Zero-shot cross-modal transfer of Reinforcement Learning policies through a Global Workspace

Open in new window