Skip to content
MIT HAN Lab
  • Home
  • Publications
  • Awards
  • Media
  • Blog
  • Projects
    • NAS & AutoML
    • Model Compression
    • Efficient Training
    • Tiny NLP
    • Tiny Point Cloud
    • Tiny GAN
    • Tiny Video
    • Efficient Software Libraries
    • Efficient Hardware​
    • AI-Designed Hardware
    • Quantum System
  • Team
Search
MIT HAN Lab
Close menu
  • Home
  • Publications
  • Awards
  • Media
  • Blog
  • Projects
    • NAS & AutoML
    • Model Compression
    • Efficient Training
    • Tiny NLP
    • Tiny Point Cloud
    • Tiny GAN
    • Tiny Video
    • Efficient Software Libraries
    • Efficient Hardware​
    • AI-Designed Hardware
    • Quantum System
  • Team
MIT HAN Lab
Search Toggle menu

Author: Ligeng Zhu

November 28, 2022November 28, 2022Efficient Applications on Edge, Training

On-Device Training Under 256KB Memory

Historically, DNNs training happens on the cloud due to the huge memory cost. Edge platforms used to only perform inference, but it […]

Blogs

  • On-Device Training Under 256KB Memory
  • Reducing the carbon footprint of AI using the Once-for-All network
  • Auto Hardware-Aware Neural Network Specialization on ImageNet in Minutes
  • Efficiently Understanding Videos, Point Cloud and Natural Language on NVIDIA Jetson Xavier NX
  • ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware

Recent Comments

    Archives

    • November 2022
    • July 2020
    • May 2020
    • March 2019

    Categories

    • Efficient Applications on Edge
    • Neural Architecture Search
    • Training

    Useful Links

    Youtube

    GitHub

    Homepage of Prof.Song Han

    © 2025 MIT HAN Lab. Proudly powered by Sydney