Software Platform Lab @ SNU

Uncategorized

Cruise’s accepted at ICDCS 2019.

Cruise’s accepted at ICDCS 2019! Cruise is a system that automates system configuration of distributed machine learning.

Eunji’s giving talks at Stanford, UC Berkeley, Google, Microsoft, and NVIDIA.

Our JANUS work has been getting much attention. Eunji’s giving talks at Stanford, UC Berkeley, Google, Microsoft, and NVIDIA next week and the week after next week. Very exciting!

The JANUS team’s starting a collaboration with the Google Brain TensorFlow team.

The JANUS team’s starting a collaboration with the Google Brain TensorFlow team. We plan to integrate JANUS’s speculative execution and recursive function support into TensorFlow. This is very exciting.

Parallax’s accepted at EuroSys 2019!

“Parallax: Sparsity-aware Data Parallel Training of Deep Neural Networks” was accepted at EuroSys 2019! Congrats to the Parallax team!

JANUS’s accpeted at NSDI 2019!

Our paper entitled “JANUS: Fast and Flexible Deep Learning via Symbolic Graph Execution of Imperative Programs” got accepted at NSDI 2019! Congrats to Eunji, Sungwoo, Gyeong-In, Joo Seong, Dongjin, and Byung-Gon!

The Nemo team will give a tutorial about Nemo at KOSSCON 2018.

Wonook and Jangho will give a tutorial about Nemo at KOSSCON 2018.
https://www.kosscon.kr/program/tutorial#T5_2

Our paper’s accepted to Systems for ML Workshop at NIPS 2018!

Our paper, Making Classical Machine Learning Pipelines Differentiable: A Neural Translation Approach, was accepted to Systems for ML Workshop at NIPS 2018! This is joint work with Microsoft. Congrats, Gyeong-In!

Parallax’s open sourced!

We’re excited to announce that we made Parallax publicly available at https://github.com/snuspl/parallax. Parallax is a tool for automatically parallelizing deep learning training.

Our PRETZEL paper’s accepted to OSDI 2018!

Our PRETZEL paper’s accepted to OSDI 2018! This is joint work with Politecnico di Milano and Microsoft. Congrats to Yunseong and Byung-Gon!

Our on-flash stateful stream processing paper’s accepted to APSys 2018!

Our paper, High-performance Stateful Stream Processing on Solid-State Drives, was accepted to APSys 2018! Congrats to Gyewon, Jeongyoon, Jangho, and Taegeon!