f8d26c1cb55baf088e4168c1cb48a569

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

• Loss: 1.9690
• Data Size: 1.0
• Epoch Runtime: 117.3079
• Bleu: 11.9628

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	9.9740	0	10.3733	0.2238
No log	1	434	6.6899	0.0078	11.5288	0.5063
No log	2	868	3.6586	0.0156	13.5108	2.0945
No log	3	1302	2.6957	0.0312	15.7120	4.2052
No log	4	1736	2.3330	0.0625	18.6447	5.4895
0.1287	5	2170	2.1177	0.125	26.6666	6.3649
2.0813	6	2604	1.9724	0.25	39.4061	7.1694
1.8452	7	3038	1.8180	0.5	65.3488	10.0895
1.5981	8.0	3472	1.7214	1.0	117.8719	9.4866
1.2642	9.0	3906	1.7066	1.0	116.8944	9.2617
1.3141	10.0	4340	1.7610	1.0	116.5585	14.7399
0.8267	11.0	4774	1.8069	1.0	117.2968	11.8996
0.6527	12.0	5208	1.8896	1.0	117.4821	12.3884
0.5034	13.0	5642	1.9690	1.0	117.3079	11.9628

Framework versions

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