04f9ea514ccf8bc7303a7a2b03a264c7

This model is a fine-tuned version of google/mt5-large on the Helsinki-NLP/opus_books [fr-ru] dataset. It achieves the following results on the evaluation set:

• Loss: 1.7333
• Data Size: 1.0
• Epoch Runtime: 87.6990
• Bleu: 7.7583

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	24.1081	0	7.2208	0.0152
No log	1	204	24.5195	0.0078	7.8803	0.0132
No log	2	408	21.2534	0.0156	10.9378	0.0156
No log	3	612	16.3771	0.0312	14.2698	0.0220
No log	4	816	14.3314	0.0625	16.7211	0.0229
No log	5	1020	8.9428	0.125	22.1585	0.0325
1.2013	6	1224	3.1389	0.25	32.2510	0.1983
3.4015	7	1428	2.1476	0.5	52.0391	2.2204
2.5765	8.0	1632	1.9066	1.0	89.1632	4.4577
2.3174	9.0	1836	1.8226	1.0	88.5794	4.9668
2.1827	10.0	2040	1.7764	1.0	88.2255	5.3471
2.0352	11.0	2244	1.7506	1.0	86.7921	6.2440
1.9055	12.0	2448	1.7290	1.0	88.4878	6.6651
1.8262	13.0	2652	1.7236	1.0	89.1292	7.0670
1.7493	14.0	2856	1.7159	1.0	87.5957	7.3231
1.6841	15.0	3060	1.7062	1.0	87.9193	7.5114
1.5845	16.0	3264	1.7046	1.0	87.4291	7.7670
1.53	17.0	3468	1.7150	1.0	88.6632	7.5829
1.4929	18.0	3672	1.7126	1.0	87.8333	7.7418
1.4343	19.0	3876	1.7235	1.0	86.8167	7.5971
1.3673	20.0	4080	1.7333	1.0	87.6990	7.7583

Framework versions

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