Documentation and Examples

[joaogui1]

So, are you guys planning on releasing the documentation and adding more examples? It's kind of hard to understand what the library has to offer and how to use it right now

[joaogui1]

Pinging @chris-chris @tomhennigan

[mtthss]

Hi, thanks for the comment.
We indeed plan to release more examples and add more extensive documentation.

[chris-chris]

@joaogui1 @mtthss
I'm implementing more examples using rlax and haiku step by step.
currently working on A3C :)

[aslanides]

We currently have two example agents using Haiku & rlax in bsuite baselines.

[joaogui1]

Maybe you could copy them here or reference them in the README?

[aslanides]

See https://github.com/deepmind/bsuite/tree/master/bsuite/baselines

[joaogui1]

I had already seen it, my suggestion is more about helping other people interested in the library who won't check all repos that depend on it

[joaogui1]

I wrote this example here: DQN
It uses the double q learning function, shows how to use vmap with your loss and implements a experience buffer.
What do you guys be interested in adding it to examples?

[mtthss]

Hello @joaogui1 ,

• I added to the README a link to the bsuite examples, thanks for the suggestion!
• Will take a look at your proposed DQN next and get back to you tomorrow.

[mtthss]

Hello @joaogui1 ,
The DQN example would be a nice contribution!
If you put together a PR, we can discuss the details there?

[mtthss]

Thanks @joaogui1 for the contribution!

We now have fully worked online Qlearning and DQN agents, plus references in the README to several BSuite agents that use rlax.