91458963c080aaefe5c5cfcc2aabda0d

This model is a fine-tuned version of Helsinki-NLP/opus-mt-tc-bible-big-deu_eng_fra_por_spa-mul on the Helsinki-NLP/opus_books [fi-pl] dataset. It achieves the following results on the evaluation set:

• Loss: 2.4792
• Data Size: 1.0
• Epoch Runtime: 7.0986
• Bleu: 0.3930

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	8.4671	0	0.9391	0.0223
No log	1	70	6.3484	0.0078	1.7160	0.0451
No log	2	140	5.1718	0.0156	1.4729	0.0378
No log	3	210	4.3375	0.0312	2.1408	0.0297
No log	4	280	3.7902	0.0625	2.6638	0.0321
No log	5	350	3.3177	0.125	3.2226	0.0401
No log	6	420	2.9722	0.25	4.2486	0.1059
0.5252	7	490	2.6636	0.5	5.8323	0.1686
2.555	8.0	560	2.3982	1.0	7.9018	0.2215
2.2613	9.0	630	2.2990	1.0	6.9135	0.3053
1.9936	10.0	700	2.2692	1.0	6.9034	0.3934
1.7935	11.0	770	2.2646	1.0	6.7004	0.3889
1.6998	12.0	840	2.3026	1.0	6.7524	0.4402
1.4693	13.0	910	2.3682	1.0	6.7452	0.4094
1.333	14.0	980	2.4108	1.0	6.7377	0.3946
1.1935	15.0	1050	2.4792	1.0	7.0986	0.3930

Framework versions

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