0c69f53a85f0b0ad9e328cd03bbf2f4d

This model is a fine-tuned version of google/mt5-small on the Helsinki-NLP/opus_books [es-no] dataset. It achieves the following results on the evaluation set:

• Loss: 2.7977
• Data Size: 1.0
• Epoch Runtime: 14.9374
• Bleu: 2.6526

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	24.9943	0	1.8690	0.0008
No log	1	89	24.9503	0.0078	1.9069	0.0007
No log	2	178	24.3999	0.0156	2.3205	0.0007
No log	3	267	23.1938	0.0312	2.6395	0.0008
No log	4	356	20.4866	0.0625	3.1138	0.0008
No log	5	445	17.5628	0.125	3.7849	0.0009
1.5531	6	534	13.6649	0.25	5.4983	0.0035
5.8809	7	623	9.5366	0.5	8.2629	0.0043
8.7533	8.0	712	5.7502	1.0	14.4154	0.0101
5.9357	9.0	801	4.0891	1.0	13.6731	0.1965
5.2307	10.0	890	3.6627	1.0	13.6807	0.4606
4.671	11.0	979	3.4887	1.0	13.9874	0.5414
4.3579	12.0	1068	3.3796	1.0	14.2997	0.6425
4.2043	13.0	1157	3.3079	1.0	14.6394	0.8816
4.1276	14.0	1246	3.2438	1.0	14.6013	0.9970
4.0182	15.0	1335	3.2049	1.0	15.1011	1.1170
3.8662	16.0	1424	3.1669	1.0	13.6995	1.2289
3.845	17.0	1513	3.1273	1.0	15.4472	1.3058
3.7523	18.0	1602	3.0947	1.0	14.2841	1.3628
3.6918	19.0	1691	3.0584	1.0	14.5412	1.3936
3.6462	20.0	1780	3.0379	1.0	14.5958	1.4721
3.566	21.0	1869	3.0200	1.0	15.0143	1.5464
3.543	22.0	1958	2.9990	1.0	15.1976	1.6810
3.4995	23.0	2047	2.9857	1.0	13.9438	1.6704
3.4371	24.0	2136	2.9665	1.0	14.0578	1.7620
3.432	25.0	2225	2.9575	1.0	14.7007	1.7714
3.3597	26.0	2314	2.9417	1.0	14.4075	1.9053
3.334	27.0	2403	2.9322	1.0	15.0994	1.9238
3.3543	28.0	2492	2.9205	1.0	15.1935	1.9365
3.2807	29.0	2581	2.9114	1.0	15.6820	1.9693
3.2708	30.0	2670	2.9043	1.0	13.9084	2.0454
3.1975	31.0	2759	2.8952	1.0	13.6897	2.0319
3.2045	32.0	2848	2.8804	1.0	14.0787	2.1034
3.1728	33.0	2937	2.8758	1.0	14.5912	2.1898
3.1786	34.0	3026	2.8676	1.0	15.6326	2.1644
3.1158	35.0	3115	2.8648	1.0	14.7389	2.2290
3.0955	36.0	3204	2.8586	1.0	14.8302	2.2062
3.0841	37.0	3293	2.8514	1.0	13.9386	2.2507
3.06	38.0	3382	2.8463	1.0	14.5196	2.2633
3.0343	39.0	3471	2.8391	1.0	14.7396	2.3475
3.0359	40.0	3560	2.8346	1.0	15.0044	2.3006
2.9838	41.0	3649	2.8281	1.0	15.0614	2.3554
2.966	42.0	3738	2.8225	1.0	15.7609	2.4191
2.934	43.0	3827	2.8153	1.0	15.9585	2.3923
2.936	44.0	3916	2.8206	1.0	15.8730	2.4333
2.9356	45.0	4005	2.8119	1.0	16.4149	2.4879
2.861	46.0	4094	2.8116	1.0	14.0475	2.5121
2.866	47.0	4183	2.8057	1.0	14.4774	2.6440
2.8413	48.0	4272	2.8038	1.0	14.4328	2.6129
2.8186	49.0	4361	2.7991	1.0	14.7876	2.5588
2.8101	50.0	4450	2.7977	1.0	14.9374	2.6526

Framework versions

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