1a8a9024ef68a2a54aeef06b5a7a29e9

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

• Loss: 1.1049
• Data Size: 1.0
• Epoch Runtime: 30.8157
• Bleu: 19.4043

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	5.4283	0	2.0844	2.9855
No log	1	35	4.4102	0.0078	2.7388	6.8520
No log	2	70	3.8369	0.0156	6.9642	9.6703
No log	3	105	3.2416	0.0312	14.8893	16.5760
No log	4	140	2.6959	0.0625	20.9497	24.1882
No log	5	175	2.1209	0.125	29.3386	31.4165
No log	6	210	1.8658	0.25	24.9732	35.7694
No log	7	245	1.4869	0.5	25.9546	37.8895
0.5288	8.0	280	1.1504	1.0	43.7308	16.9612
1.5408	9.0	315	1.0206	1.0	30.3582	18.4935
1.1092	10.0	350	0.9925	1.0	35.9729	18.6189
1.1092	11.0	385	1.0105	1.0	39.9857	18.6486
0.8356	12.0	420	1.0368	1.0	30.9963	19.0504
0.6771	13.0	455	1.0582	1.0	33.9931	19.3082
0.6771	14.0	490	1.1049	1.0	30.8157	19.4043

Framework versions

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