07d38e6db51d22b45668dfc750fc451a

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

• Loss: 2.4027
• Data Size: 1.0
• Epoch Runtime: 206.1746
• Bleu: 6.5197

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	7.4178	0	17.3964	0.5153
No log	1	808	4.7268	0.0078	19.1863	2.7702
No log	2	1616	3.9526	0.0156	21.1123	3.0918
No log	3	2424	3.1249	0.0312	24.9135	5.8358
0.1108	4	3232	15.1188	0.0625	31.9091	0.1210
9.8362	5	4040	6.8800	0.125	43.3509	0.0028
2.5215	6	4848	2.3426	0.25	66.6294	11.5187
2.0946	7	5656	2.1343	0.5	114.0027	6.1022
1.8789	8.0	6464	1.9672	1.0	210.5752	6.6744
1.511	9.0	7272	1.9822	1.0	213.4882	6.9122
1.2266	10.0	8080	2.0686	1.0	204.4838	6.8228
1.0071	11.0	8888	2.2011	1.0	207.8282	6.6827
0.7724	12.0	9696	2.4027	1.0	206.1746	6.5197

Framework versions

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