a04f75b168fb5e65cc4aa043c0986e46

This model is a fine-tuned version of google/long-t5-local-large on the Helsinki-NLP/opus_books [de-ru] dataset. It achieves the following results on the evaluation set:

• Loss: 1.5194
• Data Size: 1.0
• Epoch Runtime: 190.7035
• Bleu: 2.9426

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	202.5984	0	14.1871	0.0032
No log	1	434	146.0637	0.0078	16.4882	0.0009
No log	2	868	82.6320	0.0156	17.8014	0.0042
No log	3	1302	36.9967	0.0312	21.5719	0.0014
No log	4	1736	12.9721	0.0625	27.5246	0.0322
2.352	5	2170	8.9631	0.125	38.5097	0.0120
13.7923	6	2604	6.8370	0.25	60.6155	0.0158
9.8565	7	3038	5.2524	0.5	102.4748	0.1683
7.1531	8.0	3472	3.9123	1.0	189.4097	0.1544
5.7235	9.0	3906	3.3748	1.0	187.1634	0.3366
4.9217	10.0	4340	2.8952	1.0	189.0930	0.5316
4.2887	11.0	4774	2.6399	1.0	189.0462	0.6685
3.7893	12.0	5208	2.5477	1.0	188.3899	0.6963
3.5148	13.0	5642	2.2363	1.0	188.5408	0.9609
3.1863	14.0	6076	2.1948	1.0	187.5332	0.9438
2.9695	15.0	6510	2.1323	1.0	191.2049	1.0293
2.8219	16.0	6944	2.0521	1.0	189.4276	1.1523
2.6797	17.0	7378	1.9762	1.0	187.4849	1.2506
2.5808	18.0	7812	1.9738	1.0	188.5891	1.2668
2.4829	19.0	8246	1.9159	1.0	189.5944	1.2928
2.3538	20.0	8680	1.8791	1.0	189.3800	1.2319
2.2968	21.0	9114	1.8656	1.0	188.7305	1.2929
2.2477	22.0	9548	1.8339	1.0	190.0467	1.2751
2.1886	23.0	9982	1.8181	1.0	188.2904	1.2143
2.1226	24.0	10416	1.7675	1.0	187.7876	1.5454
2.0982	25.0	10850	1.7498	1.0	187.9860	1.6132
2.0616	26.0	11284	1.7464	1.0	189.7731	1.6805
1.9994	27.0	11718	1.7334	1.0	188.4008	1.6033
1.9536	28.0	12152	1.7069	1.0	188.5661	1.7287
1.9417	29.0	12586	1.6890	1.0	187.8541	1.7209
1.912	30.0	13020	1.6810	1.0	189.3464	1.7065
1.8667	31.0	13454	1.6775	1.0	190.3884	1.7976
1.8477	32.0	13888	1.6604	1.0	190.3526	1.8207
1.8402	33.0	14322	1.6425	1.0	189.3657	1.9363
1.815	34.0	14756	1.6353	1.0	190.6804	2.0192
1.7951	35.0	15190	1.6228	1.0	191.0806	2.0446
1.7637	36.0	15624	1.6211	1.0	190.5929	1.9894
1.7506	37.0	16058	1.6261	1.0	187.5604	2.0763
1.721	38.0	16492	1.5991	1.0	188.9553	2.2292
1.6896	39.0	16926	1.5949	1.0	190.0516	2.2357
1.6767	40.0	17360	1.5716	1.0	190.8610	2.3121
1.6509	41.0	17794	1.5777	1.0	191.4814	2.2918
1.6528	42.0	18228	1.5765	1.0	189.4791	2.3562
1.6221	43.0	18662	1.5635	1.0	189.4849	2.4379
1.5932	44.0	19096	1.5540	1.0	190.0615	2.5319
1.5844	45.0	19530	1.5539	1.0	190.9674	2.6273
1.5697	46.0	19964	1.5391	1.0	189.2361	2.5885
1.5423	47.0	20398	1.5403	1.0	190.4720	2.5408
1.5435	48.0	20832	1.5302	1.0	190.5557	2.6584
1.536	49.0	21266	1.5258	1.0	189.1135	2.8198
1.5052	50.0	21700	1.5194	1.0	190.7035	2.9426

Framework versions

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