54907d09f479e129c0d513d1eafbcc47

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

• Loss: 1.0872
• Data Size: 1.0
• Epoch Runtime: 13.4024
• Bleu: 6.2624

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	5.6807	0	1.5963	0.1010
No log	1	204	3.5220	0.0078	2.3839	0.0316
No log	2	408	3.0094	0.0156	2.1145	0.0072
No log	3	612	2.4518	0.0312	2.5577	0.0791
No log	4	816	2.1240	0.0625	3.0818	0.1892
No log	5	1020	1.9020	0.125	3.7342	0.7233
0.1793	6	1224	1.7328	0.25	5.1171	1.3163
1.7493	7	1428	1.5585	0.5	7.8978	1.9708
1.5298	8.0	1632	1.3895	1.0	14.3990	2.6825
1.3821	9.0	1836	1.2845	1.0	13.2500	3.3474
1.287	10.0	2040	1.2225	1.0	13.0935	3.7747
1.1731	11.0	2244	1.1713	1.0	13.6184	4.2759
1.1063	12.0	2448	1.1367	1.0	13.5218	4.7828
1.0303	13.0	2652	1.1098	1.0	13.1975	5.0388
0.9674	14.0	2856	1.0930	1.0	13.7111	5.1666
0.9303	15.0	3060	1.0710	1.0	14.2294	5.4132
0.8663	16.0	3264	1.0760	1.0	13.3073	5.6127
0.8239	17.0	3468	1.0641	1.0	13.1533	5.6693
0.7997	18.0	3672	1.0643	1.0	13.1749	5.7542
0.7471	19.0	3876	1.0548	1.0	13.7517	5.9914
0.7037	20.0	4080	1.0699	1.0	13.3547	6.0320
0.6667	21.0	4284	1.0757	1.0	13.5546	6.2629
0.6365	22.0	4488	1.0879	1.0	13.2568	6.1387
0.6119	23.0	4692	1.0872	1.0	13.4024	6.2624

Framework versions

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