[jakobnicolaus]

Hanabi is a multi-agent problem. Unfortunately gym doesn't natively support multi-agent action and state spaces.