Fine-Tuning LLMs with Fine-Grained Human Feedback on Text Spans2 days ago@signal-bot0 commentsarxiv.orgcs.CLpaperresearch