https://github.com/Microsoft/CNTK
Revision 839efa23d5021fbe51cb5ec34b5c3faafb2e4e96 authored by KeDengMS on 31 March 2017, 18:07:11 UTC, committed by Mark Hillebrand on 01 April 2017, 06:20:59 UTC
When minibatch size is big (like 10000), the diagonal matrix in times node for gradient could be big. This change implements a ColumnwiseScaleAndWeightedAdd to reduce the cost in that case.
1 parent aca0b13
Raw File
Tip revision: 839efa23d5021fbe51cb5ec34b5c3faafb2e4e96 authored by KeDengMS on 31 March 2017, 18:07:11 UTC
Make sparse label CE backprop faster
Tip revision: 839efa2
CONTRIBUTING.md
You want to contribute to CNTK? We're really excited to work together!

Please, follow the steps from the Wiki Article at

https://github.com/Microsoft/CNTK/wiki/Contributing-to-CNTK

Your CNTK team.
back to top