86216830e59ebeae2d54426b0e421a42

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

• Loss: 2.5618
• Data Size: 1.0
• Epoch Runtime: 55.9854
• Bleu: 7.8963

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	10.0525	0	5.3191	0.2877
No log	1	204	4.6881	0.0078	6.9191	1.9941
No log	2	408	4.0168	0.0156	6.9221	2.7411
No log	3	612	3.7356	0.0312	9.1861	3.5176
No log	4	816	3.1467	0.0625	10.9352	4.2581
No log	5	1020	2.7263	0.125	14.3815	4.6251
0.2633	6	1224	2.4107	0.25	20.7555	5.1189
2.6811	7	1428	2.7302	0.5	32.9229	4.7323
2.0875	8.0	1632	2.1668	1.0	57.4955	7.6951
1.6264	9.0	1836	2.1788	1.0	56.8797	9.7831
1.2996	10.0	2040	2.2509	1.0	56.2120	7.6752
0.9938	11.0	2244	2.3547	1.0	57.6545	7.0823
0.7527	12.0	2448	2.5618	1.0	55.9854	7.8963

Framework versions

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