- 21 10月, 2018 6 次提交
-
-
由 Tomasz Patejko 提交于
MKLDNN conv + elementwise_add fusion: skip connection attribute renamed. Comments about patterns added. test=develop
-
由 Michal Gallus 提交于
MKLDNN conv + elementwise_add fusion: Fix output_data to point to the right tensor, also fix transpiler integration
-
由 Tomasz Patejko 提交于
-
由 Tomasz Patejko 提交于
-
由 Tomasz Patejko 提交于
MKLDNN conv + elementwise_add fusion: output and elemwise param share data in conv primitive. Output is properly allocated
-
由 Tomasz Patejko 提交于
-
- 29 9月, 2018 1 次提交
-
-
由 Yu Yang 提交于
-
- 14 9月, 2018 1 次提交
-
-
由 Michał Gallus 提交于
-
- 13 9月, 2018 1 次提交
-
-
由 chuanqiw 提交于
-
- 11 9月, 2018 3 次提交
-
-
由 Krzysztof Binias 提交于
-
由 Michal Gallus 提交于
-
由 Krzysztof Binias 提交于
-
- 10 9月, 2018 1 次提交
-
-
由 Krzysztof Binias 提交于
-
- 21 8月, 2018 1 次提交
-
-
由 Michał Gallus 提交于
* Fuse Convolution and Eltwise Add into Conv+Bias * Reduce bias branching at conv_mkldnn_op * Add MKLDNN build checks for Conv Bias * Conv-bias: check if bias input exist befor assignment * Conv-bias: Remove Bias dim check from infershape It was causing conv3d test to crash upon\ncalling HasInput(Bias)
-
- 17 8月, 2018 2 次提交
-
-
由 Michal Gallus 提交于
Also replace memory types from int64_t to size_t
-
由 Michal Gallus 提交于
-
- 09 8月, 2018 1 次提交
-
-
由 Sylwester Fraczek 提交于
-
- 31 7月, 2018 1 次提交
-
-
由 gongweibao 提交于
-
- 13 7月, 2018 2 次提交
-
-
由 Jacek Czaja 提交于
-
由 Jacek Czaja 提交于
-
- 11 7月, 2018 1 次提交
-
-
由 Jacek Czaja 提交于
* - Rebase of conv reuse - clag formatter fixes - Fix to conv reuse - Yet another fix - Fix - Fix - clagn format * - comment update
-
- 11 6月, 2018 1 次提交
-
-
由 mozga-intel 提交于
-
- 18 4月, 2018 2 次提交
-
-
由 Abhinav Arora 提交于
-
由 Abhinav Arora 提交于
* Fix conv_op.h * Fix conv_mkldnn_op * Fix cpplint issues in detection_map_op
-
- 10 3月, 2018 1 次提交
-
-
由 pzelazko-intel 提交于
* MKLDNN pool2d OP kernel added * conv2d and pool2d MKLDNN kernels renamed * MKLDNN conv2d kernel refactoring
-
- 07 3月, 2018 1 次提交
-
-
由 pzelazko-intel 提交于
* MKLDNN conv2 OP kernel added * TODOs added * mkldnn conv2d OP refactor * CanCUDNNBeUsed and CanMKLDNNBeUsed moved
-