f64ecc4aeb51b053cecae47d815c724a

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

• Loss: 2.6936
• Data Size: 1.0
• Epoch Runtime: 4.7638
• Bleu: 0.6405

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	8.4134	0	0.9973	0.0133
No log	1	58	7.3532	0.0078	1.6767	0.0499
No log	2	116	6.8972	0.0156	1.2262	0.1500
No log	3	174	6.4281	0.0312	1.2670	0.1453
No log	4	232	5.8248	0.0625	1.5846	0.0648
No log	5	290	5.0911	0.125	1.7803	0.1366
0.4997	6	348	4.4645	0.25	2.3705	0.0906
0.6531	7	406	3.9468	0.5	3.2073	0.1052
2.883	8.0	464	3.5137	1.0	5.1638	0.1142
3.611	9.0	522	3.2992	1.0	4.9019	0.1352
3.3946	10.0	580	3.1590	1.0	4.4863	0.2642
3.2628	11.0	638	3.0634	1.0	4.5552	0.1885
3.1405	12.0	696	2.9958	1.0	4.7116	0.3218
2.9724	13.0	754	2.9324	1.0	4.8064	0.3044
2.8922	14.0	812	2.8916	1.0	4.6985	0.3649
2.8167	15.0	870	2.8495	1.0	4.8551	0.3981
2.7602	16.0	928	2.8170	1.0	5.1181	0.4298
2.712	17.0	986	2.7919	1.0	5.2093	0.4582
2.6443	18.0	1044	2.7571	1.0	5.0125	0.5180
2.5507	19.0	1102	2.7485	1.0	4.9770	0.5086
2.4848	20.0	1160	2.7351	1.0	5.3925	0.5671
2.4457	21.0	1218	2.7130	1.0	5.2506	0.5826
2.398	22.0	1276	2.7127	1.0	4.9315	0.6359
2.3506	23.0	1334	2.7075	1.0	4.8651	0.5891
2.3122	24.0	1392	2.6896	1.0	4.6772	0.6432
2.2335	25.0	1450	2.7065	1.0	4.7600	0.6275
2.1836	26.0	1508	2.6983	1.0	4.8372	0.6173
2.127	27.0	1566	2.6906	1.0	4.6869	0.6627
2.0998	28.0	1624	2.6936	1.0	4.7638	0.6405

Framework versions

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