20d638d02ef3a013a4d941bb1213614b

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

• Loss: 1.7550
• Data Size: 1.0
• Epoch Runtime: 327.3207
• Bleu: 7.2327

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	7.7231	0	21.4023	0.0169
No log	1	688	3.5766	0.0078	24.0239	0.3067
No log	2	1376	2.8469	0.0156	30.7525	0.5278
No log	3	2064	2.2278	0.0312	39.3824	0.7233
0.1223	4	2752	1.9983	0.0625	53.2779	4.1132
0.1894	5	3440	1.8789	0.125	68.8534	5.3701
2.144	6	4128	1.7926	0.25	106.3536	6.0453
1.9787	7	4816	1.7305	0.5	180.6113	6.5822
1.8253	8.0	5504	1.6614	1.0	327.5792	6.7351
1.6459	9.0	6192	1.6412	1.0	322.5477	6.9843
1.502	10.0	6880	1.6297	1.0	325.1674	7.0326
1.3639	11.0	7568	1.6529	1.0	325.3689	7.4017
1.2501	12.0	8256	1.6803	1.0	323.8920	7.5099
1.1506	13.0	8944	1.7121	1.0	324.7622	7.1286
1.0853	14.0	9632	1.7550	1.0	327.3207	7.2327

Framework versions

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