An 8B autoregressive LM implements a language-switching backdoor via a three-phase circuit with early trigger composition, orthogonal mid-layer propagation, and final-layer MLP conversion, routed through a single-position serial bottleneck.
Title resolution pending
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
years
2026 5representative citing papers
QAOD projects away question-aligned directions from answer representations to isolate domain-agnostic factuality signals, enabling efficient hallucination detection with top in-domain AUROC and up to 21% better OOD transfer.
Linear probes on frozen LLM hidden states recover an approximate remaining-output-length signal that is decodable at prompt-end, transfers across datasets, and shifts upward at retraction tokens.
Stable personality vectors in LLMs function as intrinsic guardrails, with ablation increasing emergent misalignment above 40% and amplification reducing it below 3%, enabling zero-shot transfer from aligned to corrupted models.
LLMs compress concreteness into a consistent 1D direction in mid-to-late layers that separates literal from figurative noun uses and supports efficient classification plus steering.
citing papers explorer
-
Language-Switching Triggers Take a Latent Detour Through Language Models
An 8B autoregressive LM implements a language-switching backdoor via a three-phase circuit with early trigger composition, orthogonal mid-layer propagation, and final-layer MLP conversion, routed through a single-position serial bottleneck.
-
When Answers Stray from Questions: Hallucination Detection via Question-Answer Orthogonal Decomposition
QAOD projects away question-aligned directions from answer representations to isolate domain-agnostic factuality signals, enabling efficient hallucination detection with top in-domain AUROC and up to 21% better OOD transfer.
-
How Much is Left? LLMs Linearly Encode Their Remaining Output Length
Linear probes on frozen LLM hidden states recover an approximate remaining-output-length signal that is decodable at prompt-end, transfers across datasets, and shifts upward at retraction tokens.
-
Intrinsic Guardrails: How Semantic Geometry of Personality Interacts with Emergent Misalignment in LLMs
Stable personality vectors in LLMs function as intrinsic guardrails, with ablation increasing emergent misalignment above 40% and amplification reducing it below 3%, enabling zero-shot transfer from aligned to corrupted models.
-
Exploring Concreteness Through a Figurative Lens
LLMs compress concreteness into a consistent 1D direction in mid-to-late layers that separates literal from figurative noun uses and supports efficient classification plus steering.