Danijar Hafner
Danijar Hafner
The reported scores are the average episode returns within the last 10k steps, that is, all episodes that finished between 390k and 400k environment frames. The scores after that are...
Continuing this conversation in https://github.com/danijar/dreamerv3/issues/175.
Good point. I'm not maintaining the pip package right now, but ultimately I think having separate `embodied` and agent packages makes sense.
Which version of ale-py are you using? See: https://github.com/Farama-Foundation/Arcade-Learning-Environment/issues/594
@chrisranderson This looks nice. Is there a way to write these summaries from within the graph, something like a `tf.summary.video(name, tensor_of_frames)`?
I also wrote a cleaned-up TF 2 version of Alex' GIF summary a few days ago. It works for both eager tensors, Numpy arrays, and with the note below also...
Any update on this?
Okay, thanks! Looking forward to this.
After 8 years of waiting, I've given up on TensorBoard and wrote my own with video support: [github.com/danijar/scope](https://github.com/danijar/scope)
Good idea! We currently don't do this and it still works fine but it's possibly that that would improve performance.