140ead7ee609d0cb161e9a508a2943eb

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

• Loss: 2.5545
• Data Size: 1.0
• Epoch Runtime: 51.3540
• Bleu: 3.5740

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	6.2093	0	3.2720	0.5923
No log	1	70	5.4909	0.0078	4.0283	1.3356
No log	2	140	4.8171	0.0156	8.6520	2.6316
No log	3	210	4.5573	0.0312	14.1054	3.2901
No log	4	280	4.0254	0.0625	19.8296	3.6352
No log	5	350	3.6888	0.125	19.8321	4.5991
No log	6	420	3.1952	0.25	23.8985	5.5649
0.6326	7	490	2.8361	0.5	35.7372	4.0599
2.9931	8.0	560	2.5080	1.0	49.5361	3.0505
2.677	9.0	630	2.4057	1.0	48.5187	3.2036
2.3192	10.0	700	2.3842	1.0	44.9186	3.3466
2.1093	11.0	770	2.3942	1.0	50.3657	3.4821
1.9723	12.0	840	2.4181	1.0	44.9340	3.4754
1.6807	13.0	910	2.4843	1.0	47.6983	3.3642
1.5699	14.0	980	2.5545	1.0	51.3540	3.5740

Framework versions

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