64584b7c6826bf6aa72d546ba8abf557

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

• Loss: 1.7005
• Data Size: 1.0
• Epoch Runtime: 24.1667
• Bleu: 11.8501

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	24.2349	0	2.1315	0.0258
No log	1	33	23.5409	0.0078	2.5022	0.0259
No log	2	66	21.6421	0.0156	4.7012	0.0237
No log	3	99	22.7850	0.0312	7.1662	0.0176
1.2843	4	132	21.1930	0.0625	10.2092	0.0164
1.2843	5	165	20.8555	0.125	12.5009	0.0168
1.2843	6	198	17.9570	0.25	13.9833	0.0236
4.287	7	231	18.8896	0.5	15.6441	0.0204
14.796	8.0	264	14.5836	1.0	23.3535	0.0213
14.796	9.0	297	9.6964	1.0	22.6342	0.0273
17.6945	10.0	330	6.3427	1.0	20.0977	0.0227
9.8726	11.0	363	6.1424	1.0	20.8876	0.0207
9.8726	12.0	396	6.0423	1.0	21.3110	0.0182
8.0695	13.0	429	5.9941	1.0	21.3595	0.0174
7.5223	14.0	462	5.3786	1.0	20.5379	0.0597
7.5223	15.0	495	5.1967	1.0	21.3496	0.0828
6.7361	16.0	528	3.1778	1.0	22.7931	0.3027
4.6037	17.0	561	2.0157	1.0	20.7813	1.1779
4.6037	18.0	594	1.8168	1.0	22.1433	6.8359
2.4518	19.0	627	1.7521	1.0	23.4117	7.5031
1.9653	20.0	660	1.7062	1.0	20.7907	7.8407
1.9653	21.0	693	1.6971	1.0	20.7467	8.5446
1.743	22.0	726	1.6720	1.0	21.6457	9.1907
1.5908	23.0	759	1.6659	1.0	23.3958	9.8446
1.5908	24.0	792	1.6796	1.0	21.4205	10.2767
1.4478	25.0	825	1.6936	1.0	22.3413	10.6892
1.3382	26.0	858	1.7027	1.0	22.7737	11.3011
1.3382	27.0	891	1.7005	1.0	24.1667	11.8501

Framework versions

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