ae3f6f9a8fb3c68f9bee3e66321b3d68

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

• Loss: 3.8653
• Data Size: 1.0
• Epoch Runtime: 26.1910
• Bleu: 4.4308

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	7.3951	0	2.4994	0.3194
No log	1	91	6.7682	0.0078	3.4514	0.2915
No log	2	182	6.2207	0.0156	4.5152	0.3771
No log	3	273	5.8200	0.0312	5.9047	0.4447
No log	4	364	5.3328	0.0625	7.6367	0.6466
No log	5	455	4.8644	0.125	8.9408	1.3717
No log	6	546	4.3926	0.25	11.7857	1.9901
0.469	7	637	3.9825	0.5	16.7760	2.5710
3.3066	8.0	728	3.5313	1.0	28.6812	3.2830
2.5231	9.0	819	3.3818	1.0	26.4448	3.4047
1.8958	10.0	910	3.4478	1.0	27.1888	3.7493
1.4316	11.0	1001	3.5424	1.0	27.6848	4.6297
0.987	12.0	1092	3.7127	1.0	26.6558	5.0275
0.695	13.0	1183	3.8653	1.0	26.1910	4.4308

Framework versions

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