Research

Paper

AI LLM February 26, 2026

Probing for Knowledge Attribution in Large Language Models

Authors

Ivo Brink, Alexander Boer, Dennis Ulmer

Abstract

Large language models (LLMs) often generate fluent but unfounded claims, or hallucinations, which fall into two types: (i) faithfulness violations - misusing user context - and (ii) factuality violations - errors from internal knowledge. Proper mitigation depends on knowing whether a model's answer is based on the prompt or its internal weights. This work focuses on the problem of contributive attribution: identifying the dominant knowledge source behind each output. We show that a probe, a simple linear classifier trained on model hidden representations, can reliably predict contributive attribution. For its training, we introduce AttriWiki, a self-supervised data pipeline that prompts models to recall withheld entities from memory or read them from context, generating labelled examples automatically. Probes trained on AttriWiki data reveal a strong attribution signal, achieving up to 0.96 Macro-F1 on Llama-3.1-8B, Mistral-7B, and Qwen-7B, transferring to out-of-domain benchmarks (SQuAD, WebQuestions) with 0.94-0.99 Macro-F1 without retraining. Attribution mismatches raise error rates by up to 70%, demonstrating a direct link between knowledge source confusion and unfaithful answers. Yet, models may still respond incorrectly even when attribution is correct, highlighting the need for broader detection frameworks.

Metadata

arXiv ID: 2602.22787
Provider: ARXIV
Primary Category: cs.CL
Published: 2026-02-26
Fetched: 2026-02-27 04:35

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2602.22787v1</id>\n    <title>Probing for Knowledge Attribution in Large Language Models</title>\n    <updated>2026-02-26T09:21:12Z</updated>\n    <link href='https://arxiv.org/abs/2602.22787v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2602.22787v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Large language models (LLMs) often generate fluent but unfounded claims, or hallucinations, which fall into two types: (i) faithfulness violations - misusing user context - and (ii) factuality violations - errors from internal knowledge. Proper mitigation depends on knowing whether a model's answer is based on the prompt or its internal weights. This work focuses on the problem of contributive attribution: identifying the dominant knowledge source behind each output. We show that a probe, a simple linear classifier trained on model hidden representations, can reliably predict contributive attribution. For its training, we introduce AttriWiki, a self-supervised data pipeline that prompts models to recall withheld entities from memory or read them from context, generating labelled examples automatically. Probes trained on AttriWiki data reveal a strong attribution signal, achieving up to 0.96 Macro-F1 on Llama-3.1-8B, Mistral-7B, and Qwen-7B, transferring to out-of-domain benchmarks (SQuAD, WebQuestions) with 0.94-0.99 Macro-F1 without retraining. Attribution mismatches raise error rates by up to 70%, demonstrating a direct link between knowledge source confusion and unfaithful answers. Yet, models may still respond incorrectly even when attribution is correct, highlighting the need for broader detection frameworks.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <published>2026-02-26T09:21:12Z</published>\n    <arxiv:primary_category term='cs.CL'/>\n    <author>\n      <name>Ivo Brink</name>\n    </author>\n    <author>\n      <name>Alexander Boer</name>\n    </author>\n    <author>\n      <name>Dennis Ulmer</name>\n    </author>\n  </entry>"
}