SSDTrain: An Activation Offloading Framework to SSDs for Faster Large Language Model Training
Paper
•
2408.10013
•
Published
Kun Wu
K-Wu
·
AI & ML interests
GPU compilers and libraries
Recent Activity
authored
a paper
3 days ago
SSDTrain: An Activation Offloading Framework to SSDs for Faster Large
Language Model Training
authored
a paper
3 days ago
Code generation and runtime techniques for enabling data-efficient deep
learning training on GPUs
authored
a paper
3 days ago
LSM-GNN: Large-scale Storage-based Multi-GPU GNN Training by Optimizing
Data Transfer Scheme
Organizations
None yet
authored
3
papers
3 days ago
authored
5
papers
over 1 year ago
PyTorch-Direct: Enabling GPU Centric Data Access for Very Large Graph Neural Network Training with Irregular Accesses
Paper
•
2101.07956
•
Published
PIGEON: Optimizing CUDA Code Generator for End-to-End Training and Inference of Relational Graph Neural Networks
Paper
•
2301.06284
•
Published
•
1
Graph Neural Network Training with Data Tiering
Paper
•
2111.05894
•
Published
TEMPI: An Interposed MPI Library with a Canonical Representation of CUDA-aware Datatypes
Paper
•
2012.14363
•
Published
Large Graph Convolutional Network Training with GPU-Oriented Data Communication Architecture
Paper
•
2103.03330
•
Published