1. Token-wise Influential Training Data Retrieval for Large Language Models
- Author
-
Lin, Huawei, Long, Jikai, Xu, Zhaozhuo, and Zhao, Weijie
- Subjects
Computer Science - Computation and Language ,Computer Science - Artificial Intelligence ,Computer Science - Cryptography and Security ,Computer Science - Information Retrieval - Abstract
Given a Large Language Model (LLM) generation, how can we identify which training data led to this generation? In this paper, we proposed RapidIn, a scalable framework adapting to LLMs for estimating the influence of each training data. The proposed framework consists of two stages: caching and retrieval. First, we compress the gradient vectors by over 200,000x, allowing them to be cached on disk or in GPU/CPU memory. Then, given a generation, RapidIn efficiently traverses the cached gradients to estimate the influence within minutes, achieving over a 6,326x speedup. Moreover, RapidIn supports multi-GPU parallelization to substantially accelerate caching and retrieval. Our empirical result confirms the efficiency and effectiveness of RapidIn., Comment: Accepted to ACL 2024. Keywords: Influence Function, Influence Estimation, Training Data Attribution
- Published
- 2024