ef4412b8383c0462187d67a5c1233fe9

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

• Loss: 1.9628
• Data Size: 1.0
• Epoch Runtime: 90.0474
• Bleu: 6.1524

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	20.4472	0	7.9210	0.0178
No log	1	419	20.9087	0.0078	9.5759	0.0133
No log	2	838	17.1391	0.0156	9.4612	0.0106
0.4543	3	1257	12.3211	0.0312	11.4169	0.0134
0.4543	4	1676	8.7521	0.0625	14.7170	0.0148
0.7366	5	2095	4.9832	0.125	20.5346	0.0285
0.5405	6	2514	2.7422	0.25	30.9037	1.6316
3.2715	7	2933	2.4198	0.5	52.6366	2.5227
2.831	8.0	3352	2.2523	1.0	91.4911	3.2185
2.6982	9.0	3771	2.1786	1.0	91.3632	3.5651
2.5973	10.0	4190	2.1253	1.0	91.7645	3.9337
2.4425	11.0	4609	2.0948	1.0	90.2739	4.2723
2.4152	12.0	5028	2.0658	1.0	91.5591	4.2959
2.2987	13.0	5447	2.0349	1.0	89.9808	4.8075
2.3058	14.0	5866	2.0176	1.0	90.4267	4.8226
2.1796	15.0	6285	2.0030	1.0	90.8174	5.0150
2.1521	16.0	6704	1.9857	1.0	89.5404	5.2649
2.108	17.0	7123	1.9821	1.0	90.7355	5.3942
2.0292	18.0	7542	1.9723	1.0	91.2213	5.6152
2.0119	19.0	7961	1.9651	1.0	90.3207	5.8017
1.9497	20.0	8380	1.9563	1.0	90.5436	5.7484
1.934	21.0	8799	1.9537	1.0	93.1253	5.7969
1.8906	22.0	9218	1.9542	1.0	91.1583	5.8890
1.8719	23.0	9637	1.9542	1.0	90.4493	5.9792
1.8063	24.0	10056	1.9491	1.0	90.4401	5.9122
1.8019	25.0	10475	1.9573	1.0	89.7545	6.0718
1.7436	26.0	10894	1.9389	1.0	91.5961	6.1435
1.7352	27.0	11313	1.9556	1.0	91.0348	6.0922
1.7133	28.0	11732	1.9518	1.0	91.3117	6.1735
1.6857	29.0	12151	1.9535	1.0	90.8807	6.2681
1.6422	30.0	12570	1.9628	1.0	90.0474	6.1524

Framework versions

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