We released a high-quality parallel sentiment corpus on 10 low-resource languages, NusaX dataset. Check the paper.