4b082f3f0ffeb2a6019a7cc94c048543

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

• Loss: 2.0289
• Data Size: 1.0
• Epoch Runtime: 308.2418
• Bleu: 10.3974

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	10.0965	0	23.6622	0.1237
No log	1	1286	9.7308	0.0078	27.4981	0.1034
0.3339	2	2572	10.2505	0.0156	28.7657	0.1188
0.432	3	3858	9.7780	0.0312	33.6258	0.1033
0.5747	4	5144	5.9837	0.0625	41.8691	0.5828
4.6949	5	6430	3.2677	0.125	57.9934	9.0637
3.5816	6	7716	2.7448	0.25	92.7076	5.7746
3.1415	7	9002	2.5153	0.5	159.0431	6.3492
2.9083	8.0	10288	2.3561	1.0	292.2227	7.4540
2.7148	9.0	11574	2.2822	1.0	293.9180	7.7842
2.5949	10.0	12860	2.2177	1.0	292.4473	8.2885
2.4574	11.0	14146	2.1866	1.0	289.4107	8.6147
2.4596	12.0	15432	2.1481	1.0	289.9727	8.7403
2.3263	13.0	16718	2.1282	1.0	292.9969	8.8737
2.2863	14.0	18004	2.1034	1.0	289.8140	9.0711
2.277	15.0	19290	2.0859	1.0	306.6649	9.1682
2.1752	16.0	20576	2.0698	1.0	307.2018	9.3133
2.1554	17.0	21862	2.0599	1.0	306.9976	9.3497
2.0987	18.0	23148	2.0588	1.0	306.9791	9.5375
2.0736	19.0	24434	2.0338	1.0	308.9585	9.6984
1.9908	20.0	25720	2.0317	1.0	308.4367	9.7726
1.9864	21.0	27006	2.0297	1.0	308.6215	9.6938
1.9704	22.0	28292	2.0201	1.0	307.1069	9.9086
1.9303	23.0	29578	2.0163	1.0	308.1973	9.9983
1.8742	24.0	30864	2.0154	1.0	308.4859	10.0873
1.8359	25.0	32150	2.0161	1.0	309.6682	10.0811
1.837	26.0	33436	2.0092	1.0	309.4687	10.1728
1.7884	27.0	34722	2.0116	1.0	309.2874	10.0999
1.7442	28.0	36008	2.0112	1.0	310.3224	10.2208
1.7304	29.0	37294	2.0023	1.0	310.3041	10.3232
1.713	30.0	38580	2.0114	1.0	308.7972	10.2984
1.6343	31.0	39866	2.0239	1.0	311.5922	10.2938
1.606	32.0	41152	2.0194	1.0	308.8730	10.2996
1.5992	33.0	42438	2.0289	1.0	308.2418	10.3974

Framework versions

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