Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2602.11167 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914323926876160 |
|---|---|
| author | Mao, Nathan Kaushik, Varun Shivkumar, Shreya Sharafoleslami, Parham Zhu, Kevin Dev, Sunishchal |
| author_facet | Mao, Nathan Kaushik, Varun Shivkumar, Shreya Sharafoleslami, Parham Zhu, Kevin Dev, Sunishchal |
| contents | Large Language Models (LLMs) often hallucinate, generating nonsensical or false information that can be especially harmful in sensitive fields such as medicine or law. To study this phenomenon systematically, we introduce FalseCite, a curated dataset designed to capture and benchmark hallucinated responses induced by misleading or fabricated citations. Running GPT-4o-mini, Falcon-7B, and Mistral 7-B through FalseCite, we observed a noticeable increase in hallucination activity for false claims with deceptive citations, especially in GPT-4o-mini. Using the responses from FalseCite, we can also analyze the internal states of hallucinating models, visualizing and clustering the hidden state vectors. From this analysis, we noticed that the hidden state vectors, regardless of hallucination or non-hallucination, tend to trace out a distinct horn-like shape. Our work underscores FalseCite's potential as a foundation for evaluating and mitigating hallucinations in future LLM research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_11167 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Visualizing and Benchmarking LLM Factual Hallucination Tendencies via Internal State Analysis and Clustering Mao, Nathan Kaushik, Varun Shivkumar, Shreya Sharafoleslami, Parham Zhu, Kevin Dev, Sunishchal Computation and Language Artificial Intelligence Large Language Models (LLMs) often hallucinate, generating nonsensical or false information that can be especially harmful in sensitive fields such as medicine or law. To study this phenomenon systematically, we introduce FalseCite, a curated dataset designed to capture and benchmark hallucinated responses induced by misleading or fabricated citations. Running GPT-4o-mini, Falcon-7B, and Mistral 7-B through FalseCite, we observed a noticeable increase in hallucination activity for false claims with deceptive citations, especially in GPT-4o-mini. Using the responses from FalseCite, we can also analyze the internal states of hallucinating models, visualizing and clustering the hidden state vectors. From this analysis, we noticed that the hidden state vectors, regardless of hallucination or non-hallucination, tend to trace out a distinct horn-like shape. Our work underscores FalseCite's potential as a foundation for evaluating and mitigating hallucinations in future LLM research. |
| title | Visualizing and Benchmarking LLM Factual Hallucination Tendencies via Internal State Analysis and Clustering |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2602.11167 |