Skip to content
Preprint

HalluTracer: Pre-Decoding Truthfulness Prediction via Depth-Averaged Probe-Logit

Aug 2026 · 0 citations · 50 references
Computer Science

TL;DR

The proposed method HalluTracer, which averages layer-wise probe logits to predict truthfulness before decoding, achieves the highest area under the receiver operating characteristic curve (AUROC) in 23 of 24 model--benchmark pairs among the compared methods.

Abstract

Internal-state probes enable truthfulness prediction before a large language model generates an answer. When detectors change both the layers they read and the rules used to combine them, the source of improved prediction becomes difficult to identify. We separate these choices and find that retaining more layers improves prediction even under fixed equal weighting. An exact Fisher-ratio decomposition explains why the additional benefit of linear reweighting is limited on these probe scores: information inlayer-wise differences largely overlaps with that captured by the depth mean. Estimating additional weights can then offset this small benefit when the data used to fit them are limited. These findings motivate our proposed method HalluTracer, which averages layer-wise probe logits to predict truthfulness before decoding. Across six models and four benchmarks, including TruthfulQA, HalluTracer achieves the highest area under the receiver operating characteristic curve (AUROC) in 23 of 24 model--benchmark pairs among the compared methods. The results support using evidence from across the network without requiring a correspondingly more flexible aggregation rule, clarifying the distinct roles of layer selection and weighting in pre-decoding detection.

View source

Similar papers

#artificial intelligence Preprint Sep 2026

Look Before You Leap: Factual Decoding with Internal Attribution Signals

This work identifies a factual-salient layer span within LLMs whose derived signal is selectively elevated for factual tokens and exhibits anomalous spikes at hallucination-prone steps, and proposes DescaPE, a decoding framework that leverages internal model signals to suppress hallucination-prone trajectories at infer...

Hayeong Ryu, Jungmin Yun, B. Lim et al. · 0 citations
Preprint Aug 2026

XTC: Head-Aware Sampling by Excluding Top Choices

This work introduces XTC (Exclude Top Choices), a lightweight head-aware decoding operator that targets this regime of open-ended generation in which several continuations are plausible but too much probability mass remains concentrated on the most generic choice.

Philippe Weidmann, Allen G. Roush, Judah Goldfeder et al. · 0 citations
#artificial intelligence Preprint Sep 2026

Greedy Decoding Is Not Precision-Invariant: Cross-Precision Output Divergence in LLM Inference

An empirical error-propagation analysis is developed and finds that 22 layers of accumulated body error do not distinguish flipping from non-flipping steps; the outcome depends primarily on the top-two logit margin at the LM head relative to the directional perturbation between the top-two candidates.

Gao-Yuan Du, A. Khan, Rex Zhou et al. · 1 citation
#natural language process... Preprint Aug 2026

HalluTracer: Hallucination Detection via Depth-Averaging Truth Signals

HalluTracer is introduced, a detection framework that reads and aggregates truthfulness evidence across every layer of the forward pass before the model emits any answer token, recasts hallucination detection from a layer-selection problem into a depth-aggregation problem governed by the geometric sparsity of the truth...

Zhi-Hao Guo, Zong-Han Wu, Huan Huo et al. · 0 citations
#natural language process... Preprint Sep 2026

SFAD: Speculative Factuality-Aware Decoding

SFAD is presented, a speculative decoding framework that enhances contextual faithfulness without inference degradation and substantially improves faithfulness while achieving $2.48\times$ speedup, offering a practical solution for efficient LLMs.

Guan-Qiao Chen, Di Wang, Lijie Hu · 0 citations
#natural language process... Preprint Oct 2026

Identifying Introspection From the Inside

This paper identifies mechanistic signatures of faithful self-report in a controlled setting and proposes that it is possible to distinguish between the two patterns of computation by examining the structure of the networks themselves.

David Atkinson, Dillon Plunkett, David Bau · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.