64848445a073ecf83e4d032a73a73ec9

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

• Loss: 1.8152
• Data Size: 1.0
• Epoch Runtime: 256.4530
• Bleu: 10.4892

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	7.2798	0	21.2076	0.5041
No log	1	1000	2.7993	0.0078	23.2646	5.5699
No log	2	2000	2.5094	0.0156	25.8455	6.3666
No log	3	3000	2.2946	0.0312	30.1900	6.5796
0.0866	4	4000	2.1342	0.0625	37.4448	7.5861
2.0707	5	5000	1.9838	0.125	52.0502	8.5033
0.1158	6	6000	1.8358	0.25	81.5101	8.9089
0.1451	7	7000	1.6992	0.5	140.9315	9.9023
1.4123	8.0	8000	1.5730	1.0	257.4517	10.4943
1.1511	9.0	9000	1.5898	1.0	255.5487	9.5679
0.9267	10.0	10000	1.6187	1.0	257.8549	10.7581
0.7436	11.0	11000	1.7209	1.0	256.6519	11.2735
0.5794	12.0	12000	1.8152	1.0	256.4530	10.4892

Framework versions

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