6431b2ccc01c99fe4ecf26fb75516f3e

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

• Loss: 1.4337
• Data Size: 1.0
• Epoch Runtime: 378.7898
• Bleu: 6.6781

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	3.4100	0	28.4126	0.2439
No log	1	684	2.4432	0.0078	31.9325	1.5877
No log	2	1368	2.3153	0.0156	37.8507	1.9091
No log	3	2052	2.2078	0.0312	47.4179	3.6468
No log	4	2736	2.0897	0.0625	59.4032	2.4129
2.4659	5	3420	1.9834	0.125	79.1582	2.4795
2.2792	6	4104	1.8620	0.25	120.1279	3.0764
2.0256	7	4788	1.7290	0.5	205.3425	3.7368
1.8058	8.0	5472	1.5855	1.0	376.6263	4.5428
1.6135	9.0	6156	1.4931	1.0	370.8074	5.1258
1.4937	10.0	6840	1.4448	1.0	371.5985	5.4812
1.3742	11.0	7524	1.4114	1.0	371.6593	5.7704
1.2846	12.0	8208	1.3902	1.0	370.8356	6.0013
1.1784	13.0	8892	1.3771	1.0	373.1421	6.2228
1.123	14.0	9576	1.3724	1.0	376.4735	6.3688
1.0134	15.0	10260	1.3827	1.0	377.0905	6.4729
0.9525	16.0	10944	1.3917	1.0	377.7570	6.6213
0.8781	17.0	11628	1.4171	1.0	379.7279	6.6494
0.8119	18.0	12312	1.4337	1.0	378.7898	6.6781

Framework versions

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