Research notes, dataset releases, and project updates on multilingual AI, machine translation, and culturally aligned language models.
Alexandria is a human-translated Dialectal Arabic machine translation dataset for English-Arabic MT, low-resource machine translation, Arabic dialect benchmarking, and LLM evaluation across 13 Arab countries.
WVS2Persona is a Hugging Face dataset that turns World Values Survey Wave 7 respondent records into textual personas for culture-aware AI, persona-based prompting, and cultural alignment research.
Unsupervised machine translation still matters in the LLM era. This post explains how self-mined in-context examples can improve translation for low-resource languages without parallel data.
This article summarizes an LLM cultural alignment pipeline in NileChat, covering cultural bias in LLMs, synthetic data generation for LLMs, and the evaluation of cultural value alignment in LLM systems.