Latest research in Multi-agent communication is focussed on getting agents to use language in a real task, allowing them to form their own representations and learn from interactions. Inspired by Ludwig Wittgenstein's - "Meaning is use"
Implements a Tensorflow 2.0 version of the language-game in this paper - https://arxiv.org/abs/1612.07182
