e357eeaec15604254da69c50ce39f181

This model is a fine-tuned version of Helsinki-NLP/opus-mt-en-sv on the Helsinki-NLP/opus_books [de-pt] dataset. It achieves the following results on the evaluation set:

• Loss: 3.0428
• Data Size: 1.0
• Epoch Runtime: 3.2821
• Bleu: 2.9934

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	7.4490	0	0.8513	0.0782
No log	1	27	6.8525	0.0078	1.2793	0.0832
No log	2	54	6.5618	0.0156	1.0775	0.0705
No log	3	81	6.3630	0.0312	1.1832	0.0926
No log	4	108	6.0918	0.0625	1.2253	0.0963
No log	5	135	5.6843	0.125	1.4588	0.1280
No log	6	162	5.2201	0.25	1.5422	0.5293
No log	7	189	4.6378	0.5	2.1339	0.8190
1.0889	8.0	216	4.0538	1.0	2.9963	1.0888
1.0889	9.0	243	3.7419	1.0	2.8456	1.3706
3.9701	10.0	270	3.5294	1.0	2.5534	1.6618
3.9701	11.0	297	3.4039	1.0	2.6783	1.8156
3.3816	12.0	324	3.2989	1.0	2.7060	2.0567
2.9955	13.0	351	3.2061	1.0	2.5990	2.1425
2.9955	14.0	378	3.1629	1.0	2.6498	2.3677
2.6975	15.0	405	3.1155	1.0	2.7766	2.5658
2.6975	16.0	432	3.0908	1.0	2.8570	2.4425
2.444	17.0	459	3.0701	1.0	2.8450	2.6497
2.444	18.0	486	3.0420	1.0	2.7861	2.6421
2.2063	19.0	513	3.0234	1.0	2.8805	2.7241
2.2063	20.0	540	3.0025	1.0	2.7423	2.7504
2.0156	21.0	567	3.0133	1.0	2.6118	2.8534
2.0156	22.0	594	3.0246	1.0	2.7367	2.8574
1.8305	23.0	621	3.0338	1.0	2.9701	2.9770
1.8305	24.0	648	3.0428	1.0	3.2821	2.9934

Framework versions

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