1fa145fa8b67260aa444d6a25f73ece2

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

• Loss: 2.0124
• Data Size: 1.0
• Epoch Runtime: 29.7277
• Bleu: 3.2004

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	6.3244	0	2.6261	0.1038
No log	1	58	5.4778	0.0078	3.3770	0.1298
No log	2	116	4.1332	0.0156	4.0611	0.1329
No log	3	174	3.4639	0.0312	6.3753	0.1487
No log	4	232	3.1592	0.0625	7.3335	0.4491
No log	5	290	2.9883	0.125	10.2179	0.5884
0.315	6	348	2.8525	0.25	15.1703	0.5602
0.4281	7	406	2.6831	0.5	20.7401	0.9285
2.0256	8.0	464	2.5202	1.0	33.3593	1.0862
2.6656	9.0	522	2.4199	1.0	31.5950	1.4794
2.5431	10.0	580	2.3434	1.0	30.9194	1.5444
2.4782	11.0	638	2.2809	1.0	31.4412	1.8279
2.3984	12.0	696	2.2351	1.0	31.5773	1.9576
2.262	13.0	754	2.1898	1.0	31.0741	2.1180
2.2022	14.0	812	2.1611	1.0	32.1944	2.3627
2.1468	15.0	870	2.1368	1.0	31.0629	2.3570
2.0988	16.0	928	2.1098	1.0	31.1569	2.5070
2.0671	17.0	986	2.0967	1.0	30.5432	2.7344
2.0133	18.0	1044	2.0732	1.0	30.4927	2.8833
1.9444	19.0	1102	2.0581	1.0	30.2351	2.9093
1.8903	20.0	1160	2.0422	1.0	30.6885	3.0504
1.8688	21.0	1218	2.0334	1.0	30.6839	2.9917
1.8279	22.0	1276	2.0312	1.0	29.5904	3.0427
1.7933	23.0	1334	2.0237	1.0	31.0217	3.1269
1.7654	24.0	1392	2.0120	1.0	31.0672	3.1205
1.7111	25.0	1450	2.0115	1.0	31.6891	3.1282
1.6784	26.0	1508	2.0074	1.0	29.6392	3.1184
1.6338	27.0	1566	2.0056	1.0	32.5466	3.1139
1.6191	28.0	1624	2.0007	1.0	30.7501	3.2825
1.5961	29.0	1682	2.0035	1.0	30.3288	3.2258
1.5631	30.0	1740	2.0008	1.0	30.8249	3.2848
1.5464	31.0	1798	2.0033	1.0	30.2752	3.1802
1.4867	32.0	1856	2.0124	1.0	29.7277	3.2004

Framework versions

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