Look Before You Leap: Factual Decoding with Internal Attribution Signals

arXiv:2609.15745 · cs.CL, cs.AI · Submitted 2026-09-14 · Read on arXiv

cs.CL, cs.AI

Submitted: 2026-09-14

Updated: 2026-09-14

Comments: EMNLP 2026 Findings

Code: https://github.com/hayeonggg/DESCAPE

License: http://creativecommons.org/licenses/by/4.0/

The gist: Hallucination remains a critical challenge in large language models (LLMs), where early factual errors compound through autoregressive generation in a snowballing effect that neither post-hoc

Terminology

Abstract

Hallucination remains a critical challenge in large language models (LLMs), where early factual errors compound through autoregressive generation in a snowballing effect that neither post-hoc correction nor weight-level intervention can effectively preempt. We propose DescaPE (DEcoding Signal Control Against Path Error-snowballing), a decoding framework that leverages internal model signals to suppress hallucination-prone trajectories at inference time. Through sliding-window MLP ablation, we identify a factual-salient layer span within LLMs whose derived signal is selectively elevated for factual tokens and exhibits anomalous spikes at hallucination-prone steps. We train a lightweight probe to approximate this signal from a single forward pass and integrate it into candidate scoring to penalize high-risk continuations while rewarding factually grounded ones. Experiments across five factuality benchmarks on three LLMs demonstrate that DescaPE achieves factuality improvements over decoding-time baselines in multiple settings, while incurring only 1.10x latency overhead in our efficiency evaluation. Our code is available at https://github.com/hayeonggg/DESCAPE.

Sources

Related papers