90f71679ad42a657d80dacf2f8670bc7

This model is a fine-tuned version of facebook/mbart-large-50-many-to-one-mmt on the Helsinki-NLP/opus_books [en-nl] dataset. It achieves the following results on the evaluation set:

• Loss: 2.3664
• Data Size: 1.0
• Epoch Runtime: 242.6549
• Bleu: 8.3584

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	6.3426	0	20.3799	0.4361
No log	1	966	4.9545	0.0078	22.3447	0.7729
No log	2	1932	4.2581	0.0156	25.9428	1.5656
0.1	3	2898	3.7704	0.0312	29.6188	2.4000
3.596	4	3864	3.3447	0.0625	36.9926	3.2119
3.1505	5	4830	2.9704	0.125	50.1914	4.3825
2.7232	6	5796	2.6179	0.25	77.9443	5.1534
2.3169	7	6762	2.3342	0.5	133.0972	6.1354
1.9759	8.0	7728	2.1105	1.0	243.3934	7.2763
1.6833	9.0	8694	2.0346	1.0	242.4714	7.7840
1.434	10.0	9660	2.0598	1.0	241.4339	8.9138
1.154	11.0	10626	2.1293	1.0	240.5921	8.7272
0.98	12.0	11592	2.2297	1.0	241.4105	8.6509
0.7896	13.0	12558	2.3664	1.0	242.6549	8.3584

Framework versions

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