mbart-latin-to-italian

This model is a fine-tuned version of facebook/mbart-large-50 on the Latino-italiano traduzioni dirette dataset. It achieves the following results on the evaluation set:

• Loss: 2.8924
• Bleu: 5.0678
• Gen Len: 101.961

Model description

This model intends to translate from Latin quotes to Italian.

Intended uses & limitations

This model has been created as an in-class exercice, so its coverage might be limited.

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5.6e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	92	3.1007	3.5723	103.6976
No log	2.0	184	2.8924	5.0678	101.961

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0