SymLoc: 跨HaluEval和TruthfulQA的幻觉符号定位
SymLoc: Symbolic Localization of Hallucination across HaluEval and TruthfulQA
AI总结：
针对LLMs在符号触发因素下的幻觉问题，提出首个符号定位框架SymLoc，利用符号语言知识追踪模型层中幻觉发展，发现早期层符号语义处理崩溃，且符号知识是理解幻觉机制的关键。
摘要：
大语言模型（LLMs）在面对修饰语、否定、数字、例外和命名实体等符号触发因素时，仍难以解决幻觉问题。然而，我们对这些符号幻觉的起源缺乏清晰理解，因此系统处理此类触发因素并在模型内部定位幻觉的出现至关重要。尽管先前的工作使用LSC和激活方差分析等统计技术探索了定位方法，但这些方法平等对待所有标记，忽视了符号语言知识在触发幻觉中的作用。迄今为止，尚无方法研究符号元素如何在模型层中具体导致幻觉失败，也未将符号语言知识作为定位框架的基础。我们提出首个符号定位框架SymLoc，利用符号语言和语义知识有意义地追踪所有模型层中幻觉的发展。通过关注模型如何处理符号触发因素，我们使用HaluEval和TruthfulQA分析了五个模型。我们的符号知识方法揭示，这些语言元素的注意力方差在早期层（2-4层）爆炸至临界不稳定状态，其中否定触发灾难性方差水平，表明符号语义处理从一开始就崩溃。通过符号语言知识的视角，尽管模型规模更大，幻觉率仍持续较高（Gemma变体的幻觉率为78.3%-83.7%），且符号语义触发因素的注意力在更深层中急剧下降。我们的研究结果表明，幻觉本质上是符号语言处理失败，而非一般生成问题，揭示符号语义知识是理解和定位LLMs幻觉机制的关键。
英文摘要：
LLMs still struggle with hallucination, especially when confronted with symbolic triggers like modifiers, negation, numbers, exceptions, and named entities. Yet, we lack a clear understanding of where these symbolic hallucinations originate, making it crucial to systematically handle such triggers and localize the emergence of hallucination inside the model. While prior work explored localization using statistical techniques like LSC and activation variance analysis, these methods treat all tokens equally and overlook the role symbolic linguistic knowledge plays in triggering hallucinations. So far, no approach has investigated how symbolic elements specifically drive hallucination failures across model layers, nor has symbolic linguistic knowledge been used as the foundation for a localization framework. We propose the first symbolic localization framework that leverages symbolic linguistic and semantic knowledge to meaningfully trace the development of hallucinations across all model layers. By focusing on how models process symbolic triggers, we analyze five models using HaluEval and TruthfulQA. Our symbolic knowledge approach reveals that attention variance for these linguistic elements explodes to critical instability in early layers (2-4), with negation triggering catastrophic variance levels, demonstrating that symbolic semantic processing breaks down from the very beginning. Through the lens of symbolic linguistic knowledge, despite larger model sizes, hallucination rates remain consistently high (78.3%-83.7% across Gemma variants), with steep attention drops for symbolic semantic triggers throughout deeper layers. Our findings demonstrate that hallucination is fundamentally a symbolic linguistic processing failure, not a general generation problem, revealing that symbolic semantic knowledge provides the key to understanding and localizing hallucination mechanisms in LLMs.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
