New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Code review #1
Comments
These look so much better than the RL code I've seen in Python-based frameworks. One thing I noticed, though, is that functions like |
Fixed it. Can we add those two functions (or just |
How about turning this into a Julia package? Makes the dependencies easier. It would be nice to start training a standard model in a couple of lines, and even better if I can quickly demo a trained model. Would be great to make sure they are all GPU compatible as well. |
Cool, I'll start working on it. |
What's going on with lines like https://github.com/tejank10/Flux-baselines/blob/master/dqn/duel-dqn.jl#L46? Is it a matter of Flux not being able to deal with the complicated broadcast expression if you write it out with dots? (If so we should make sure that's fixed in Flux-on-v0.7). |
Works with dots, fixed it |
I am trying to adapt this code to a problem of mine but I get an error (on julia 1.0). Just wondering if anyone could give a hint how to deal with this. The code line is
and π comes directly from the Flux model (I use the AC example)
Not sure why the mean function doesn't recognize the ::TrackedArray or how I could adjust it to work. thanks for any hints |
If you're on Julia 1.0, |
major thanks. I also noted I need to change |
Hey @MikeInnes, if you are back could you please review the code? New models which I have added are Dueling DQN, Advantage Actor-Critic, and DDPG. Also, all the previous work done on DQN is added to dqn directory.
The text was updated successfully, but these errors were encountered: