dda9a487749a74412beb822d916b19ee

This model is a fine-tuned version of facebook/mbart-large-50-one-to-many-mmt on the Helsinki-NLP/opus_books [en-no] dataset. It achieves the following results on the evaluation set:

• Loss: 3.4443
• Data Size: 1.0
• Epoch Runtime: 25.1438
• Bleu: 12.9591

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	9.1304	0	2.3210	0.6058
No log	1	87	8.0377	0.0078	2.9462	0.8765
No log	2	174	7.3611	0.0156	4.4665	1.1308
No log	3	261	6.8949	0.0312	5.7543	1.3981
No log	4	348	6.2328	0.0625	7.9787	1.6412
0.2835	5	435	5.2273	0.125	9.9717	2.1739
1.2167	6	522	4.2237	0.25	12.5629	3.1708
1.3321	7	609	3.5774	0.5	15.7342	4.7647
1.8028	8.0	696	3.1783	1.0	27.0919	9.6086
2.25	9.0	783	3.0740	1.0	26.5085	13.1514
1.7315	10.0	870	3.0921	1.0	25.3620	11.1121
1.2774	11.0	957	3.2290	1.0	25.7309	14.7836
0.9691	12.0	1044	3.3519	1.0	25.5203	11.8644
0.6823	13.0	1131	3.4443	1.0	25.1438	12.9591

Framework versions

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