Mesa: A Memory-saving Training Framework for Transformers.

Zizheng Pan Peng Chen Haoyu He Jing Liu Jianfei Cai Bohan Zhuang

Published in: CoRR (2021)

Keyphrases

neural network
theoretical framework
real time
real world
probabilistic model
main contribution
bayesian framework
knowledge base
case study
support vector
search algorithm
active learning
memory requirements
limited memory
framework enables