@mcbeukman
Pinned post
I鈥檝e had so much fun making bricky, and playing with it: a compiler for LEGO庐 models, built for AI agents.
The agent writes a Python build script. bricky places every part, reports each mistake, and renders the model.
Thanks to @omar_aflak for the idea of a LEGO庐 compiler.View quoted post on X
Finally, here are the participants of our AutoRL debate!
We鈥檙e switching things up from a traditional panel: three controversial positions, with participants randomly assigned to argue for or against each one.
Expect opposing views, lively discussions, and surprising insights!View quoted post on X
1/ As compute continues to grow and simulators continue to improve, it is becoming feasible to train RL agents for billions or trillions of timesteps. However, this is only useful if agents can continue learning over such long training horizons, which is far from given 馃憞View quoted post on X
