2cb96db85a0b3aa79cf17cf7316f507e

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

• Loss: 2.1290
• Data Size: 1.0
• Epoch Runtime: 24.1905
• Bleu: 4.5965

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	7.3838	0	2.4479	0.1340
No log	1	390	6.0867	0.0078	3.0070	0.1444
No log	2	780	5.5103	0.0156	3.0680	0.2464
No log	3	1170	4.8498	0.0312	3.5405	0.3908
No log	4	1560	4.2201	0.0625	4.2382	0.8343
0.2621	5	1950	3.7420	0.125	5.7792	1.2614
0.5536	6	2340	3.3187	0.25	8.0047	1.7998
3.1241	7	2730	2.9250	0.5	13.6797	2.5626
2.6886	8.0	3120	2.5791	1.0	24.3796	3.3368
2.4328	9.0	3510	2.4166	1.0	23.8098	3.6219
2.256	10.0	3900	2.3173	1.0	23.8307	3.8409
2.128	11.0	4290	2.2387	1.0	23.0042	4.0513
2.011	12.0	4680	2.1867	1.0	23.8362	4.2675
1.9335	13.0	5070	2.1611	1.0	22.8356	4.2757
1.8487	14.0	5460	2.1390	1.0	24.6442	4.3811
1.7208	15.0	5850	2.1169	1.0	22.8496	4.4555
1.6473	16.0	6240	2.1034	1.0	24.6025	4.5328
1.5977	17.0	6630	2.0967	1.0	23.2151	4.5667
1.5234	18.0	7020	2.1110	1.0	24.4015	4.5911
1.4693	19.0	7410	2.1283	1.0	24.1252	4.5897
1.4073	20.0	7800	2.1240	1.0	24.1978	4.6644
1.351	21.0	8190	2.1290	1.0	24.1905	4.5965

Framework versions

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