2 ms·Capturing token IDs during agentic interaction for better reinforcement learning3 points by visha1v 3mo agoaozelai 3mo ago[flagged]