WASD: Locating Critical Neurons as Sufficient Conditions for Explaining and Controlling LLM Behavior

arXiv:2603.18474v2 Announce Type: replace Abstract: Precise behavioral control of large language models (LLMs) is critical for complex applications. However, existing methods often incur high training costs, lack natural language controllability, or compromise semantic coherence. To bridge this gap, we propose WASD (unWeaving Actionable Sufficient Directives), a novel framework that explains model behavior by identifying sufficient neural conditions for token generation. Our method represents candidate conditions as neuron-activation predicates and iteratively searches for a minimal set that guarantees the current output under input perturbations. Experiments on SST-2 and CounterFact with the Gemma-2-2B model demonstrate that our approach produces explanations that are more stable, accurate, and concise than conventional attribution graphs. Moreover, through a case study on controlling cross-lingual output generation, we validated the practical effectiveness of WASD in controlling model behavior.

Related

• GroupGPT: A Token-efficient and Privacy-preserving Agentic Framework for Multi-User Chat Assistant
• Towards Real-world Human Behavior Simulation: Benchmarking Large Language Models on Long-horizon, Cross-scenario, Heterogeneous Behavior Traces
• ModeX: Evaluator-Free Best-of-N Selection for Open-Ended Generation
• PIKA: Expert-Level Synthetic Datasets for Post-Training Alignment from Scratch
• Distributed Multi-Layer Editing for Rule-Level Knowledge in Large Language Models

Source: arXiv cs.CL | 2026-04-10