16c088bd50f60dc3d8b58cf6db4d8e97

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

• Loss: 2.1585
• Data Size: 1.0
• Epoch Runtime: 111.6070
• Bleu: 8.2384

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	5.9554	0	9.4906	0.4722
No log	1	437	4.6232	0.0078	10.6367	1.1591
No log	2	874	3.8821	0.0156	12.2133	2.2390
No log	3	1311	3.4263	0.0312	15.0625	3.3700
No log	4	1748	3.0455	0.0625	19.3313	4.5330
2.8613	5	2185	2.7022	0.125	26.0513	5.3314
2.5206	6	2622	2.3944	0.25	37.4514	6.4168
2.1287	7	3059	2.1496	0.5	62.9429	7.2491
1.6994	8.0	3496	1.9230	1.0	112.8877	8.4187
1.3225	9.0	3933	1.9081	1.0	112.4188	8.8377
0.9777	10.0	4370	1.9337	1.0	110.6205	9.0018
0.7374	11.0	4807	1.9846	1.0	112.6945	8.6039
0.5269	12.0	5244	2.0883	1.0	111.7913	8.2357
0.3633	13.0	5681	2.1585	1.0	111.6070	8.2384

Framework versions

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