yousef-khoubrane/tulu-v2-sft-mixture-role-split-english-darija

This dataset is derived from Tulu-v2-SFT-Mixture-English-Darija. It splits the original data by role (user and assistant), creating a separate row for each role with its corresponding English and Darija content. Dataset: Original source (e.g., flan_v2, sharegpt). ID: Unique identifier from the original dataset. Role: user or assistant. English: English content for the role. Darija: Translated Darija content for the role.