From Unaligned to Aligned: Scaling Multilingual LLMs with Multi-Way Parallel Corpora

Open in new window