f108a1b2c1c40ccfff0ff9f1bbe49536

This model is a fine-tuned version of facebook/mbart-large-50-one-to-many-mmt on the Helsinki-NLP/opus_books [fi-no] dataset. It achieves the following results on the evaluation set:

• Loss: 4.3941
• Data Size: 1.0
• Epoch Runtime: 26.0804
• Bleu: 5.8488

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	9.5660	0	2.3212	0.2197
No log	1	85	8.4854	0.0078	3.0738	0.1704
No log	2	170	7.5612	0.0156	3.6157	0.3846
No log	3	255	6.9790	0.0312	4.7630	0.6933
No log	4	340	6.3720	0.0625	6.5844	0.8409
0.4266	5	425	5.7273	0.125	8.4043	1.1656
0.4266	6	510	4.9046	0.25	10.9274	1.3784
1.486	7	595	4.3110	0.5	15.1796	2.9648
3.8665	8.0	680	3.9113	1.0	28.1134	5.0586
3.111	9.0	765	3.7840	1.0	27.1383	5.1805
2.6286	10.0	850	3.7717	1.0	25.5010	6.0196
2.1677	11.0	935	3.8344	1.0	25.7779	6.2014
1.7239	12.0	1020	3.9871	1.0	27.3611	6.6675
1.3949	13.0	1105	4.2133	1.0	26.2326	5.5582
1.1029	14.0	1190	4.3941	1.0	26.0804	5.8488

Framework versions

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