@elliotarledge The network encoder is the bottleneck. This project was done with PufferDrive 2.0. It runs at ~300-400k SPS on a 4090 or 5090. IIRC, using one of the standard PufferLib MLPs gets you 1-1.5 M SPS, but the entity encoder is more performant. Starting from PufferLib 5.0 + spending a little time optimizing the regularizer implementation should get you to ~800k or higher SPS, but the network is still a bottleneck (overview below).

Sweden, here we come! 🇸🇪 I’m heading to ECCV, where we’ll host our workshop this Tuesday: emerging-ad.github.io/ We have a fantastic lineup of speakers; see you there!

Refactoring efforts pay off 🙂 (same functionality; less and more readable code)

Great video for an introduction to simulating fluids. youtube.com/watch?v=rSKMYc1C…

Looking forward to presenting our work on spiced self-play online on the 31st of July! Sign-up link below

for those interested, full interview youtube.com/watch?v=n-hWHV2Z…

We provide a couple of comparisons on the webpage! spiced-self-play.com/ e.g., see yellow car here

Project page: spiced-self-play.com/ Arxiv: arxiv.org/abs/2606.19370

New Paper: Human-like Autonomy Emerges from Self-Play and a Pinch of Human Data. We trained self-play RL on 60 years of simulation on 1 GPU in ~15 hours. Regularizing with 30 minutes of demonstration data produces much more human-like driving policies!