方言?我几乎不了解:风格控制与刻板印象相关的挑战
Vernacular? I Barely Know Her: Challenges with Style Control and Stereotyping
AI总结:
本研究针对大语言模型在教育场景的风格控制需求,评估5款前沿模型在两项风格控制任务中的表现,发现模型存在风格控制能力不足、输出不一致及易生成文化不敏感内容等问题,并展开了全面分析。
摘要:
大语言模型(Large Language Models,LLMs)正越来越多地被应用于教育与学习类场景。已有研究表明,针对学习者需求调整文本风格能够提升理解效果、促进包容性并助力知识蒸馏。为探究当前LLM在风格控制方面的能力与局限,我们在两项风格控制任务中评估了5个前沿模型:GPT-3.5、GPT-4、GPT-4o、Llama-3以及Mistral-instruct-7B。我们发现第一项任务中模型表现存在显著不一致性:面向一年级学生的任务,模型输出的阅读水平平均处于5至8年级区间,标准差最高达27.6。第二项任务中,模型表现从0.02提升至0.26,具有统计学显著性。但我们也发现,即便参考文本中不存在刻板印象,LLM在执行任务时仍经常生成文化不敏感内容。我们对结果进行了全面的分析与讨论。
英文摘要:
Large Language Models (LLMs) are increasingly being used in educational and learning applications. Research has demonstrated that controlling for style, to fit the needs of the learner, fosters increased understanding, promotes inclusion, and helps with knowledge distillation. To understand the capabilities and limitations of contemporary LLMs in style control, we evaluated five state-of-the-art models: GPT-3.5, GPT-4, GPT-4o, Llama-3, and Mistral-instruct- 7B across two style control tasks. We observed significant inconsistencies in the first task, with model performances averaging between 5th and 8th grade reading levels for tasks intended for first-graders, and standard deviations up to 27.6. For our second task, we observed a statistically significant improvement in performance from 0.02 to 0.26. However, we find that even without stereotypes in reference texts, LLMs often generated culturally insensitive content during their tasks. We provide a thorough analysis and discussion of the results.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。