Daniel Bashir

Awesome, on refactoring: let me throw out another PR or at least make an issue with ideas once I've played around more. Some thoughts off the top of my head...

I wouldn't call this anything definitive since I haven't done hyperparameter sweeps or anything, but using standard values and trying MLP, KAN, and the efficient version (with 16 seeds this...

Gotcha, are you using any bells and whistles or just standard reinforce? My second plot (above comment) was with rtg—I'm still not sure why the efficient_kan version didn't run for...

Yeah, I understood that you meant the y axis haha—thanks, it slipped by me that I wasn't using truncation! And gotcha, I'll play around and see if doing that gives...