TY  - RPRT
TI  - Tartan: Accelerating Fully-Connected and Convolutional Layers in Deep Learning Networks by Exploiting Numerical Precision Variability
AU  - Alberto Delmas
AU  - Sayeh Sharify
AU  - Patrick Judd
AU  - Andreas Moshovos
PY  - 2017
UR  - https://arxiv.org/abs/1707.09068
ID  - 1707.09068
ER  -