Improving LLMs with Ask-when-Needed for Clearer Instructions
Discover how the Ask-when-Needed framework helps LLM agents clarify unclear instructions, boosting accuracy and reducing errors in AI tool use.
ComboStoc: Boosting Diffusion Models with Combinatorial Stochasticity
Discover how ComboStoc enhances diffusion generative models by tackling combinatorial complexity for faster training and better data generation.
OpenAI Limits Access to GPT-5.5 Cyber Amid Safety Concerns
OpenAI restricts GPT-5.5 Cyber access to critical defenders, balancing innovation with cybersecurity risks and ethical deployment.
Data-Centric Foundation Models in Healthcare AI: Survey
Explore data-centric foundation models transforming healthcare AI, addressing data quality, privacy, and ethics for better clinical outcomes.
HalluHunter: Automated Detection of Factual Errors in LLMs
Discover HalluHunter, an iterative method that uncovers factual errors in large language models to improve accuracy and reliability.