Dopamine is a framework for research and development which can be used for implementing reinforcement learning algorithms. It targets to fulfill the requirement for a tiny, only understandable codebase in which developer can experiment in a free behavior with new ideas (speculative research).
It is a framework which can be used to implement reinforcement-learning dedicatedly while other framework or platform does not provide this feature.
It can be used to implement Reinforcement learning with easy experimentation, having flexible development, compact and reliable, and also provide reproducibility in results.