3048e4c2ab0d9efc3daa857ab2714c1b

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

• Loss: 1.8672
• Data Size: 1.0
• Epoch Runtime: 220.4552
• Bleu: 8.1452

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	5.4846	0	15.2137	1.9323
No log	1	419	4.0330	0.0078	18.1488	5.8458
No log	2	838	3.3093	0.0156	23.2445	9.1132
0.1266	3	1257	2.8515	0.0312	29.8031	11.9876
0.1266	4	1676	2.3942	0.0625	35.7858	14.4652
0.1895	5	2095	2.0690	0.125	53.0679	5.8884
0.3167	6	2514	1.9319	0.25	73.9145	6.4234
2.1187	7	2933	1.8440	0.5	123.6452	7.0374
1.896	8.0	3352	1.7652	1.0	221.2216	7.6377
1.7204	9.0	3771	1.7368	1.0	216.4819	7.9181
1.5795	10.0	4190	1.7352	1.0	217.6422	8.0089
1.3851	11.0	4609	1.7597	1.0	216.3156	8.1047
1.2893	12.0	5028	1.7809	1.0	216.2436	8.1628
1.1339	13.0	5447	1.8185	1.0	217.2292	8.2207
1.0763	14.0	5866	1.8672	1.0	220.4552	8.1452

Framework versions

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