Learning from others' mistakes: Finetuning machine translation models with span-level error annotations

Open in new window