Layer-wise Vulnerabilities in LLMs Exposed by Mechanistic Steering
Discover how mechanistic steering reveals layer-specific vulnerabilities in LLMs, enhancing defenses against adversarial jailbreak attacks.
MindTrellis: AI-Powered Interactive Knowledge Graph Tool
Discover MindTrellis, an AI-driven platform for collaborative knowledge structuring through interactive visual exploration and dynamic graph building.
ArgRE: Formal Conflict Resolution in Multi-Agent Negotiation
Discover ArgRE, a formal argumentation system enhancing conflict resolution and auditability in multi-agent requirements negotiation for complex software.
Efficient Language Modeling with Heterogeneous Expert Mixtures
Discover how Mixture of Heterogeneous Grouped Experts (MoHGE) boosts language model efficiency and balances GPU use for better AI performance.
MOSAIC: AI Code Generation Without Test Cases for Science
Discover MOSAIC, a novel AI framework enabling accurate scientific code generation without test cases using knowledge distillation and structured reasoning...