Text this: Exploration and value function factorisation in single and multi-agent reinforcement learning