3D-VCD reduces hallucinations in 3D-LLM embodied agents by contrasting predictions from original and distorted 3D scene representations at inference time.
hub Mixed citations
A Survey of Hallucination in Large Foundation Models
Mixed citation behavior. Most common role is background (60%).
26
Pith papers citing it
Background
60% of classified citations
abstract
Hallucination in a foundation model (FM) refers to the generation of content that strays from factual reality or includes fabricated information. This survey paper provides an extensive overview of recent efforts that aim to identify, elucidate, and tackle the problem of hallucination, with a particular focus on ``Large'' Foundation Models (LFMs). The paper classifies various types of hallucination phenomena that are specific to LFMs and establishes evaluation criteria for assessing the extent of hallucination. It also examines existing strategies for mitigating hallucination in LFMs and discusses potential directions for future research in this area. Essentially, the paper offers a comprehensive examination of the challenges and solutions related to hallucination in LFMs.
hub tools
citation-role summary
background 4
dataset 1
citation-polarity summary
representative citing papers
Physics-IQ benchmark reveals that generative video models exhibit limited physical understanding unrelated to their visual quality.
A novel FMECA-based framework was developed and validated for systematic assessment of patient safety risks in LLM-generated clinical discharge summaries, demonstrating moderate-to-substantial inter-rater agreement and good usability.
GraphScout trains LLMs to autonomously synthesize structured training data from knowledge graphs via flexible exploration tools, enabling a 4B model to outperform larger LLMs by 16.7% on average with fewer inference tokens and strong cross-domain transfer.
Dimension-level evaluation reveals that 25-58% of LLM outputs with perfect holistic scores still show measurable intent deficits across languages and domains.
DisAAD trains a 1%-sized proxy model via adversarial distillation to quantify uncertainty in black-box LLMs by aligning with their output distributions.
OSCAR exploits the generative-discriminative gap in LVLMs to build online preference data with MCTS and dual-granularity rewards for DPO-based calibration, claiming SOTA hallucination reduction and improved multimodal performance.
SHADE adaptively combines coverage and spectral signals to estimate semantic alphabet size from few LLM samples, yielding better performance than baselines in low-sample regimes for alphabet estimation and QA error detection.
Hallucination neurons in LLMs are domain-specific, with cross-domain classifiers dropping from AUROC 0.783 within-domain to 0.563 across domains.
Only 39% of LLM safety benchmark repositories run without modification, 6% include ethical warnings, and adoption tracks author prominence and runnability rather than code quality metrics.
SENTINEL reduces MLLM object hallucinations by over 90% via sentence-level early intervention with detector-bootstrapped preference data and C-DPO loss, outperforming prior SOTA on hallucination and capability benchmarks.
PCD redirects robotic policies toward object-relevant visual features via contrastive decoding on masked inputs, improving generalization without retraining or weight access.
SEPs approximate semantic entropy from single-generation hidden states to enable cheap and robust hallucination detection in LLMs.
LLM-generated code matches human-written code in overall readability but exhibits different issue patterns, and prompt engineering has limited impact on improving it.
LLMs exhibit substantial heterogeneity and non-determinism in SLR evidence screening, abstracts are decisive for performance, and they show no reliable superiority over classical classifiers on two real SLRs.
SinkTrack anchors LLMs to initial context by modifying the attention sink token with injected features, yielding gains on textual and multimodal tasks.
LLMs reflect users' privacy preferences in access control decisions with up to 86% agreement and can promote safer behavior, but personalization trades off higher individual match for potentially less secure results when users over-permission.
Temperature and persona variations shape consensus speed in LLM multi-agent coding but produce no robust accuracy gains over single agents on human-annotated tutoring transcripts.
The paper surveys hallucination in LLMs with an innovative taxonomy, factors, detection methods, benchmarks, mitigation strategies, and open research directions.
DAVinCI combines claim attribution to model internals and external sources with entailment-based verification to improve LLM factual reliability by 5-20% on fact-checking datasets.
Dense retrieval plus query reformulation and reranking reaches 60.49% accuracy on MedQA USMLE, outperforming other setups while domain-specialized models make better use of the retrieved evidence.
An experiment with 276 participants finds that vision language model assistance improves human game testers' defect identification, especially with design documentation, while AI errors create challenges.
A survey that compiles and taxonomizes more than 32 existing hallucination mitigation techniques for LLMs while analyzing their challenges and limitations.
A systematic review of memory designs, evaluation methods, applications, limitations, and future directions for LLM-based agents.
citing papers explorer
-
3D-VCD: Hallucination Mitigation in 3D-LLM Embodied Agents through Visual Contrastive Decoding
3D-VCD reduces hallucinations in 3D-LLM embodied agents by contrasting predictions from original and distorted 3D scene representations at inference time.
-
Do generative video models understand physical principles?
Physics-IQ benchmark reveals that generative video models exhibit limited physical understanding unrelated to their visual quality.
-
Evaluating Patient Safety Risks in Generative AI: Development and Validation of a FMECA Framework for Generated Clinical Content
A novel FMECA-based framework was developed and validated for systematic assessment of patient safety risks in LLM-generated clinical discharge summaries, demonstrating moderate-to-substantial inter-rater agreement and good usability.
-
GraphScout: Empowering Large Language Models with Intrinsic Exploration Ability for Agentic Graph Reasoning
GraphScout trains LLMs to autonomously synthesize structured training data from knowledge graphs via flexible exploration tools, enabling a 4B model to outperform larger LLMs by 16.7% on average with fewer inference tokens and strong cross-domain transfer.
-
Dimension-Level Intent Fidelity Evaluation for Large Language Models: Evidence from Structured Prompt Ablation
Dimension-level evaluation reveals that 25-58% of LLM outputs with perfect holistic scores still show measurable intent deficits across languages and domains.
-
Estimating the Black-box LLM Uncertainty with Distribution-Aligned Adversarial Distillation
DisAAD trains a 1%-sized proxy model via adversarial distillation to quantify uncertainty in black-box LLMs by aligning with their output distributions.
-
Online Self-Calibration Against Hallucination in Vision-Language Models
OSCAR exploits the generative-discriminative gap in LVLMs to build online preference data with MCTS and dual-granularity rewards for DPO-based calibration, claiming SOTA hallucination reduction and improved multimodal performance.
-
Mind the Unseen Mass: Unmasking LLM Hallucinations via Soft-Hybrid Alphabet Estimation
SHADE adaptively combines coverage and spectral signals to estimate semantic alphabet size from few LLM samples, yielding better performance than baselines in low-sample regimes for alphabet estimation and QA error detection.
-
Do Hallucination Neurons Generalize? Evidence from Cross-Domain Transfer in LLMs
Hallucination neurons in LLMs are domain-specific, with cross-domain classifiers dropping from AUROC 0.783 within-domain to 0.563 across domains.
-
Benchmark of Benchmarks: Unpacking Influence and Code Repository Quality in LLM Safety Benchmarks
Only 39% of LLM safety benchmark repositories run without modification, 6% include ethical warnings, and adoption tracks author prominence and runnability rather than code quality metrics.
-
Mitigating Object Hallucinations via Sentence-Level Early Intervention
SENTINEL reduces MLLM object hallucinations by over 90% via sentence-level early intervention with detector-bootstrapped preference data and C-DPO loss, outperforming prior SOTA on hallucination and capability benchmarks.
-
Policy Contrastive Decoding for Robotic Foundation Models
PCD redirects robotic policies toward object-relevant visual features via contrastive decoding on masked inputs, improving generalization without retraining or weight access.
-
Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMs
SEPs approximate semantic entropy from single-generation hidden states to enable cheap and robust hallucination detection in LLMs.
-
The Readability Spectrum: Patterns, Issues, and Prompt Effects in LLM-Generated Code
LLM-generated code matches human-written code in overall readability but exhibits different issue patterns, and prompt engineering has limited impact on improving it.
-
Beyond Accuracy: LLM Variability in Evidence Screening for Software Engineering SLRs
LLMs exhibit substantial heterogeneity and non-determinism in SLR evidence screening, abstracts are decisive for performance, and they show no reliable superiority over classical classifiers on two real SLRs.
-
SinkTrack: Attention Sink based Context Anchoring for Large Language Models
SinkTrack anchors LLMs to initial context by modifying the attention sink token with injected features, yielding gains on textual and multimodal tasks.
-
Can LLMs Make (Personalized) Access Control Decisions?
LLMs reflect users' privacy preferences in access control decisions with up to 86% agreement and can promote safer behavior, but personalization trades off higher individual match for potentially less secure results when users over-permission.
-
Temperature and Persona Shape LLM Agent Consensus With Minimal Accuracy Gains in Qualitative Coding
Temperature and persona variations shape consensus speed in LLM multi-agent coding but produce no robust accuracy gains over single agents on human-annotated tutoring transcripts.
-
A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions
The paper surveys hallucination in LLMs with an innovative taxonomy, factors, detection methods, benchmarks, mitigation strategies, and open research directions.
-
Trust but Verify: Introducing DAVinCI -- A Framework for Dual Attribution and Verification in Claim Inference for Language Models
DAVinCI combines claim attribution to model internals and external sources with entailment-based verification to improve LLM factual reliability by 5-20% on fact-checking datasets.
-
A Systematic Study of Retrieval Pipeline Design for Retrieval-Augmented Medical Question Answering
Dense retrieval plus query reformulation and reranking reaches 60.49% accuracy on MedQA USMLE, outperforming other setups while domain-specialized models make better use of the retrieved evidence.
-
Human-AI Collaborative Game Testing with Vision Language Models
An experiment with 276 participants finds that vision language model assistance improves human game testers' defect identification, especially with design documentation, while AI errors create challenges.
-
A Comprehensive Survey of Hallucination Mitigation Techniques in Large Language Models
A survey that compiles and taxonomizes more than 32 existing hallucination mitigation techniques for LLMs while analyzing their challenges and limitations.
-
A Survey on the Memory Mechanism of Large Language Model based Agents
A systematic review of memory designs, evaluation methods, applications, limitations, and future directions for LLM-based agents.
-
Data-Centric Foundation Models in Computational Healthcare: A Survey
The paper surveys data-centric strategies for foundation models in computational healthcare and supplies a curated list of related models and datasets.
- Chain of Evidence: Pixel-Level Visual Attribution for Iterative Retrieval-Augmented Generation