Yanghua Peng
6 papers · 2022–2026 · 4 conferences · across top CS/AI conferences
Achievements
Jump to papers ↓+3 more ↓ Show less ↑
π Cross-Pollinator (15) π§ Keyword Pioneer π Conference Polyglot (3) π Renaissance Researcher (6) π Interdisciplinary Bridge
πΊοΈ
Taxonomy Completionist
(13)
π₯
Mega-Team
(32)
π₯
Unstoppable
(5)
Conferences
NSDI (3)
AAAI (1)
CVPR (1)
NIPS (1)
Top co-authors
Keywords
distributed training
(4)
large language model
(2)
image generation
(1)
video generation
(1)
text-to-image generation
(1)
flow matching
(1)
model parallelism
(1)
fault tolerance
(1)
weight prediction
(1)
mixture of expert
(1)
feature retrieval
(1)
model scaling
(1)
flow-based model
(1)
text-to-video generation
(1)
data parallelism
(1)
rectified flow
(1)
subgraph sampling
(1)
gpu training
(1)
visual generation
(1)
graph neural network
(1)
Papers
OmniScale: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo
AAAI 2026
Goku: Flow Based Video Generative Foundation Models
CVPR 2025
ByteCheckpoint: A Unified Checkpointing System for Large Foundation Model Development
NSDI 2025
MegaScale: Scaling Large Language Model Training to More Than 10,000 GPUs
NSDI 2024
BGL: GPU-Efficient GNN Training by Optimizing Graph Data I/O and Preprocessing
NSDI 2023
SAPipe: Staleness-Aware Pipeline for Data Parallel DNN Training
NIPS 2022