Welcome to MMRazor’s documentation!

Get Started

• Prerequisites
• Installation
  • Prepare environment
  • Install MMRazor
  • A from-scratch setup script
• Model Zoo
  • Baselines

Quick Run

• Train different type algorithms
  • NAS
  • Pruning
  • Distillation
• Train with different devices
  • Training with CPU
  • Train with single/multiple GPUs
  • Train with multiple machines
  • Launch multiple jobs on a single machine
• Test a model
  • NAS
  • Pruning
  • Distillation

Tutorials

• Tutorial 1: Overview
  • Major features:
  • Design and implement
  • Key Concepts
• Tutorial 2: Learn about Configs
  • Config Name Style
  • Config System
• Toturial 3: Customize Architectures
  • Develop searchable model components
  • Develop common model components
• Toturial 4: Customize NAS algorithms
• Tutorial 5: Customize Pruning algorithms
• Toturial 6: Customize KD algorithms
• Tutorial 7: Customize mixed algorithms with our algorithm components
• Tutorial 8: Apply existing algorithms to new tasks

Switch Language

• English
• 简体中文

API Reference

• mmrazor.apis
  • mmcls
  • mmdet
  • mmseg
• mmrazor.core
  • hooks
  • optimizer
  • runners
  • searcher
  • utils
• mmrazor.datasets
  • datasets
• mmrazor.models
  • algorithms
  • architectures
  • distillers
  • losses
  • mutables
  • mutators
  • ops
  • pruners
• mmrazor.utils

Indices and tables

• Index

• Search Page