Search for a command to run...
Towards Omnimodal Expressions and Reasoning in Referring Audio-Visual Segmentation