5b409f4438fef7ba7ab2132dbaaa0d91

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

• Loss: 1.3111
• Data Size: 1.0
• Epoch Runtime: 180.7120
• Bleu: 10.3155

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	23.3626	0	13.8304	0.0125
No log	1	434	23.9957	0.0078	15.3958	0.0093
No log	2	868	20.1921	0.0156	17.7159	0.0108
No log	3	1302	10.4964	0.0312	21.5673	0.0223
No log	4	1736	3.9784	0.0625	28.8893	0.1610
0.5362	5	2170	2.5155	0.125	39.9764	0.5591
2.8305	6	2604	1.8329	0.25	59.9116	6.1797
2.1372	7	3038	1.5705	0.5	101.8626	6.1786
1.8663	8.0	3472	1.4538	1.0	190.6350	7.8930
1.6803	9.0	3906	1.3960	1.0	186.8617	9.0823
1.5558	10.0	4340	1.3584	1.0	187.2073	9.5962
1.4758	11.0	4774	1.3342	1.0	189.5848	9.9947
1.364	12.0	5208	1.3198	1.0	187.0171	10.0214
1.3104	13.0	5642	1.3096	1.0	190.2128	9.9382
1.2567	14.0	6076	1.3096	1.0	187.6537	9.9939
1.1979	15.0	6510	1.2991	1.0	189.2613	10.2176
1.1351	16.0	6944	1.2906	1.0	181.2897	10.2235
1.1014	17.0	7378	1.3017	1.0	179.3653	10.2882
1.0424	18.0	7812	1.3031	1.0	179.7579	10.2756
0.9774	19.0	8246	1.3075	1.0	180.9290	10.2699
0.9627	20.0	8680	1.3111	1.0	180.7120	10.3155

Framework versions

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