SciIntegrity-Bench shows seven LLMs exhibit a 34.2% integrity failure rate in dilemmatic scenarios, with all models fabricating synthetic data in missing-data cases and an intrinsic completion bias persisting after prompt changes.
hub
Detecting hallucinations in large language models using semantic entropy.Nature, 630(8017):625–630
14 Pith papers cite this work. Polarity classification is still indexing.
14
Pith papers citing it
hub tools
citation-role summary
background 1
baseline 1
citation-polarity summary
representative citing papers
SIOP enables turn-level credit assignment in LLM agents via semantic clustering of final answers as latent outcomes, improving performance on reasoning benchmarks without verifiers.
Presents a distributional model of linguistic confidence, Faithfulness Divergence metric, and RALC pipeline that boosts faithfulness and calibration on QA benchmarks across LLM families.
PUMA detects reasoning-level semantic redundancy to enable early exit in chains of thought, achieving 26.2% average token reduction across five LRMs and five benchmarks while preserving accuracy and CoT quality.
Temporal knowledge drift is encoded as a geometrically orthogonal direction in LLM residual streams, independent of correctness and uncertainty.
Uncertainty estimation for LLM hallucinations can be done effectively with partial generations or input-only predictors, reducing the need for full autoregressive sampling.
NeuroCogMap maps LLM internal representations into stable functional parcels tied to cognitive functions, failure modes, and human cortical activity during language tasks.
CoMet decomposes MLLM uncertainty into context-specific and multiplicity-specific terms estimated by a trained post-hoc module, improving performance on open-ended multimodal benchmarks and hallucination detection.
PASTA combines data augmentation and a self-learning DPO process to integrate new factual knowledge from news articles into LLMs, raising accuracy from 0.02 to 0.82 on post-cutoff questions while preserving general capabilities.
Identifies two gaps in entropy-based uncertainty for LLM post-training and proposes GCPO to align geometry-aware disagreement measures with reward-based calibration for better gradient regulation.
A lightweight max-pooling network with MLP detects LLM hallucinations competitively without semantic consistency computations by adaptively aggregating internal token features.
Researchers create a human-labeled dataset of obvious and elusive multimodal hallucinations and use learned activation-space probes to control their verifiability in MLLMs.
LLMs memorize citations hierarchically: titles and first authors are recalled at lower redundancy levels than venues or years, with accuracy scaling log-linearly and saturating near verbatim reproduction above roughly 1200 citations.
Semi-CoT selects low-entropy pseudo-CoT chains from unlabeled questions via answer-level semantic entropy and shows high pseudo-answer precision but only small or negative gains on math reasoning benchmarks.
citing papers explorer
-
SciIntegrity-Bench: A Benchmark for Evaluating Academic Integrity in AI Scientist Systems
SciIntegrity-Bench shows seven LLMs exhibit a 34.2% integrity failure rate in dilemmatic scenarios, with all models fabricating synthetic data in missing-data cases and an intrinsic completion bias persisting after prompt changes.
-
Self-Induced Outcome Potential: Turn-Level Credit Assignment for Agents without Verifiers
SIOP enables turn-level credit assignment in LLM agents via semantic clustering of final answers as latent outcomes, improving performance on reasoning benchmarks without verifiers.
-
Retrieval-Augmented Linguistic Calibration
Presents a distributional model of linguistic confidence, Faithfulness Divergence metric, and RALC pipeline that boosts faithfulness and calibration on QA benchmarks across LLM families.
-
Stop When Reasoning Converges: Semantic-Preserving Early Exit for Reasoning Models
PUMA detects reasoning-level semantic redundancy to enable early exit in chains of thought, achieving 26.2% average token reduction across five LRMs and five benchmarks while preserving accuracy and CoT quality.
-
The Geometry of Forgetting: Temporal Knowledge Drift as an Independent Axis in LLM Representations
Temporal knowledge drift is encoded as a geometrically orthogonal direction in LLM residual streams, independent of correctness and uncertainty.
-
Towards Generation-Efficient Uncertainty Estimation in Large Language Models
Uncertainty estimation for LLM hallucinations can be done effectively with partial generations or input-only predictors, reducing the need for full autoregressive sampling.
-
NeuroCogMap Reveals Cognitive Organization of Large Language Models
NeuroCogMap maps LLM internal representations into stable functional parcels tied to cognitive functions, failure modes, and human cortical activity during language tasks.
-
CoMet: Context and Multiplicity Decomposition for Multimodal Uncertainty Estimation
CoMet decomposes MLLM uncertainty into context-specific and multiplicity-specific terms estimated by a trained post-hoc module, improving performance on open-ended multimodal benchmarks and hallucination detection.
-
PASTA: A Paraphrasing And Self-Training Approach for Knowledge Updating in LLMs
PASTA combines data augmentation and a self-learning DPO process to integrate new factual knowledge from news articles into LLMs, raising accuracy from 0.02 to 0.82 on post-cutoff questions while preserving general capabilities.
-
Why Semantic Entropy Fails: Geometry-Aware and Calibrated Uncertainty for Policy Optimization
Identifies two gaps in entropy-based uncertainty for LLM post-training and proposes GCPO to align geometry-aware disagreement measures with reward-based calibration for better gradient regulation.
-
Max-pooling Network Revisited: Analyzing the Role of Semantic Probability in Multiple Instance Learning for Hallucination Detection
A lightweight max-pooling network with MLP detects LLM hallucinations competitively without semantic consistency computations by adaptively aggregating internal token features.
-
Steering the Verifiability of Multimodal AI Hallucinations
Researchers create a human-labeled dataset of obvious and elusive multimodal hallucinations and use learned activation-space probes to control their verifiability in MLLMs.
-
Hierarchical Memorization in Large Language Models: Evidence from Citation Generation
LLMs memorize citations hierarchically: titles and first authors are recalled at lower redundancy levels than venues or years, with accuracy scaling log-linearly and saturating near verbatim reproduction above roughly 1200 citations.
-
Revisiting Chain-of-Thought Reasoning under Limited Supervision: Semi-supervised Chain-of-Thought Learning
Semi-CoT selects low-entropy pseudo-CoT chains from unlabeled questions via answer-level semantic entropy and shows high pseudo-answer precision but only small or negative gains on math reasoning benchmarks.