arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.09365cs.AIcs.CL
收藏

概念蒸馏:通过假设到理论提示从强模型到弱模型

Concept Distillation from Strong to Weak Models via Hypotheses-to-Theories Prompting

Microsoft(微软)

Emmanuel Aboah Boateng, Cassiano O. Becker, Nabiha Asghar, Kabir Walia, Ashwin Srinivasan, Ehi Nosakhare, Soundar Srinivasan, Victor Dibia

AI总结:

提出概念蒸馏(CD)自动提示优化方法,通过收集弱模型错误、强模型生成规则并验证过滤,显著提升弱模型在NL2Code和数学推理任务上的性能,如Mistral-7B准确率提高20%。

摘要:

手工制作高质量提示词以优化语言模型性能是一个复杂且劳动密集的过程。此外,当迁移到更新、更小或更弱的模型(可能出于延迟或成本考虑)时,需要更新提示词以重新优化任务性能。我们提出概念蒸馏(CD),一种自动提示优化技术,用于增强较弱模型在复杂任务上的表现。CD包括:(1)收集弱模型使用基础提示词(初始化)所犯的错误,(2)使用强模型为这些错误生成原因并为弱模型创建规则/概念(归纳),以及(3)基于验证集性能过滤这些规则并将其整合到基础提示词中(演绎/验证)。我们在NL2Code和数学推理任务上评估了CD,观察到小型和较弱语言模型性能的显著提升。值得注意的是,Mistral-7B在Multi-Arith上的准确率提高了20%,Phi-3-mini-3.8B在HumanEval上的准确率提高了34%。与其他自动化方法相比,CD提供了一种有效且成本高效的策略,用于提升弱模型在复杂任务上的表现,并实现跨不同语言模型的无缝工作负载迁移,而不损害性能。

英文摘要:

Hand-crafting high quality prompts to optimize the performance of language models is a complicated and labor-intensive process. Furthermore, when migrating to newer, smaller, or weaker models (possibly due to latency or cost gains), prompts need to be updated to re-optimize the task performance. We propose Concept Distillation (CD), an automatic prompt optimization technique for enhancing weaker models on complex tasks. CD involves: (1) collecting mistakes made by weak models with a base prompt (initialization), (2) using a strong model to generate reasons for these mistakes and create rules/concepts for weak models (induction), and (3) filtering these rules based on validation set performance and integrating them into the base prompt (deduction/verification). We evaluated CD on NL2Code and mathematical reasoning tasks, observing significant performance boosts for small and weaker language models. Notably, Mistral-7B's accuracy on Multi-Arith increased by 20%, and Phi-3-mini-3.8B's accuracy on HumanEval rose by 34%. Compared to other automated methods, CD offers an effective, cost-efficient strategy for improving weak models' performance on complex tasks and enables seamless workload migration across different language models without compromising performance.

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。

↑