f71cbbbfe981e7200584100d7966fa2b

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

• Loss: 0.9341
• Data Size: 1.0
• Epoch Runtime: 28.5646
• Bleu: 7.3742

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	5.0592	0	2.6505	0.0115
No log	1	434	4.4341	0.0078	3.0459	0.0569
No log	2	868	3.9381	0.0156	3.3078	0.0099
No log	3	1302	3.2692	0.0312	3.6265	0.0726
No log	4	1736	2.7618	0.0625	4.4822	0.1281
0.1296	5	2170	2.5026	0.125	6.7853	0.3005
2.4159	6	2604	2.2260	0.25	8.8421	0.6399
2.0546	7	3038	1.9135	0.5	15.0381	1.2476
1.7035	8.0	3472	1.5633	1.0	27.1699	2.6298
1.472	9.0	3906	1.3763	1.0	27.4828	3.4125
1.3375	10.0	4340	1.2564	1.0	27.1508	3.7303
1.2342	11.0	4774	1.1741	1.0	27.0430	4.2354
1.1346	12.0	5208	1.1151	1.0	26.5754	4.4166
1.078	13.0	5642	1.0643	1.0	26.6082	4.9190
1.0196	14.0	6076	1.0308	1.0	26.8679	5.3985
0.9742	15.0	6510	1.0061	1.0	27.4594	5.3076
0.9201	16.0	6944	0.9779	1.0	26.3045	5.9532
0.8864	17.0	7378	0.9632	1.0	26.8915	6.0912
0.8432	18.0	7812	0.9518	1.0	26.3667	6.2207
0.8019	19.0	8246	0.9407	1.0	26.7003	6.2461
0.7776	20.0	8680	0.9374	1.0	26.3287	6.5594
0.7523	21.0	9114	0.9198	1.0	26.7803	6.7620
0.7127	22.0	9548	0.9247	1.0	26.7151	6.5384
0.6878	23.0	9982	0.9187	1.0	26.8602	6.8646
0.666	24.0	10416	0.9250	1.0	26.6537	7.0121
0.6503	25.0	10850	0.9153	1.0	27.0551	6.7534
0.6202	26.0	11284	0.9236	1.0	27.1023	6.8676
0.5995	27.0	11718	0.9233	1.0	26.7890	7.0623
0.5655	28.0	12152	0.9331	1.0	27.1751	6.9835
0.5552	29.0	12586	0.9341	1.0	28.5646	7.3742

Framework versions

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