969391e3edca42bb67265d889bb0f2de

This model is a fine-tuned version of google/umt5-base on the Helsinki-NLP/opus_books [en-no] dataset. It achieves the following results on the evaluation set:

• Loss: 2.3579
• Data Size: 1.0
• Epoch Runtime: 22.2381
• Bleu: 9.1771

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	12.6089	0	2.3110	0.0463
No log	1	87	12.3311	0.0078	2.5011	0.0285
No log	2	174	12.0040	0.0156	3.1168	0.0282
No log	3	261	12.0963	0.0312	4.6377	0.0334
No log	4	348	11.7581	0.0625	5.7346	0.0341
0.7837	5	435	11.2765	0.125	7.7011	0.0423
4.1787	6	522	9.1949	0.25	10.8425	0.0825
4.9454	7	609	8.1111	0.5	15.0277	0.1050
6.2734	8.0	696	5.4070	1.0	25.0478	1.0300
5.9352	9.0	783	3.4979	1.0	22.1364	6.2949
4.3914	10.0	870	2.9098	1.0	23.6211	5.3635
3.6986	11.0	957	2.7066	1.0	23.8962	6.2296
3.4338	12.0	1044	2.5942	1.0	21.7949	6.4528
3.1597	13.0	1131	2.5501	1.0	22.6695	6.8768
2.9923	14.0	1218	2.4953	1.0	22.8472	7.2702
2.8177	15.0	1305	2.4511	1.0	22.0388	7.5479
2.7483	16.0	1392	2.4291	1.0	22.3877	7.5955
2.6706	17.0	1479	2.4134	1.0	22.5722	7.8085
2.6043	18.0	1566	2.4011	1.0	23.2514	8.0208
2.5313	19.0	1653	2.3669	1.0	24.2886	8.0977
2.4446	20.0	1740	2.3642	1.0	22.3310	8.1760
2.3425	21.0	1827	2.3593	1.0	21.8849	8.3472
2.3573	22.0	1914	2.3436	1.0	23.0895	8.4311
2.2714	23.0	2001	2.3440	1.0	23.6487	8.6941
2.2094	24.0	2088	2.3409	1.0	22.1558	8.7230
2.1613	25.0	2175	2.3410	1.0	22.5434	8.7608
2.1342	26.0	2262	2.3345	1.0	22.8285	8.8175
2.0927	27.0	2349	2.3514	1.0	23.8641	8.7929
2.0233	28.0	2436	2.3373	1.0	23.6712	8.9700
1.9949	29.0	2523	2.3308	1.0	23.2570	8.9758
1.9759	30.0	2610	2.3329	1.0	23.2268	9.0151
1.901	31.0	2697	2.3509	1.0	23.0121	9.1556
1.8725	32.0	2784	2.3408	1.0	23.4961	9.2192
1.8202	33.0	2871	2.3579	1.0	22.2381	9.1771

Framework versions

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