Glance-or-Gaze: Adaptive Visual Search for LMMs
Discover how Glance-or-Gaze uses reinforcement learning to enhance Large Multimodal Models' focus and accuracy in complex visual search tasks.
AdaFRUGAL: Adaptive Memory-Efficient Training for LLMs
Discover AdaFRUGAL, an adaptive framework that optimizes LLM training with dynamic memory control, reducing GPU usage and boosting efficiency.
Addressing Demographic Bias in LLM Safety Alignment
Explore the Selective Safety Trap in LLMs and discover how MiJaBench audits demographic biases to improve AI safety for all groups.
Training-Free Adaptation of LLMs with Legacy Clinical Models
Discover CAPT, a training-free method to adapt new LLMs using legacy clinical models, boosting clinical AI performance without costly retraining.
q3-MuPa: Fast, Quiet Multi-Parametric MRI with Diffusion Models
Discover q3-MuPa, a rapid and silent multi-parametric MRI method using physics-informed diffusion models for accurate, noise-free quantitative imaging.