The experiment will be broken into a series of phases in which a new agent (so to speak) is introduced to one contiguous narrative. Technically we will maintain a Minecraft server (version 1.12.2) and let these agents interact with the server and others as they learn according to the human examples we provide. There really isn't a goal here, we just want to see what the hell you do when you stress tess these agents. We'd be really happy to see the community contribute other agents to the narrative, but for now we're attempting to stress test the standard set of algorithms.
sdan/Herobraine
Artificially intelligent Herobrine. An experiment to create intelligence using deep reinforcement learning.
Python