d72d3cfe8861ed9bca9e3faa6a2a75d2

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

• Loss: 2.4183
• Data Size: 1.0
• Epoch Runtime: 56.2133
• Bleu: 6.3918

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	10.2193	0	5.3934	0.1363
No log	1	204	7.7337	0.0078	7.1698	0.1956
No log	2	408	5.4899	0.0156	7.0925	0.2652
No log	3	612	3.6764	0.0312	9.1531	1.5411
No log	4	816	3.0859	0.0625	11.2122	2.4468
No log	5	1020	2.6803	0.125	14.3287	3.5942
0.2748	6	1224	2.4689	0.25	21.0019	4.7902
2.3488	7	1428	2.2746	0.5	33.1998	6.0747
2.0178	8.0	1632	2.1681	1.0	58.8386	7.0957
1.5977	9.0	1836	2.1980	1.0	58.6352	6.9815
1.3391	10.0	2040	2.2276	1.0	55.7050	10.6408
0.9947	11.0	2244	2.2963	1.0	55.9216	7.0433
0.7692	12.0	2448	2.4183	1.0	56.2133	6.3918

Framework versions

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