Picture for Noah Golmant

Noah Golmant

On the Computational Inefficiency of Large Batch Sizes for Stochastic Gradient Descent

Add code
Nov 30, 2018
Figure 1 for On the Computational Inefficiency of Large Batch Sizes for Stochastic Gradient Descent
Figure 2 for On the Computational Inefficiency of Large Batch Sizes for Stochastic Gradient Descent
Figure 3 for On the Computational Inefficiency of Large Batch Sizes for Stochastic Gradient Descent
Figure 4 for On the Computational Inefficiency of Large Batch Sizes for Stochastic Gradient Descent
Viaarxiv icon

Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions

Add code
Dec 03, 2017
Figure 1 for Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions
Figure 2 for Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions
Figure 3 for Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions
Figure 4 for Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions
Viaarxiv icon