ScalingOpt is an optimization-centric community and knowledge hub for efficient AI. We focus on optimization methods, scalable algorithms, and resource-aware model design, with curated resources, tutorials, and research summaries.
Website: https://tianshijing.github.io/ScalingOpt/
If this repository is helpful, please consider giving it a star. Your support helps us reach more researchers and grow this resource.
- A comprehensive optimizer directory with detailed metadata
- Research blogs and tutorials for modern optimization
- Benchmark summaries and practical training guidance
- Community-driven curation and continuous updates
[2026.3] Mano v2 is now officially released! Developed by Yufei Gu and Juanxi Tian, stay tuned for updates.
[2026.2] ScalingOpt's latest optimizer, Mano, is now officially released! Authored by Yufei Gu.
[2026.1] We are always welcoming new members to join us. ScalingOpt has now supplemented the A-Summary-Sheet-of-Optimization-in-Deep-Learning (collecting over 100 types of optimizers), authored by Yifeng Liu.
[2025.12] ScalingOpt team has published its first blog post: Backbone-Optimizer Coupling Bias: The Hidden Co-Design Principle. Welcome to read it!
[2025.11] ScalingOpt has been officially released and is currently undergoing continuous improvement and expansion of its content. Please stay tuned!
Join our growing community of optimization researchers and practitioners.
- GitHub Discussions: Technical discussions and Q&A
- Research Collaboration: Connect with other researchers
- Blog Posts: Share your optimization insights
- Tutorial Contributions: Help others learn optimization
We thank the optimization research community for their groundbreaking work and contributions. Special thanks to:
- All researchers who developed the optimization algorithms featured in this platform