5e59f57dffe512f90060cd5e07b0c8fd

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

• Loss: 1.9345
• Data Size: 1.0
• Epoch Runtime: 34.5669
• Bleu: 10.9462

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	5.8294	0	1.9136	1.6688
No log	1	31	4.5176	0.0078	2.4182	5.0418
No log	2	62	4.1661	0.0156	9.8658	6.7587
No log	3	93	3.9249	0.0312	15.0261	9.0812
No log	4	124	3.6367	0.0625	22.9748	10.5419
No log	5	155	3.3171	0.125	28.8917	13.1800
No log	6	186	2.8972	0.25	31.2639	17.0151
0.6346	7	217	2.4808	0.5	31.6536	20.5573
0.6346	8.0	248	2.1165	1.0	39.0252	14.4510
1.8936	9.0	279	1.9540	1.0	40.5633	9.0780
2.1393	10.0	310	1.8616	1.0	30.1519	9.9327
2.1393	11.0	341	1.8266	1.0	35.1690	10.2229
1.7686	12.0	372	1.8304	1.0	37.8512	10.3581
1.4687	13.0	403	1.8382	1.0	42.0593	10.5740
1.4687	14.0	434	1.8662	1.0	31.1458	10.9923
1.1998	15.0	465	1.9345	1.0	34.5669	10.9462

Framework versions

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