9820033747fd2bc74e50b940454ea100

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

• Loss: 2.8058
• Data Size: 1.0
• Epoch Runtime: 32.4416
• Bleu: 0.1985

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	220.5446	0	2.6730	0.0036
No log	1	58	204.5957	0.0078	4.3093	0.0036
No log	2	116	191.1346	0.0156	3.9760	0.0036
No log	3	174	165.1904	0.0312	5.5890	0.0029
No log	4	232	132.2721	0.0625	7.4449	0.0026
No log	5	290	88.7089	0.125	10.7634	0.0013
11.694	6	348	40.9796	0.25	14.3784	0.0011
8.7816	7	406	17.6196	0.5	19.4515	0.0038
20.3087	8.0	464	10.8403	1.0	33.8301	0.0051
17.5162	9.0	522	9.2789	1.0	30.4905	0.0295
14.7784	10.0	580	8.4976	1.0	31.4386	0.0104
13.0537	11.0	638	7.9470	1.0	30.9553	0.0278
11.9767	12.0	696	7.6253	1.0	31.7243	0.0265
10.465	13.0	754	6.7038	1.0	30.7987	0.0490
9.8629	14.0	812	6.1852	1.0	31.0795	0.0481
9.1462	15.0	870	5.7621	1.0	31.1096	0.0682
8.6466	16.0	928	5.5545	1.0	30.6564	0.0612
8.209	17.0	986	5.2784	1.0	30.7827	0.0320
7.8017	18.0	1044	5.1448	1.0	31.3435	0.0801
7.2001	19.0	1102	4.7046	1.0	30.7288	0.0777
6.8989	20.0	1160	4.5510	1.0	30.7220	0.0732
6.6344	21.0	1218	4.4494	1.0	30.8027	0.1000
6.42	22.0	1276	4.3321	1.0	30.9185	0.0775
6.1573	23.0	1334	4.1810	1.0	30.7196	0.0709
6.0025	24.0	1392	4.1812	1.0	30.7153	0.0762
5.637	25.0	1450	3.9501	1.0	32.2425	0.0785
5.5103	26.0	1508	3.7917	1.0	31.7518	0.1297
5.3807	27.0	1566	3.7109	1.0	32.4725	0.0926
5.2493	28.0	1624	3.6484	1.0	32.1762	0.0779
5.0971	29.0	1682	3.7100	1.0	32.0763	0.0838
4.9723	30.0	1740	3.3917	1.0	32.5480	0.0943
4.8582	31.0	1798	3.4487	1.0	31.8977	0.0952
4.6482	32.0	1856	3.3220	1.0	32.1636	0.1071
4.5502	33.0	1914	3.3602	1.0	31.9986	0.0914
4.4548	34.0	1972	3.2248	1.0	31.7632	0.1156
4.3663	35.0	2030	3.2001	1.0	31.7785	0.1274
4.2816	36.0	2088	3.1693	1.0	32.2818	0.0931
4.2056	37.0	2146	3.0939	1.0	32.5222	0.0870
4.0655	38.0	2204	3.2121	1.0	32.3663	0.0873
3.9848	39.0	2262	2.9937	1.0	32.6995	0.1050
3.9345	40.0	2320	2.9854	1.0	32.4806	0.1002
3.8787	41.0	2378	2.9914	1.0	32.3101	0.1151
3.7917	42.0	2436	2.9243	1.0	32.5237	0.1507
3.7642	43.0	2494	2.8550	1.0	32.1596	0.1668
3.66	44.0	2552	2.9300	1.0	32.1260	0.1921
3.6051	45.0	2610	2.8908	1.0	31.8458	0.1565
3.5778	46.0	2668	2.8238	1.0	32.6554	0.1940
3.534	47.0	2726	2.8003	1.0	31.9395	0.1799
3.4921	48.0	2784	2.8139	1.0	32.5989	0.1728
3.4553	49.0	2842	2.8279	1.0	33.0087	0.2065
3.3782	50.0	2900	2.8058	1.0	32.4416	0.1985

Framework versions

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