95db879e959115e5bb32232064d629e4

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

• Loss: 1.6274
• Data Size: 1.0
• Epoch Runtime: 27.9245
• Bleu: 10.6838

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	6.9403	0	1.8082	0.0340
No log	1	33	7.4309	0.0078	2.1002	0.0322
No log	2	66	8.1900	0.0156	9.5218	0.0259
No log	3	99	10.2432	0.0312	15.7303	0.0175
0.5553	4	132	8.5060	0.0625	20.6668	0.0254
0.5553	5	165	5.0524	0.125	28.8793	0.0729
0.5553	6	198	3.3192	0.25	28.1629	0.4519
1.3857	7	231	2.2782	0.5	28.6872	1.0765
1.9739	8.0	264	1.6795	1.0	33.1974	7.4020
1.9739	9.0	297	1.5514	1.0	35.4250	8.2417
1.9021	10.0	330	1.5321	1.0	28.2163	9.3060
1.5089	11.0	363	1.5006	1.0	30.7004	9.4752
1.5089	12.0	396	1.4921	1.0	35.0547	10.7190
1.2617	13.0	429	1.5280	1.0	27.9345	10.2581
1.0589	14.0	462	1.5600	1.0	31.1559	11.1917
1.0589	15.0	495	1.5960	1.0	33.7256	10.6167
0.8995	16.0	528	1.6274	1.0	27.9245	10.6838

Framework versions

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