Author name: Xiaodan Du, Ruize Xu, David Yunis, Yael Vinker, Greg Shakhnarovich

arXiv:2603.19500v2 Announce Type: replace
Abstract: We develop a method for producing vector sketches one part at a time. To do this, we train a multi-modal language model-based agent using a novel multi-turn process-reward reinforcement learning foll…