EfficientZero Remastered
RL Research Reproduction and Release
EfficientZero Remastered is an open-source effort to reproduce the state-of-the-art paper EfficientZero.
Our team has updated the code to improve its performance in both cloud and local environments, and has added fixes to stabilize the learning process. The trained models used for testing are publicly available on Hugging Face 🤗.
Ultimately, we hope this project helps make reinforcement learning more accessible to everyone.
This project was made possible by a research grant from Stability AI.
June 2023 update: We've completed our productionization, and the code is now available on GitHub! Read more about our findings in our blog post.
