Ziming Liu
  • Bio
  • Papers
  • Experience
  • Projects
  • Recent & Upcoming Talks
    • Example Talk
  • Projects
    • RAS
    • OpenDiT
    • EnergonAI
  • Publications
    • Region-Adaptive Sampling for Diffusion Transformers
    • WeiPipe: Weight Pipeline Parallelism for Communication-Effective Long-Context Large Model Training
    • Concerto: Automatic Communication Optimization and Scheduling for Large-Scale Deep Learning
    • WallFacer: Harnessing Multi-dimensional Ring Parallelism for Efficient Long Sequence Model Training
    • HeteGen: Efficient Heterogeneous Parallel Inference for Large Language Models on Resource-Constrained Devices
    • AutoChunk: Automated Activation Chunk for Memory-Efficient Long Sequence Inference
    • DSP: Dynamic Sequence Parallelism for Multi-Dimensional Transformers
    • Hanayo: Harnessing Wave-like Pipeline Parallelism for Enhanced Large Model Training Efficiency
    • ATP: Adaptive Tensor Parallelism for Foundation Models
    • EnergonAI: An Inference System for 10-100 Billion Parameter Transformer Models
  • Projects
  • Blog
    • 🎉 Easily create your own simple yet highly customizable blog
    • 🧠 Sharpen your thinking with a second brain
    • 📈 Communicate your results effectively with the best data visualizations
    • 👩🏼‍🏫 Teach academic courses
    • ✅ Manage your projects
  • Experience
  • Teaching
    • Learn JavaScript
    • Learn Python

RAS

Feb 14, 2025 · 1 min read
Go to Project Site

An open-source implementation of Regional Adaptive Sampling (RAS), a novel diffusion model sampling strategy that introduces regional variability in sampling steps.

Last updated on Feb 14, 2025
AIGC Diffusion Transformer Multimodal Model
Ziming Liu
Authors
Ziming Liu
Ph.D. Candidate
OpenDiT Feb 1, 2024 →

2025 Ziming Liu, NUS HPC-AI lab. All rights reserved.

Published with Hugo Blox Builder — the free, open source website builder that empowers creators.