A research paper introduces Reframe, a post-training framework combining agentic reinforcement learning and multi-teacher on-policy distillation to help visual agents use tool-returned observations and ground actions better.
Read it at HF Papers →
← Front page