7773ad480c417a27889ffce6154bb8ff

This model is a fine-tuned version of google/umt5-xl on the Helsinki-NLP/opus_books [de-it] dataset. It achieves the following results on the evaluation set:

• Loss: 1.7256
• Data Size: 1.0
• Epoch Runtime: 351.4545
• Bleu: 9.3610

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	5.6193	0	24.3287	1.7357
No log	1	684	3.6946	0.0078	27.3071	7.8813
No log	2	1368	2.9833	0.0156	34.4032	11.4621
No log	3	2052	2.5257	0.0312	44.6903	14.2871
No log	4	2736	2.2360	0.0625	57.7191	6.0669
2.5665	5	3420	1.9873	0.125	78.1888	6.6060
2.3373	6	4104	1.8818	0.25	116.1459	7.2287
2.0869	7	4788	1.7860	0.5	200.8131	7.8675
1.9294	8.0	5472	1.7030	1.0	356.3443	8.6230
1.7207	9.0	6156	1.6619	1.0	351.5749	8.8814
1.5659	10.0	6840	1.6628	1.0	351.1586	9.1244
1.4302	11.0	7524	1.6654	1.0	351.9081	9.3690
1.2979	12.0	8208	1.6881	1.0	352.9931	9.3489
1.1766	13.0	8892	1.7256	1.0	351.4545	9.3610

Framework versions

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