提出可扩展的多阶段影响函数，解释大模型下游预测对预训练数据的依赖。
Scalable Multi-Stage Influence Function for Large Language Models via Eigenvalue-Corrected Kronecker-Factored Parameterization
- 用修正特征值的克罗内克分解参数化提升计算效率
- 在30亿参数模型上实现全参数微调下的多阶段影响分析
- 适合需要解释大模型决策过程的研究者和开发者
预训练的大语言模型（LLM）通常通过微调适应下游任务。由于大部分知识来自预训练阶段，将微调后模型的预测归因于其预训练数据可提供重要洞见。影响函数被提出用于基于训练数据解释模型预测，但现有方法无法计算多阶段影响，且难以扩展至十亿级参数的LLM。本文提出多阶段影响函数，用于在全参数微调范式下，将微调后LLM的下游预测归因于预训练数据。为提升效率与实用性，我们采用修正特征值的克罗内克分解（EK-FAC）参数化进行高效近似。实验验证了EK-FAC近似的优越可扩展性及所提方法的有效性。在真实世界模型dolly-v2-3b上的案例研究展示了其解释能力，多个实例揭示了多阶段影响估计提供的洞察。代码已公开于https://github.com/colored-dye/multi_stage_influence_function。
原文摘要 · Abstract (English)
Pre-trained large language models (LLMs) are commonly fine-tuned to adapt to downstream tasks. Since the majority of knowledge is acquired during pre-training, attributing the predictions of fine-tuned LLMs to their pre-training data may provide valuable insights. Influence functions have been proposed as a means to explain model predictions based on training data. However, existing approaches fail to compute ``multi-stage'' influence and lack scalability to billion-scale LLMs. In this paper, we propose the multi-stage influence function to attribute the downstream predictions of fine-tuned LLMs to pre-training data under the full-parameter fine-tuning paradigm. To enhance the efficiency and practicality of our multi-stage influence function, we leverage Eigenvalue-corrected Kronecker-Factored (EK-FAC) parameterization for efficient approximation. Empirical results validate the superior scalability of EK-FAC approximation and the effectiveness of our multi-stage influence function. Additionally, case studies on a real-world LLM, dolly-v2-3b, demonstrate its interpretive power, with exemplars illustrating insights provided by multi-stage influence estimates. Our code is public at https://github.com/colored-dye/multi_stage_influence_function.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。