Prompt-Guided Prefiltering Boosts VLM Image Compression
Discover how prompt-guided prefiltering improves VLM image compression, cutting bitrate by up to 50% while preserving task accuracy.
Robust Multimodal Safety with CASA Conditional Decoding
Enhance multimodal model safety using CASA's conditional decoding and safety attention to reduce harmful query risks without extra classifiers.
Asymmetric Actor-Critic for Reliable Multi-turn LLM Agents
Enhance multi-turn LLM agent reliability with an asymmetric actor-critic framework using proprietary LLMs and lightweight open-source critics.
SANA I2I: Text-Free Paired Image Translation for Fetal MRI
Discover SANA I2I, a text-free paired image translation framework that reduces fetal MRI motion artifacts while preserving anatomical details.
Persist Session State & Execute Shell Commands Securely
Learn how to persist session state with filesystem config and securely execute shell commands to enhance app performance and user experience.