Note A collection of moroccan darija texts (155M tokens). Can be used for pretraining Moroccan Darija LMs.

Note A culturally aligned translation benchmark for evaluating Machine Translation for Moroccan Darija.

Note A collection of 350,000 high quality (english, moroccan darija) pairs.

Note A collection of 12,743 parallel text and speech samples for Moroccan Darija, including its transcription in both Latin and Arabic scripts and English translations.

Note A collection of social media data.

Note A collection of 190,000 synthetically generated (using Gemini-2.0-Flash) text in 26 topics. Can be used to train text classification models.

Note A collection of 10,044 parallel text samples of Moroccan Darija sourced from Darija Wikipedia.

Note A collection of 551 parallel text and speech samples of Moroccan Darija sourced from Wikipedia Darija.

Note A sentence embedding score benchmark for Moroccan Darija.