22e8edafce68bc5399381c9b6ef9ee29

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

• Loss: 2.3631
• Data Size: 1.0
• Epoch Runtime: 206.2697
• Bleu: 7.3005

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	3.9893	0	17.4741	9.1950
No log	1	808	3.0547	0.0078	19.2018	5.3794
No log	2	1616	2.7553	0.0156	21.1794	6.0326
No log	3	2424	2.4236	0.0312	24.9660	8.9770
0.0787	4	3232	2.3228	0.0625	31.2339	12.2186
2.3082	5	4040	2.2259	0.125	43.7439	16.4083
2.0997	6	4848	2.1222	0.25	67.5827	10.5571
1.877	7	5656	2.0145	0.5	113.8543	9.0539
1.7437	8.0	6464	1.9279	1.0	207.2858	8.7008
1.3908	9.0	7272	1.9469	1.0	206.6341	9.7038
1.1273	10.0	8080	2.0514	1.0	206.1858	7.7756
0.9145	11.0	8888	2.1763	1.0	205.2806	7.3002
0.6943	12.0	9696	2.3631	1.0	206.2697	7.3005

Framework versions

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