Pith. sign in

REVIEW 2 cited by

TruthX: Alleviating Hallucinations by Editing Large Language Models in Truthful Space

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2402.17811 v2 pith:PYDMROJ6 submitted 2024-02-27 cs.CL cs.AIcs.LG

TruthX: Alleviating Hallucinations by Editing Large Language Models in Truthful Space

classification cs.CL cs.AIcs.LG
keywords truthfultruthxeditingtruthfulnessrepresentationsinternalllmsspace
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Large Language Models (LLMs) sometimes suffer from producing hallucinations, especially LLMs may generate untruthful responses despite knowing the correct knowledge. Activating the truthfulness within LLM is the key to fully unlocking LLM's knowledge potential. In this paper, we propose TruthX, an inference-time intervention method to activate the truthfulness of LLM by identifying and editing the features within LLM's internal representations that govern the truthfulness. TruthX employs an auto-encoder to map LLM's representations into semantic and truthful latent spaces respectively, and applies contrastive learning to identify a truthful editing direction within the truthful space. During inference, by editing LLM's internal representations in truthful space, TruthX effectively enhances the truthfulness of LLM. Experiments show that TruthX improves the truthfulness of 13 advanced LLMs by an average of 20% on TruthfulQA benchmark. Further analyses suggest that TruthX can control LLM to produce truthful or hallucinatory responses via editing only one vector in LLM's internal representations.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. DICA: Dual-Indicator Guided Contrastive Alignment in Multimodal Large Language Models

    cs.AI 2026-07 conditional novelty 6.0

    Monitoring attention entropy and image-output correlation during decoding, then applying targeted contrastive corrections, reduces hallucination in multimodal LLMs without retraining.

  2. ToxiREX: A Dataset on Toxic REasoning in ConteXt

    cs.CL 2026-06 unverdicted novelty 6.0

    ToxiREX is a new dataset of 128k Reddit comments in six languages with hierarchical annotations for implicit toxicity in conversational context based on an existing reasoning schema.