4f44b16855ea244aeb2f4e3570233743

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

• Loss: 2.9369
• Data Size: 1.0
• Epoch Runtime: 6.5193
• Bleu: 2.4252

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	7.7042	0	1.1171	0.0917
No log	1	91	7.0587	0.0078	1.6726	0.1812
No log	2	182	6.4934	0.0156	1.3671	0.2227
No log	3	273	6.0683	0.0312	1.5600	0.0832
No log	4	364	5.5922	0.0625	1.6840	0.3601
No log	5	455	5.0549	0.125	2.1190	0.4097
No log	6	546	4.4864	0.25	2.5782	0.6015
0.4957	7	637	3.9661	0.5	3.9885	1.4232
3.8124	8.0	728	3.5344	1.0	6.3736	1.7445
3.3883	9.0	819	3.2933	1.0	6.3176	1.9267
3.0917	10.0	910	3.1310	1.0	6.1658	2.1607
2.8475	11.0	1001	3.0214	1.0	6.3216	2.2193
2.6654	12.0	1092	2.9420	1.0	6.3733	2.2780
2.4722	13.0	1183	2.8980	1.0	6.5454	2.3527
2.2706	14.0	1274	2.8747	1.0	6.1674	2.4034
2.1071	15.0	1365	2.8530	1.0	6.1465	2.4375
1.9545	16.0	1456	2.8542	1.0	6.2864	2.4483
1.787	17.0	1547	2.8480	1.0	6.4477	2.4202
1.6386	18.0	1638	2.8774	1.0	6.7741	2.3760
1.5028	19.0	1729	2.8870	1.0	7.5152	2.4360
1.3884	20.0	1820	2.8991	1.0	6.5545	2.5406
1.2868	21.0	1911	2.9369	1.0	6.5193	2.4252

Framework versions

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