Identifying Good and Bad Neurons for Task-Level Controllable LLMs

TL;DR

This paper introduces NeuronLLM, a framework that identifies both supportive and inhibitive neurons in large language models to better understand and control their task performance.

Contribution

NeuronLLM applies the biological concept of functional antagonism and contrastive learning to holistically model neurons, addressing limitations of previous methods focused only on positive roles.

Findings

NeuronLLM outperforms existing methods across multiple NLP tasks.

It provides new insights into the functional organization of neurons in LLMs.

The framework is effective across different model sizes and architectures.

Abstract

Large Language Models have demonstrated remarkable capabilities on multiple-choice question answering benchmarks, but the complex mechanisms underlying their large-scale neurons remain opaque, posing significant challenges for understanding and steering LLMs. While recent studies made progress on identifying responsible neurons for certain abilities, these ability-specific methods are infeasible for task-focused scenarios requiring coordinated use of multiple abilities. Moreover, these approaches focus only on supportive neurons that correlate positively with task completion, while neglecting neurons with other roles-such as inhibitive roles-and misled neuron attribution due to fortuitous behaviors in LLMs (i.e., correctly answer the questions by chance rather than genuine understanding). To address these challenges, we propose NeuronLLM, a novel task-level LLM understanding framework that adopts the biological principle of functional antagonism for LLM neuron identification. The key insight is that task performance is jointly determined by neurons with two opposing roles: good neurons that facilitate task completion and bad neurons that inhibit it. NeuronLLM achieves a holistic modeling of neurons via contrastive learning of good and bad neurons, while leveraging augmented question sets to mitigate the fortuitous behaviors in LLMs. Comprehensive experiments on LLMs of different sizes and families show the superiority of NeuronLLM over existing methods in four NLP tasks, providing new insights into LLM functional organization.