mbart-large-50-many-to-many-mmt-finetuned-pt-to-en-arabic-hindi

This model is a fine-tuned version of facebook/mbart-large-50-many-to-many-mmt on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.8373
• Bleu: 5.0308
• Gen Len: 52.3902

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: 0.0002
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	77	2.9700	1.8614	49.3137
No log	2.0	154	2.8490	2.477	49.817
No log	3.0	231	2.9014	2.9323	51.0426
No log	4.0	308	3.0640	3.7026	62.0698
No log	5.0	385	3.2683	3.9123	49.7686
No log	6.0	462	3.4488	4.2838	52.1925
1.7841	7.0	539	3.5876	4.6755	53.1631
1.7841	8.0	616	3.7173	4.4499	48.1984
1.7841	9.0	693	3.7868	4.8237	51.9214
1.7841	10.0	770	3.8373	5.0308	52.3902

Framework versions

• Transformers 4.40.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1