Author
tarif mouad
Recent research
- AI & ComputingOpen access
O-MoE: Optimized Mixture of Experts for Defensive and Specialized AI Systems
O-MoE is a theoretical framework designed to enhance Mixture of Experts (MoE) architectures by integrating: - **Sequential defensive layers** to counter jailbreak and bias. - **Explicit domain-expert mapping** for specialized tasks. - **Hierarchical Top-K allocation** for adaptiv...
- AI & ComputingOpen access
O-MoE: Optimized Mixture of Experts for Defensive and Specialized AI Systems
O-MoE is a theoretical framework designed to enhance Mixture of Experts (MoE) architectures by integrating: - **Sequential defensive layers** to counter jailbreak and bias. - **Explicit domain-expert mapping** for specialized tasks. - **Hierarchical Top-K allocation** for adaptiv...
- AI & ComputingOpen access
O-MoE: Optimized Mixture of Experts for Defensive and Specialized AI Systems
O-MoE is a theoretical framework designed to enhance Mixture of Experts (MoE) architectures by integrating: - **Sequential defensive layers** to counter jailbreak and bias. - **Explicit domain-expert mapping** for specialized tasks. - **Hierarchical Top-K allocation** for adaptiv...