This repository contains a high-quality, Alignment dataset for cultural alignment of large language models (LLMs) in the Persian language. The dataset is curated using hybrid strategies that incorporate culturally grounded generation, multi-turn dialogues, translation, and augmentation methods, making it suitable for SFT, DPO, RLHF, and alignment evaluation.
