diff options
| author | Narumol Prangnawarat <narumol.prangnawarat@arm.com> | 2023-05-22 10:57:47 +0100 |
|---|---|---|
| committer | TeresaARM <teresa.charlinreyes@arm.com> | 2023-05-23 21:07:47 +0000 |
| commit | 270641b6feb12cd68c812060b9828ac7d87d4826 (patch) | |
| tree | c04481ca07a0053a55e694c7a13aa7fea54dd158 /src/backends/neon/workloads/NeonUnidirectionalSequenceLstmFloatWorkload.cpp | |
| parent | d04873f16c261e166b0d90cecd77ae8e736e2e67 (diff) | |
| download | armnn-270641b6feb12cd68c812060b9828ac7d87d4826.tar.gz | |
MLCE-1022 Fix failure on UnidirectionalSequenceLstm Operator
* Fix failure to parse UnidirectionalSequenceLstm Operator on CpuAcc
* Fix failure to parse UnidirectionalSequenceLstm Operator on GpuAcc
* Fix IsLayerSupported tests when there are multiple otutputs
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Ia690f34d3c7fae87bd36c97056a3ff71baa865f6
Diffstat (limited to 'src/backends/neon/workloads/NeonUnidirectionalSequenceLstmFloatWorkload.cpp')
| -rw-r--r-- | src/backends/neon/workloads/NeonUnidirectionalSequenceLstmFloatWorkload.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/backends/neon/workloads/NeonUnidirectionalSequenceLstmFloatWorkload.cpp b/src/backends/neon/workloads/NeonUnidirectionalSequenceLstmFloatWorkload.cpp index 8dba719d91..7bdb2d5a5a 100644 --- a/src/backends/neon/workloads/NeonUnidirectionalSequenceLstmFloatWorkload.cpp +++ b/src/backends/neon/workloads/NeonUnidirectionalSequenceLstmFloatWorkload.cpp @@ -517,7 +517,7 @@ NeonUnidirectionalSequenceLstmFloatWorkloadValidate(const TensorInfo& input, const LstmInputParamsInfo& paramsInfo) { TensorShape inputLayerShape = input.GetShape(); - TensorShape outputLayerShape = outputStateIn.GetShape(); + TensorShape outputLayerShape = output.GetShape(); unsigned int maxTime = descriptor.m_TimeMajor ? inputLayerShape[0] : inputLayerShape[1]; unsigned int batchSize = descriptor.m_TimeMajor ? inputLayerShape[1] : inputLayerShape[0]; |