"Simple random search provides a competitive approach to reinforcement learning", by Mania, Guy and @beenwrekt
Paper: https://arxiv.org/abs/1803.07055
Code: https://github.com/modestyachts/ARS …
Blog:http://www.argmin.net/2018/03/20/mujocoloco/ …
-
-
Humans have a lot of 'np.random' in their brains as well.
-
As stated, this looks like it's not even wrong. Since you likely had something deeper in mind, what do you mean by this?
- Show replies
New conversation -
-
-
Neural nets are stuck by symmetry around the saddle point at zero without random init :)
-
In a world without RNG, Glorot initialization is copy-pasting the weights he selected by hand, one at a time.
- Show replies
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Couldn't resist but to relate dot com bubble in 2000 and AI madness now. Same place, same "investors", same rules ... np.random is better than set of static HTML pages and guy that knows which "button" to press to make illusion of working prototype!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.