search

Denys88 / rl_games

RL implementations
MIT License
930 stars 155 forks source link

Is there any description of the structure of this package? #255

Open xinyu-66 opened 1 year ago

xinyu-66 commented 1 year ago

Hi, I want to know if there is any documentations about the structure of this package? There are many inheritances among the files, I still do not know which file is used for updating observations, running episode. Is there any description documents to learn this package quickly?

Denys88 commented 1 year ago

Hi! Ill answer in a couple of days.

Denys88 commented 1 year ago

I am sorry for a huge delay. I still remember about you.

xinyu-66 commented 1 year ago

I am sorry for a huge delay. I still remember about you.

Hi, Denys, do not worry. I have already had some experience in this field, and now I am trying to add a new model for generating visual representations, which are used as input for the agent.

I have noticed that reinforcement learning is more complicated than deep learning, and it is very hard to package the pipeline completely and fully at the same time. In addition, I still want to suggest adding some explanations for the packages' structure or opening one tutorial with a very simple version without running mean std or so. In the beginning, I was used to calling neural network as model, so I took some time to understand the network_builder.py and model_builder.py.

Anyway, this package is great, and the community is still active, my wish is to give more descriptions and lower the barrier for many beginners.