7d82a74d0aee191aebfd57ce210f640d

This model is a fine-tuned version of facebook/mbart-large-50 on the Helsinki-NLP/opus_books [de-it] dataset. It achieves the following results on the evaluation set:

• Loss: 2.7553
• Data Size: 1.0
• Epoch Runtime: 174.2353
• Bleu: 7.4299

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	7.8905	0	14.6119	0.5081
No log	1	684	5.0743	0.0078	17.3059	2.1218
No log	2	1368	4.5363	0.0156	18.9241	3.3058
No log	3	2052	3.4038	0.0312	22.7737	5.1173
No log	4	2736	2.4219	0.0625	27.2705	7.6940
10.2512	5	3420	8.7716	0.125	37.4445	0.0009
6.0003	6	4104	2.8846	0.25	56.9574	6.2464
2.3332	7	4788	2.2471	0.5	95.9291	10.7889
2.124	8.0	5472	2.5019	1.0	176.0536	8.0866
1.6775	9.0	6156	2.1168	1.0	174.2897	9.2319
1.3396	10.0	6840	2.1073	1.0	172.4781	8.1862
1.0683	11.0	7524	2.2048	1.0	173.1780	7.1304
0.8422	12.0	8208	2.3845	1.0	172.7134	7.6362
0.6473	13.0	8892	2.5822	1.0	174.0057	7.5613
0.5065	14.0	9576	2.7553	1.0	174.2353	7.4299

Framework versions

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