29 #include "tests/datasets/ShapeDatasets.h" 34 #include "tests/validation/fixtures/ConvertFullyConnectedWeightsFixture.h" 58 validate(CLAccessor(_target), _reference);
half_float::half half
16-bit floating point type
1 channel, 1 F32 per channel
std::enable_if< is_container< T >::value, ContainerDataset< T > >::type make(std::string name, T &&values)
Helper function to create a ContainerDataset.
This file contains all available output stages for GEMMLowp on OpenCL.
1 channel, 1 F16 per channel
#define TEST_SUITE(SUITE_NAME)
Basic function to run CLConvertFullyConnectedWeightsKernel.
ConvertFullyConnectedWeightsValidationFixture< CLTensor, CLAccessor, CLConvertFullyConnectedWeights, T > CLConvertFullyConnectedWeightsFixture
FIXTURE_DATA_TEST_CASE(RunSmall, CLAbsoluteDifferenceFixture< uint8_t >, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), AbsoluteDifferenceU8Dataset))
validate(dst.info() ->valid_region(), dst_valid_region)
DatasetMode
Possible dataset modes.
quantized, asymmetric fixed-point 8-bit number
Accessor implementation for CLTensor objects.
Num samples, channels, height, width.
TEST_SUITE_END() DATA_TEST_CASE(Configuration
Num samples, height, width, channels.
combine(combine(combine(concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType",{DataType::U8, DataType::S16})), datasets::BorderModes()), framework::dataset::make("filter_size",{5}))
DataType
Available data types.
Basic implementation of the OpenCL tensor interface.