|
--- |
|
language: |
|
- en |
|
- ar |
|
--- |
|
Model Description : |
|
This model is a fine-tuned version of Helsinki-en-ar, adapted to translate text from English to Darija (Moroccan Arabic) . The fine-tuning process was conducted on a substantial dataset. |
|
|
|
Fine-tuning Details |
|
Source Model: Helsinki-NLP/opus-mt-en-ar |
|
Fine-tuning Objective: To adapt the pre-existing English to Arabic translation model to perform translations from English to Arabic Darija . |
|
Dataset: The model was fine-tuned using the Darija Open Dataset (DODa), an open-source project dedicated to the Moroccan dialect. DODa contains approximately 150,000 entries, making it one of the largest open-source collaborative projects for Darija <=> English translation aimed at Natural Language Processing (NLP) applications. |
|
Training Examples: More than 15,000 translation pairs from Darija to Arabic were used for fine-tuning. |
|
Training Time: The fine-tuning process took approximately 8 hours to complete |
