virea. VRM-native motion model project for interactive embodied avatars, aiming to generate continuous skeletal motion from text and dialogue context. Includes motion data preparation, VRM-centered representation, training pipelines, preview/evaluation tools, and runtime playback.