Source-Sensitive Reasoning in Turkish: Humans vs LLMs
Explore how source trust impacts Turkish evidential reasoning in humans and LLMs, revealing gaps in AI's understanding of language trust cues.
DySIB: Learning Phase Space from High-Dim Experimental Data
Discover DySIB, a novel method to extract low-dimensional phase space from high-dimensional experimental data, enhancing dynamics analysis and interpretati...
AgentWard: Secure Lifecycle Architecture for AI Agents
Discover AgentWard, a lifecycle security architecture enhancing protection for autonomous AI agents through multi-stage defense strategies.
DepthKV: Layer-Wise KV Cache Pruning for Efficient LLMs
Optimize long-context LLM inference with DepthKV's layer-dependent KV cache pruning for improved memory efficiency and model performance.
K-MetBench: Benchmarking AI for Korean Meteorology
Discover K-MetBench, a multidimensional benchmark for evaluating AI expert reasoning, locality, and multimodality in Korean meteorology applications.