Replaces heavy, unoptimized deep learning models with lightweight versions via pruning and quantization. Provides a unified benchmark for knowledge distillation and neural architecture search within the OpenMMLab ecosystem.
Tech Stack
PythonAI, ML & Data
Why mmrazor?
• Unified interface for multiple techniques
• Extensive benchmark results
• Modular design for custom algorithms
Limitations
• Dependency on OpenMMLab stack
• Complex configuration files
• Steep learning curve for beginners
3/3/2026
Last Update
241
Forks
167
Issues
Apache-2.0
License
Financial Leak Detected
Stop the "SaaS Tax"
Your team could be burning cash. Switching to mmrazor instantly boosts your runway.
