19d3c84947303e85c2fe70324f7e3edf

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

• Loss: 1.4919
• Data Size: 1.0
• Epoch Runtime: 61.4229
• Bleu: 6.0299

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	8.4172	0	5.3870	0.0277
No log	1	1000	6.1717	0.0078	6.3496	0.0897
No log	2	2000	5.1173	0.0156	6.3012	0.1106
No log	3	3000	4.3603	0.0312	7.0461	0.2314
0.1684	4	4000	3.8039	0.0625	8.8263	0.5564
3.6746	5	5000	3.3532	0.125	12.3997	1.0019
0.1948	6	6000	2.9186	0.25	19.0197	1.5298
0.2422	7	7000	2.5081	0.5	32.6737	2.2591
2.2451	8.0	8000	2.1127	1.0	59.4709	3.1630
1.9915	9.0	9000	1.9149	1.0	59.1901	3.9066
1.8117	10.0	10000	1.7829	1.0	60.0190	4.4409
1.6855	11.0	11000	1.6953	1.0	58.9931	4.8143
1.6028	12.0	12000	1.6365	1.0	60.6151	4.9521
1.5129	13.0	13000	1.5933	1.0	59.9825	5.1537
1.4309	14.0	14000	1.5561	1.0	60.8551	5.3746
1.3621	15.0	15000	1.5323	1.0	61.7483	5.4809
1.3014	16.0	16000	1.5148	1.0	61.1742	5.5865
1.2221	17.0	17000	1.5016	1.0	60.5016	5.7213
1.2173	18.0	18000	1.4925	1.0	61.5947	5.8126
1.1572	19.0	19000	1.4841	1.0	62.4802	5.8839
1.1153	20.0	20000	1.4827	1.0	61.3685	5.9235
1.0884	21.0	21000	1.4759	1.0	61.1664	5.9707
1.0506	22.0	22000	1.4868	1.0	60.2091	5.9909
0.9963	23.0	23000	1.4806	1.0	59.7321	6.0556
0.9482	24.0	24000	1.4841	1.0	60.0780	6.0219
0.9597	25.0	25000	1.4919	1.0	61.4229	6.0299

Framework versions

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