6a778a7a32b20788510307cd3a0e19df

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

• Loss: 2.0635
• Data Size: 1.0
• Epoch Runtime: 111.3556
• Bleu: 8.7399

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	6.0913	0	9.4755	0.6769
No log	1	434	4.6127	0.0078	10.5923	1.6149
No log	2	868	3.7996	0.0156	12.7595	2.6426
No log	3	1302	3.1877	0.0312	14.7574	3.5921
No log	4	1736	2.8535	0.0625	17.8665	4.4721
0.1292	5	2170	2.5716	0.125	24.3006	5.3295
2.331	6	2604	2.2628	0.25	37.6650	6.1219
1.9509	7	3038	2.0146	0.5	62.6823	7.4085
1.585	8.0	3472	1.8155	1.0	113.4126	8.4135
1.1615	9.0	3906	1.7844	1.0	110.3768	8.5937
0.8671	10.0	4340	1.8237	1.0	111.8692	8.6426
0.6216	11.0	4774	1.8903	1.0	111.0240	8.6707
0.4329	12.0	5208	1.9875	1.0	109.5296	8.8373
0.3115	13.0	5642	2.0635	1.0	111.3556	8.7399

Framework versions

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