A comprehensive index of artificial intelligence and machine-learning research with AI-generated summaries, citation metrics, and direct links to papers and code.
Unknown
UNIHD proposes a unified framework for detecting hallucinations in multimodal large language models, addressing the fragmented landscape of existing detection methods.
Unknown
This paper proposes a robust method for detecting hallucinations in large language model outputs, improving reliability of question answering systems.
Unknown
GraphRAG enhances LLM-driven RAG for finance by using graph-based retrieval to reduce hallucinations and improve contextual coherence.
Manuel Cossio
This paper provides a comprehensive taxonomy of LLM hallucinations, arguing their theoretical inevitability and emphasizing the need for robust detection, mitigation, and human oversight.
Yung-Sung Chuang, Linlu Qiu, Cheng-Yu Hsieh, et al.
Proposes Lookback Lens, a simple hallucination detector using attention weight ratios, effective across tasks and models, and reduces hallucinations via classifier-guided decoding.
Adi Simhi, Jonathan Herzig, Idan Szpektor, et al.
This paper distinguishes between two types of LLM hallucinations—HK- (model lacks knowledge) and HK+ (model has knowledge but answers incorrectly)—and shows that distinguishing them improves mitigation.
Sebastian Farquhar, Jannik Kossen, Lorenz Kuhn, et al.
Proposes semantic entropy to detect LLM confabulations by measuring uncertainty at the meaning level, enabling task-agnostic hallucination detection.
Zorik Gekhman, G. Yona, Roee Aharoni, et al.
This paper shows that supervised fine-tuning introducing new factual knowledge to LLMs is learned slowly and linearly increases hallucination, supporting that factual knowledge is acquired during pre-training.
Unknown
DataGemma reduces LLM hallucinations by grounding responses in Google's Data Commons statistical database using RIG and RAG methods.
Unknown
LMDX adapts arbitrary LLMs for document information extraction with layout encoding and grounding to prevent hallucinations.
Unknown
GPT-4.5 scales unsupervised learning with new alignment techniques to reduce hallucinations and improve natural conversation, while GPT-4.1 family excels in coding, instruction following, and long-context tasks at lower cost.