projects
/
platform
/
upstream
/
armnn.git
/ commitdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
| commitdiff |
tree
raw
|
patch
| inline |
side by side
(parent:
5a05963
)
IVGCVSW-4390 Refactor QUANTIZE to make use of Decoder/Encoder types
author
Keith Davis
<keith.davis@arm.com>
Tue, 4 Feb 2020 09:31:47 +0000
(09:31 +0000)
committer
Keith Davis
<keith.davis@arm.com>
Tue, 4 Feb 2020 09:31:47 +0000
(09:31 +0000)
* Fix bug where QSymmS8 is duplicated and F16 is missing
Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: I312815868e74a9dff913bac052f1a86585644098
src/backends/reference/RefLayerSupport.cpp
patch
|
blob
|
history
diff --git
a/src/backends/reference/RefLayerSupport.cpp
b/src/backends/reference/RefLayerSupport.cpp
index c6a3af46bf7ebf4a58a09a571440ae3794193ded..0ca19bbb06dbd75f0cbae238007f6f2e1faaf65c 100644
(file)
--- a/
src/backends/reference/RefLayerSupport.cpp
+++ b/
src/backends/reference/RefLayerSupport.cpp
@@
-1440,8
+1440,8
@@
bool RefLayerSupport::IsQuantizeSupported(const TensorInfo& input,
// Define supported input types.
std::array<DataType,5> supportedInputTypes = {
- DataType::QSymmS8,
DataType::Float32,
+ DataType::Float16,
DataType::QAsymmU8,
DataType::QSymmS8,
DataType::QSymmS16