large language models (LLMs) (AI model): news timeline & CVEs

Search-Aware Reinforcement Learning for Multi-Component Query Understanding in Roblox Game Search

A search-aware reinforcement learning framework for query understanding in Roblox Game Search improves search quality by 8.9 points over SFT policy.

A search-aware reinforcement learning framework for multi-component query understanding in Roblox Game Search improves search quality by 8.9 points over supervised fine-tuning, demonstrating component-specific optimization for query understanding.

arXiv cs.AI / cs.LG / cs.CLSep 24, 17:26 UTC · 2d agoResearch

Your Model Is Leaking: Covert Information Transfer through LLM Residual Streams

A residual-stream covert-channel attack hides sensitive information in LLM intermediate activations, recoverable with a simple linear decoder without retraining or weight modification.

This research describes a residual-stream covert-channel attack that hides sensitive information in LLM intermediate activations, enabling recovery with a simple linear decoder without model retraining or weight modification.

arXiv cs.CRSep 23, 12:27 UTC · 3d agoAI safety & security in the wild

large language models (LLMs)

Timeline2 articles

Appears with

Search-Aware Reinforcement Learning for Multi-Component Query Understanding in Roblox Game Search

Your Model Is Leaking: Covert Information Transfer through LLM Residual Streams