7c3b09f6e6fc44d4e2eeff9b701f1ee9

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [de-it] dataset. It achieves the following results on the evaluation set:

• Loss: 2.3003
• Data Size: 1.0
• Epoch Runtime: 180.6288
• Bleu: 6.8396

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	9.8790	0	16.2297	0.1133
No log	1	684	3.8498	0.0078	18.2437	2.1707
No log	2	1368	3.4255	0.0156	18.9680	3.0434
No log	3	2052	3.0299	0.0312	23.3534	3.4203
No log	4	2736	2.7125	0.0625	28.3892	4.7164
2.7378	5	3420	2.4868	0.125	37.9407	8.8320
2.4948	6	4104	2.3204	0.25	59.0587	9.3806
2.225	7	4788	2.2569	0.5	100.0037	8.3113
2.0896	8.0	5472	2.0968	1.0	180.5144	8.3858
1.7805	9.0	6156	2.1456	1.0	179.6898	8.6006
1.5797	10.0	6840	2.1373	1.0	180.4057	7.0408
1.3705	11.0	7524	2.4689	1.0	179.4391	5.8947
1.1805	12.0	8208	2.3003	1.0	180.6288	6.8396

Framework versions

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