d128c24e05d175d269e03155e4ea4529

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [it-sv] dataset. It achieves the following results on the evaluation set:

• Loss: 4.5640
• Data Size: 1.0
• Epoch Runtime: 22.5104
• Bleu: 5.3658

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	14.6706	0	2.6366	0.0857
No log	1	74	10.8074	0.0078	3.1775	0.2320
No log	2	148	8.2347	0.0156	4.5450	0.1227
0.3439	3	222	7.6195	0.0312	5.7044	0.2196
0.3439	4	296	7.1687	0.0625	7.9862	0.2007
0.4998	5	370	6.6233	0.125	9.2208	0.2233
0.4998	6	444	6.1467	0.25	11.9886	0.3066
1.3502	7	518	5.4764	0.5	15.0848	0.9251
3.488	8.0	592	4.6302	1.0	25.2315	1.6674
4.1843	9.0	666	4.3235	1.0	24.5005	2.4872
3.7551	10.0	740	4.3014	1.0	23.1926	2.7223
14.3267	11.0	814	7.1968	1.0	23.4077	0.1724
7.0015	12.0	888	4.1839	1.0	23.9296	4.4242
2.966	13.0	962	3.9380	1.0	23.7268	6.1310
2.5691	14.0	1036	3.9823	1.0	24.0302	6.3044
1.9563	15.0	1110	4.1305	1.0	24.3097	7.0976
1.6206	16.0	1184	4.3650	1.0	24.7539	5.5460
1.2324	17.0	1258	4.5640	1.0	22.5104	5.3658

Framework versions

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