MAC-Attention: Fast, Accurate Attention for Long-Context LLMs
Discover MAC-Attention, a novel scheme boosting large language models with fast, accurate long-context attention and up to 14x speed improvements.
Diversity-Aware Reverse KL Divergence for LLM Distillation
Explore Diversity-aware Reverse KL Divergence (DRKL) to improve large language model distillation with better performance and output diversity.
QUEST: Robust Query-Modulated Spherical Attention in Transformers
Discover QUEST, a stable and robust query-modulated spherical attention method enhancing Transformer training and performance across domains.
AI Agents: Adoption, Architectures & Insights from Experts
Explore AI agents' adoption, architectures, and expert insights to enhance your AI strategy with real-world practitioner takeaways.
Explainable AI for Blind Users: Trust & Accessibility
Explore how explainable AI can improve trust and accessibility for blind and low-vision users through multimodal, blame-aware designs.