- The project involves creating a transformer-based classifier for a multilingual text classification task. The chosen corpus is amazon_reviews_multi. Labels include a Star rating of 1–5 in English, German, Spanish, French, Japanese, and Chinese languages. Subset sizes (per language) are; train:200K, validation:5K, and test:5K.