DeLask dynamically skips hallucination-prone decoder layers in LLMs by measuring gradient driftance via cosine similarity and partially aggregating states instead of full skipping.
In-context sharpness as alerts: An inner representation perspective for hallucination mitigation
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
verdicts
UNVERDICTED 4representative citing papers
DCO is an inference-time intervention that decomposes attention head outputs orthogonally to a dynamic context anchor and suppresses outlier components via Z-score to improve contextual faithfulness in Llama models.
APCD adaptively branches LLM decoding paths based on token entropy and contrasts divergent paths to improve factual accuracy while preserving efficiency.
A framework detects LLM anomalies including hallucinations, jailbreaks, and backdoors by forensic inspection of layer-wise hidden state patterns, reporting over 95% accuracy with minimal computational overhead.
citing papers explorer
-
Mitigating Hallucinations in Large Language Models Via Decoder Layer Skipping
DeLask dynamically skips hallucination-prone decoder layers in LLMs by measuring gradient driftance via cosine similarity and partially aggregating states instead of full skipping.
-
Hallucinations as Orthogonal Noise: Inference-Time Manifold Alignment via Dynamic Contextual Orthogonalization
DCO is an inference-time intervention that decomposes attention head outputs orthogonally to a dynamic context anchor and suppresses outlier components via Z-score to improve contextual faithfulness in Llama models.
-
APCD: Adaptive Path-Contrastive Decoding for Reliable Large Language Model Generation
APCD adaptively branches LLM decoding paths based on token entropy and contrasts divergent paths to improve factual accuracy while preserving efficiency.
-
Exposing the Ghost in the Transformer: Abnormal Detection for Large Language Models via Hidden State Forensics
A framework detects LLM anomalies including hallucinations, jailbreaks, and backdoors by forensic inspection of layer-wise hidden state patterns, reporting over 95% accuracy with minimal computational overhead.