@piyusch@boknilev Is that good or bad? I mean technically the length of the paper shouldn't be too long that it spills into appendix. For that we have journals, right?
Have been working on this project built on top of MuJoCo-Warp, appended with a custom forked @Google Filament Render Engine which I have made to render thousands of environments in parallel on GPU natively.
Checkout the trailer video ๐๐
As it uses Google Filament render engine, you can now drop in any environments in GLB, FBX, etc. formats from your favourite asset stores like @Sketchfab, @polyhaven and add your robots into these ready to use photorealistic environments instantly.
Checkout the project at: https://t.co/dwNvOmeopz
@paraschopra This is very true and a valid problem. And not just that, reviewers at times cite these papers' results during rebuttal, and if you point out the concern that the results don't reproduce, the reviewer questions you for not including this finding in paper.
Our paper "Safe Flow Q-Learning" got accepted at @RL_Conference (RLC) 2026! ๐
RL is growing incredibly fast right now, and I'm just really happy to be contributing to the space along the way.
Looking forward to attending it in Montreal this August! ๐จ๐ฆ