I will going to build the algorithm known as DQN. I decide to use openai gym to reproduce the result described in Human-level control through deep reinforcement learning. The article can be download from here for free.
I notice that there are so many repositories about DQN, but a lot of them don’t generate any benchmark for algorithms. So I think I need a benchmark and algorithm comparision, and I decide to write my own repository.
I think writing code is going to be painful, so I need to write down the pains I have. By the way, I’m using tensorflow and gym together.
When I decide to write code about reinforcement learning, I notice there is just few website about reinforcement learning. And I want to share my experiences in RL. So when I look back, I would consider this as some kind of legacies. I would be happy if some one really get helps from this site.