projects / platform / upstream / armnn.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: db73c98) | patch
IVGCVSW-4643 Add Convolution2D tests with Bfloat16 input and Float32 output
authorNarumol Prangnawarat <narumol.prangnawarat@arm.com>
Wed, 1 Apr 2020 15:51:23 +0000 (16:51 +0100)
committerJim Flynn <jim.flynn@arm.com>
Wed, 1 Apr 2020 21:57:05 +0000 (21:57 +0000)
commite8cddebd26a1dc16b713e4aea1041fb9c8944ea7
treebf153ff5b8d733b10152c08e0c42c8c9ac8df867tree | snapshot
parentdb73c98f491d2fb17991acc13621c80cad628fb0commit | diff
IVGCVSW-4643 Add Convolution2D tests with Bfloat16 input and Float32 output

 * Modify SimpleConvolution2dNhwcTestImpl to allow different input and
out types
 * Add unit tests for Conv2D with Bfloat16 input and Float32 output to
compare the accuracy with Float32 for small value inputs and big value inputs

Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Ia59c92a22001b41f2681dafb951203a2223d3367
src/backends/backendsCommon/test/layerTests/Conv2dTestImpl.cpp diff | blob | history
src/backends/backendsCommon/test/layerTests/Conv2dTestImpl.hpp diff | blob | history
src/backends/reference/test/RefLayerTests.cpp diff | blob | history
Domain: Machine Learning / ML Framework;