bleu: 8.9750

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	8.5467	0	17.0788	0.3018
No log	1	806	4.9038	0.0078	19.8966	3.1786
No log	2	1612	4.3759	0.0156	21.8352	4.5213
No log	3	2418	3.1960	0.0312	24.9157	4.2811
0.1246	4	3224	2.6798	0.0625	30.6225	4.5840
11.2646	5	4030	7.1054	0.125	42.0271	0.0008
2.7751	6	4836	2.3923	0.25	65.4964	7.7904
2.1083	7	5642	2.1171	0.5	112.7058	9.7122
1.8501	8.0	6448	1.9743	1.0	203.5311	8.7201
1.5457	9.0	7254	1.9077	1.0	201.6626	8.9252
1.2736	10.0	8060	1.9443	1.0	204.1982	10.1011
1.055	11.0	8866	2.0638	1.0	204.1636	9.3755
0.8399	12.0	9672	2.1922	1.0	203.4054	8.5365
0.6433	13.0	10478	2.3776	1.0	204.1863	8.9750

Framework versions

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