5134fe4feaf8728a8c281a7eb1e63136

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

• Loss: 2.5673
• Data Size: 1.0
• Epoch Runtime: 45.3823
• Bleu: 4.0225

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	6.1512	0	3.2356	0.8828
No log	1	70	5.3169	0.0078	3.7765	2.0050
No log	2	140	4.7405	0.0156	8.6043	2.8797
No log	3	210	4.4349	0.0312	14.1447	3.1755
No log	4	280	3.9278	0.0625	20.2654	3.9124
No log	5	350	3.6416	0.125	20.8366	4.5577
No log	6	420	3.2453	0.25	24.5699	6.0642
0.6354	7	490	2.8171	0.5	35.7157	7.0532
2.9661	8.0	560	2.4645	1.0	58.7406	3.0399
2.5954	9.0	630	2.3717	1.0	47.6476	3.5116
2.226	10.0	700	2.3515	1.0	45.0232	3.7462
1.9643	11.0	770	2.3426	1.0	49.6419	4.0440
1.8322	12.0	840	2.3624	1.0	44.8523	4.0094
1.5688	13.0	910	2.4258	1.0	48.5923	4.0527
1.4098	14.0	980	2.4559	1.0	48.5111	3.9110
1.2588	15.0	1050	2.5673	1.0	45.3823	4.0225

Framework versions

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