Training language models to follow instructions with human feedback

  • 类型:arxiv
  • 标识:2203.02155
  • 链接:https://arxiv.org/abs/2203.02155
  • 主题:evaluation
  • 主分类:engineering
  • 形态:method
  • 被引:23575
  • 被引来源:Semantic Scholar
  • S2被引:23575
  • OpenAlex被引:4337
  • 影响力被引:2402
  • TLDR:The results show that fine-tuning with human feedback is a promising direction for aligning language models with human intent and showing improvements in truthfulness and reductions in toxic output generation while having minimal performance regressions on public NLP datasets.
  • OpenAlex ID:W4226278401
  • OpenAlex DOI:10.48550/arxiv.2203.02155
  • DOI:10.48550/arxiv.2203.02155
  • DOI来源:OpenAlex
  • 开放获取:green
  • 开放获取链接:https://arxiv.org/pdf/2203.02155
  • OpenAlex更新:2026-08-25
  • 待LLM分类:否
  • 标题中文:使用人类反馈训练语言模型遵循指令
  • TLDR中文:结果表明,使用人类反馈进行微调是使语言模型与人类意图对齐的一个有前景的方向,在真实性方面有所提升,并减少了有毒输出的生成,同时在公开 NLP 数据集上的性能回归极小。
  • 来源文件
  • [OpenAlex discover]
  • [S2 enrich]