78c43c296753607a367bd75d67a2b57d

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

• Loss: 3.0414
• Data Size: 1.0
• Epoch Runtime: 36.2348
• Bleu: 0.0885

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	209.2599	0	2.9343	0.0022
No log	1	70	187.7119	0.0078	3.6677	0.0021
No log	2	140	160.4909	0.0156	5.1776	0.0035
No log	3	210	135.7296	0.0312	6.6340	0.0033
No log	4	280	104.5299	0.0625	8.9982	0.0028
No log	5	350	57.1903	0.125	12.2910	0.0035
No log	6	420	28.1823	0.25	16.1746	0.0013
10.6872	7	490	14.6561	0.5	22.8352	0.0036
20.0982	8.0	560	10.7794	1.0	37.2661	0.0048
16.7674	9.0	630	8.8928	1.0	36.4249	0.0079
13.3888	10.0	700	8.2467	1.0	36.7575	0.0135
12.499	11.0	770	7.3715	1.0	36.2484	0.0158
11.725	12.0	840	6.9170	1.0	37.2235	0.0162
10.5242	13.0	910	6.5168	1.0	36.5129	0.0192
10.0615	14.0	980	6.2463	1.0	36.6950	0.0197
9.3333	15.0	1050	6.1689	1.0	36.1341	0.0191
8.96	16.0	1120	5.8733	1.0	36.9673	0.0211
8.6238	17.0	1190	5.5877	1.0	36.3308	0.0226
8.1025	18.0	1260	5.2607	1.0	36.7496	0.0207
7.8044	19.0	1330	5.3178	1.0	36.2681	0.0187
7.4109	20.0	1400	4.9554	1.0	36.5586	0.0334
7.2297	21.0	1470	4.7693	1.0	36.2652	0.0233
7.0329	22.0	1540	4.9162	1.0	36.9754	0.0461
6.6534	23.0	1610	4.3485	1.0	36.5360	0.0201
6.4951	24.0	1680	4.3899	1.0	36.1867	0.0439
6.1811	25.0	1750	4.3861	1.0	36.2829	0.0531
6.0734	26.0	1820	4.2726	1.0	36.4671	0.0478
5.9986	27.0	1890	4.1590	1.0	36.6922	0.0432
5.7188	28.0	1960	4.1682	1.0	36.0706	0.0406
5.5975	29.0	2030	3.9898	1.0	36.4282	0.0456
5.3677	30.0	2100	3.9291	1.0	35.9161	0.0461
5.3125	31.0	2170	3.7723	1.0	36.5840	0.0542
5.2008	32.0	2240	3.6980	1.0	36.7902	0.0614
4.999	33.0	2310	3.6551	1.0	36.8089	0.0701
4.9079	34.0	2380	3.5539	1.0	36.7155	0.0535
4.7541	35.0	2450	3.4654	1.0	36.7493	0.1156
4.6667	36.0	2520	3.4963	1.0	36.4739	0.0356
4.5992	37.0	2590	3.3897	1.0	36.0276	0.0715
4.4739	38.0	2660	3.3127	1.0	36.3668	0.0923
4.3799	39.0	2730	3.2270	1.0	37.5475	0.1129
4.2726	40.0	2800	3.2151	1.0	36.9524	0.1486
4.1815	41.0	2870	3.2228	1.0	36.4025	0.0813
4.1766	42.0	2940	3.2554	1.0	36.9824	0.0646
4.0409	43.0	3010	3.1722	1.0	36.4834	0.0750
3.9996	44.0	3080	3.1450	1.0	36.3839	0.0706
3.9136	45.0	3150	3.0476	1.0	36.4526	0.0890
3.8524	46.0	3220	3.0807	1.0	36.4076	0.0702
3.7894	47.0	3290	2.9933	1.0	37.0932	0.0943
3.6966	48.0	3360	2.9177	1.0	37.4004	0.1357
3.668	49.0	3430	2.9728	1.0	36.4877	0.0849
3.6084	50.0	3500	3.0414	1.0	36.2348	0.0885

Framework versions

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