Travis DeWolf comments

Results 54 comments of


                                            Travis DeWolf

Allow learning_rate to be a function of time

Hmm, I don't think I understand why learning rate as a node would be more biologically plausible.

Allow learning_rate to be a function of time

Hmm, I'm preferential towards only have the one place where you're controlling the learning rate, rather than having an error node and incorporating the learning rate there and then having...

Allow the user to set the initial state of a population

Just throwing in a +1 for this being desirable, not critical, but would be super nice to have.

Allow the user to set the initial state of a population

I think we've discussed this before, I thought we had it online but it might have been just in the office. If I remember correctly the problem here is setting...

Support for changing a learning rate

still an issue

Example using Python bindings to visualize Mujoco scene

Ah, okeydoke. Is there a time line on that? My goal was to remove dependencies! :D Looking through the Colab and then the `dm_control` documentation it seems like the rendering...

Example using Python bindings to visualize Mujoco scene

Oh, that would be excellent, thanks!

Example using Python bindings to visualize Mujoco scene

I took another look at the `simulate.cc` and decided to just work to try rendering from Python with that code as a reference (instead of the render_test.py) and was successful....

Example using Python bindings to visualize Mujoco scene

Ah, great, thanks! This is much better than building out my own while waiting for the updates!

Example using Python bindings to visualize Mujoco scene

@yuvaltassa any updates?