7ce224400f704b08290d41e7e801e989

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

• Loss: 3.5240
• Data Size: 1.0
• Epoch Runtime: 20.6079
• Bleu: 4.2992

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	6.9780	0	1.9714	0.2428
No log	1	58	6.4824	0.0078	3.1328	0.0968
No log	2	116	6.2821	0.0156	2.9656	0.1340
No log	3	174	5.8348	0.0312	4.9359	0.3038
No log	4	232	5.3368	0.0625	6.2168	0.4847
No log	5	290	4.6974	0.125	8.3314	0.8709
0.4511	6	348	4.2036	0.25	10.1632	1.4197
0.5902	7	406	3.7409	0.5	13.8419	2.3581
2.5447	8.0	464	3.3664	1.0	20.3870	3.1136
2.9308	9.0	522	3.2001	1.0	20.4593	3.7319
2.5123	10.0	580	3.1377	1.0	20.1334	3.9054
2.2199	11.0	638	3.1965	1.0	19.7409	4.3245
1.9263	12.0	696	3.2985	1.0	19.6490	4.1539
1.4139	13.0	754	3.4151	1.0	20.4135	4.5297
1.1894	14.0	812	3.5240	1.0	20.6079	4.2992

Framework versions

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