An agentic architecture with multimodal screening, a five-agent jury, meta-synthesis, and source attribution protocol detects biases in Romanian history textbooks more accurately than zero-shot baselines, achieving 83.3% acceptable excerpts and human preference in 64.8% of blind comparisons.
Siren’s Song in the AI Ocean: A Survey on Hallucination in Large Language Models
11 Pith papers cite this work. Polarity classification is still indexing.
11
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
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.
LLMs flag byte-identical errors 23-93 points more often when the error is presented under an external chat role than inside their own <thought> block, so role labeling, not content, gates explicit self-correction.
False success occurs in 3-76% of LLM agent failures; LLM judges reach at most 0.65 AUROC while TF-IDF detectors reach 0.83-0.95 and recover 4-8x more cases at equal flag rate.
BARRED uses dimension decomposition and asymmetric multi-agent debate to generate high-fidelity synthetic data that lets small fine-tuned models outperform proprietary LLMs and existing guardrail models on custom policies.
A context-aware Sentinel-Strategist system for RAG selectively applies defenses to block membership inference and data poisoning while recovering most retrieval utility compared to always-on defense stacks.
Token-level contrastive attribution yields informative signals for some LLM benchmark failures but is not universally applicable across datasets and models.
A factorized study finds raw hidden states and attention features hard to beat in-domain for LLM uncertainty probes, but structured compressed features are more robust under distribution shift, with pretrained probes transferring to open-ended generation.
EPGS detects high-confidence factual errors in LLMs by using embedding perturbations to measure gradient sensitivity as a proxy for sharp versus flat minima.
SIVR detects LLM hallucinations by learning from token-wise and layer-wise variance patterns in internal hidden states, outperforming baselines with better generalization and less training data.
ELERAG integrates Wikidata entity linking with hybrid RRF re-ranking into RAG and outperforms baselines on a custom Italian academic dataset while cross-encoder methods win on the general SQuAD-it dataset.
citing papers explorer
-
An Agentic Evaluation Architecture for Historical Bias Detection in Educational Textbooks
An agentic architecture with multimodal screening, a five-agent jury, meta-synthesis, and source attribution protocol detects biases in Romanian history textbooks more accurately than zero-shot baselines, achieving 83.3% acceptable excerpts and human preference in 64.8% of blind comparisons.
-
ToxiREX: A Dataset on Toxic REasoning in ConteXt
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.
-
The Self-Correction Illusion: Role Relabeling Gates Explicit Error Flagging in Large Language Models
LLMs flag byte-identical errors 23-93 points more often when the error is presented under an external chat role than inside their own <thought> block, so role labeling, not content, gates explicit self-correction.
-
From Confident Closing to Silent Failure: Characterizing False Success in LLM Agents
False success occurs in 3-76% of LLM agent failures; LLM judges reach at most 0.65 AUROC while TF-IDF detectors reach 0.83-0.95 and recover 4-8x more cases at equal flag rate.
-
BARRED: Synthetic Training of Custom Policy Guardrails via Asymmetric Debate
BARRED uses dimension decomposition and asymmetric multi-agent debate to generate high-fidelity synthetic data that lets small fine-tuned models outperform proprietary LLMs and existing guardrail models on custom policies.
-
Adaptive Defense Orchestration for RAG: A Sentinel-Strategist Architecture against Multi-Vector Attacks
A context-aware Sentinel-Strategist system for RAG selectively applies defenses to block membership inference and data poisoning while recovering most retrieval utility compared to always-on defense stacks.
-
Contrastive Attribution in the Wild: An Interpretability Analysis of LLM Failures on Realistic Benchmarks
Token-level contrastive attribution yields informative signals for some LLM benchmark failures but is not universally applicable across datasets and models.
-
From Signals to Transfer: A Factorised Study of Probe-Based Uncertainty Estimation in Large Language Models
A factorized study finds raw hidden states and attention features hard to beat in-domain for LLM uncertainty probes, but structured compressed features are more robust under distribution shift, with pretrained probes transferring to open-ended generation.
-
From Flat Facts to Sharp Hallucinations: Detecting Stubborn Errors via Gradient Sensitivity
EPGS detects high-confidence factual errors in LLMs by using embedding perturbations to measure gradient sensitivity as a proxy for sharp versus flat minima.
-
Learning Uncertainty from Sequential Internal Dispersion in Large Language Models
SIVR detects LLM hallucinations by learning from token-wise and layer-wise variance patterns in internal hidden states, outperforming baselines with better generalization and less training data.
-
Enhancing Retrieval-Augmented Generation with Entity Linking for Educational Platforms
ELERAG integrates Wikidata entity linking with hybrid RRF re-ranking into RAG and outperforms baselines on a custom Italian academic dataset while cross-encoder methods win on the general SQuAD-it dataset.