This is an episode of the game bridge BoxWorld, played by a (simplicial) deep reinforcement learning agent, as described in the paper [ Ссылка ] by James Clift, Dmitry Doryn, Daniel Murfet and James Wallbridge.
What you can see in the video is the game board, together with various activations in the agent's network, specifically attention matrices from both the 1-simplicial (ordinary) and 2-simplicial attention. For detailed explanations see the paper.
