Do Localization Methods Actually Localize Memorized Data in LLMs? A Tale of Two Benchmarks

Anonymous

TL;DR: We propose two benchmarking approaches to systematically and directly evaluate localization methods

Abstract: The concept of localization in LLMs is often mentioned in prior work; however, methods for localization have never been systematically and directly evaluated. We propose two complementary benchmarks that evaluate the ability of localization methods to pinpoint LLM components responsible for memorized data. In our INJ benchmark, we actively inject a piece of new information into a small subset of LLM weights, enabling us to directly evaluate whether localization methods can identify these "ground truth'' weights. In our DEL benchmark, we evaluate localization by measuring how much dropping out identified neurons deletes a memorized pretrained sequence. Despite their different perspectives, our two benchmarks yield consistent rankings of five localization methods. Methods adapted from network pruning perform well on both benchmarks, and all evaluated methods show promising localization ability. On the other hand, even successful methods identify neurons that are not specific to a single memorized sequence.

Paper Type: long

Research Area: Interpretability and Analysis of Models for NLP

Contribution Types: Model analysis & interpretability

Languages Studied: English