Credential Leakage Risks in LLM Agent Skills: Empirical Study
Discover critical insights into credential leakage in LLM agent skills from a large-scale study of 17,000+ skills, highlighting key vulnerabilities and fix...
Controlling LLM Sycophancy by Verbalizing Assumptions
Explore how verbalizing assumptions in LLMs helps explain and control sycophantic behavior, improving AI safety and response objectivity.
Natural Language Querying of Structured Data Using LLMs
Discover how language models enable natural language queries on structured data with high accuracy, even on resource-limited hardware.
MECO Dataset: Emotion & Cognitive Data for Older Adults
Explore MECO, a multimodal dataset offering 38 hours of emotion and cognitive data from older adults for advanced affective computing research.
JoyAI-LLM Flash: Efficient Mid-Scale LLM with Token Savings
Discover JoyAI-LLM Flash, a mid-scale language model optimizing token efficiency and performance with novel RL and sparsity techniques.