TY  - RPRT
TI  - Densifying Assumed-sparse Tensors: Improving Memory Efficiency and MPI Collective Performance during Tensor Accumulation for Parallelized Training of Neural Machine Translation Models
AU  - Derya Cavdar
AU  - Valeriu Codreanu
AU  - Can Karakus
AU  - John A. Lockman III
AU  - Damian Podareanu
AU  - Vikram Saletore
AU  - Alexander Sergeev
AU  - Don D. Smith II
AU  - Victor Suthichai
AU  - Quy Ta
AU  - Srinivas Varadharajan
AU  - Lucas A. Wilson
AU  - Rengan Xu
AU  - Pei Yang
PY  - 2019
UR  - https://arxiv.org/abs/1905.04035
ID  - 1905.04035
ER  -