e51ec449b06f2fb04171eca8d22e4e71

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

• Loss: 3.8620
• Data Size: 1.0
• Epoch Runtime: 24.8006
• Bleu: 5.5143

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	9.3261	0	2.3691	0.2400
No log	1	85	8.1859	0.0078	3.2752	0.2718
No log	2	170	7.0799	0.0156	3.6659	0.7489
No log	3	255	6.5017	0.0312	4.7162	0.9274
No log	4	340	5.9026	0.0625	6.1222	1.1520
0.3985	5	425	5.2132	0.125	8.1075	1.6933
0.3985	6	510	4.4264	0.25	10.8217	2.2844
1.3351	7	595	3.9312	0.5	14.8494	2.9898
3.4362	8.0	680	3.5939	1.0	26.8419	3.7224
2.6442	9.0	765	3.4622	1.0	26.4859	4.0347
2.1099	10.0	850	3.4875	1.0	24.7126	4.7718
1.6385	11.0	935	3.5734	1.0	25.5174	4.5416
1.2281	12.0	1020	3.7410	1.0	25.4161	4.7051
0.9387	13.0	1105	3.8620	1.0	24.8006	5.5143

Framework versions

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