4ebe552990688d8cc44d90d9ff3c6acb

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

• Loss: 1.8906
• Data Size: 1.0
• Epoch Runtime: 119.6279
• Bleu: 13.0280

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	11.4284	0	10.4456	0.2222
No log	1	447	3.9163	0.0078	12.6083	4.6158
0.0873	2	894	3.5194	0.0156	13.3579	6.6736
0.0829	3	1341	3.0251	0.0312	15.5111	7.1696
0.1249	4	1788	2.3728	0.0625	18.5220	7.0497
0.1841	5	2235	1.9471	0.125	26.7345	7.9220
3.3433	6	2682	1.8593	0.25	39.7884	8.8727
1.6819	7	3129	1.6328	0.5	66.6799	12.2916
1.3654	8.0	3576	1.5305	1.0	117.9939	14.6430
1.0809	9.0	4023	1.4889	1.0	117.0876	12.4964
0.8272	10.0	4470	1.5304	1.0	119.6083	12.5766
0.6012	11.0	4917	1.6533	1.0	118.6062	12.3317
0.4656	12.0	5364	1.7921	1.0	118.2992	14.0522
0.3406	13.0	5811	1.8906	1.0	119.6279	13.0280

Framework versions

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