4c71ce1eafdb5d36bfec031e622c7637

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

• Loss: 2.2967
• Data Size: 1.0
• Epoch Runtime: 296.9825
• Bleu: 1.5312

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	229.1420	0	21.3235	0.0102
No log	1	684	169.5335	0.0078	23.9654	0.0061
No log	2	1368	105.1773	0.0156	27.4962	0.0041
No log	3	2052	37.4363	0.0312	33.8458	0.0015
No log	4	2736	17.6935	0.0625	42.6523	0.0095
22.5653	5	3420	12.8546	0.125	61.7509	0.2321
15.8601	6	4104	9.7817	0.25	97.5144	0.0232
11.2192	7	4788	7.5349	0.5	167.9222	0.0168
7.604	8.0	5472	5.1665	1.0	311.9814	0.1142
5.8428	9.0	6156	4.0933	1.0	306.4764	0.1187
5.0192	10.0	6840	3.7044	1.0	308.3549	0.1499
4.485	11.0	7524	3.3896	1.0	300.2145	0.1799
4.0706	12.0	8208	3.1497	1.0	298.1315	0.3534
3.8092	13.0	8892	3.0523	1.0	300.3850	0.3495
3.6472	14.0	9576	2.9593	1.0	298.2718	0.4008
3.4461	15.0	10260	2.8402	1.0	297.2106	0.4580
3.3258	16.0	10944	2.7898	1.0	301.6928	0.4834
3.2073	17.0	11628	2.7334	1.0	300.2841	0.6354
3.1163	18.0	12312	2.6852	1.0	298.0317	0.6447
3.0383	19.0	12996	2.6541	1.0	297.7672	0.6590
2.9504	20.0	13680	2.6051	1.0	300.4691	0.7512
2.8632	21.0	14364	2.5677	1.0	297.4315	0.8523
2.8375	22.0	15048	2.5375	1.0	296.9936	0.8462
2.7784	23.0	15732	2.5039	1.0	299.7437	0.9535
2.7273	24.0	16416	2.4831	1.0	299.1126	0.9218
2.7019	25.0	17100	2.4658	1.0	307.1864	0.9450
2.6508	26.0	17784	2.4403	1.0	298.0001	1.0115
2.5866	27.0	18468	2.4429	1.0	300.0931	1.0111
2.5655	28.0	19152	2.4098	1.0	298.6058	1.0417
2.5107	29.0	19836	2.3942	1.0	297.7683	1.2072
2.4684	30.0	20520	2.3807	1.0	297.4598	1.1409
2.4466	31.0	21204	2.3601	1.0	298.9890	1.1560
2.4378	32.0	21888	2.3492	1.0	298.1015	1.2198
2.375	33.0	22572	2.3348	1.0	298.1212	1.2304
2.3469	34.0	23256	2.3288	1.0	296.2021	1.3018
2.3185	35.0	23940	2.3205	1.0	298.5160	1.3001
2.2792	36.0	24624	2.3057	1.0	298.3413	1.3007
2.2462	37.0	25308	2.3119	1.0	297.9856	1.3171
2.2336	38.0	25992	2.3040	1.0	300.3649	1.4047
2.178	39.0	26676	2.2890	1.0	298.0711	1.3388
2.1507	40.0	27360	2.2871	1.0	299.2080	1.4162
2.1355	41.0	28044	2.2830	1.0	296.7871	1.4717
2.113	42.0	28728	2.2790	1.0	297.9198	1.4811
2.0881	43.0	29412	2.2894	1.0	297.2793	1.5521
2.0622	44.0	30096	2.2832	1.0	298.7365	1.4748
2.0253	45.0	30780	2.2811	1.0	295.0279	1.5263
2.0143	46.0	31464	2.2967	1.0	296.9825	1.5312

Framework versions

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