Tag: Language Model Safety
Browse our exclusive articles!
ProjLens: Enhancing Safety in Multimodal Models
Discover how ProjLens improves multimodal model safety by exposing backdoor vulnerabilities and boosting robustness in large language models.
Spectral Dynamics of Hallucination in Whisper ASR Models
Explore how spectral dynamics affect hallucinations in Whisper ASR models, revealing safety risks and phase transitions across model scales.
Statute-Centric Legal QA: Structure-Aware Retrieval & Safety
Explore SearchFireSafety, a benchmark enhancing statute-centric legal QA with structure-aware retrieval and safety to reduce hallucination in legal AI mode...
MetaSAEs: Enhancing Atomic Sparse Autoencoder Latents
Discover how MetaSAEs use joint training with decomposability penalties to create more atomic, interpretable sparse autoencoder latents for safer AI models...
GUARD-SLM: Defense Against Jailbreaks in Small Language Models
Discover GUARD-SLM, a token activation-based method protecting small language models from jailbreak attacks while preserving legitimate inputs.