Add bert-base-ner in MemLeak tests (#19817)
* Add `bert-base-ner` in MemLeak tests * Fix segfault caused by `fillTensorRandom()`
This commit is contained in:
Vitaliy Urusovskij
GitHub
parent
475ad32cc4
commit
6b5a22a656
@ -10,6 +10,13 @@
|
||||
#include <random>
|
||||
|
||||
|
||||
template <typename T>
|
||||
using uniformDistribution = typename std::conditional<
|
||||
std::is_floating_point<T>::value,
|
||||
std::uniform_real_distribution<T>,
|
||||
typename std::conditional<std::is_integral<T>::value, std::uniform_int_distribution<T>, void>::type>::type;
|
||||
|
||||
|
||||
/**
|
||||
* @brief Determine if InferenceEngine blob means image or not (OV API 1.0)
|
||||
*/
|
||||
@ -84,27 +91,20 @@ void fillBlobRandom(InferenceEngine::Blob::Ptr &inputBlob) {
|
||||
/**
|
||||
* @brief Fill InferenceEngine tensor with random values (OV API 2.0)
|
||||
*/
|
||||
template<typename T, typename U>
|
||||
ov::Tensor fillTensorRandom(T &input) {
|
||||
ov::Tensor tensor{input.get_element_type(), input.get_shape()};
|
||||
std::vector<U> values(ov::shape_size(input.get_shape()));
|
||||
|
||||
std::random_device rd;
|
||||
std::mt19937 gen(rd());
|
||||
|
||||
if (std::is_floating_point<U>::value == true) {
|
||||
std::uniform_real_distribution<> distrib_f = std::uniform_real_distribution<>(0, std::numeric_limits<U>::max());
|
||||
for (size_t i = 0; i < values.size(); ++i)
|
||||
values[i] = distrib_f(gen);
|
||||
} else {
|
||||
std::uniform_int_distribution<> distrib_i = std::uniform_int_distribution<>(0, std::numeric_limits<U>::max());
|
||||
for (size_t i = 0; i < values.size(); ++i)
|
||||
values[i] = distrib_i(gen);
|
||||
template<typename T, typename T2>
|
||||
void fillTensorRandom(ov::Tensor& tensor,
|
||||
T rand_min = std::numeric_limits<uint8_t>::min(),
|
||||
T rand_max = std::numeric_limits<uint8_t>::max()) {
|
||||
std::mt19937 gen(0);
|
||||
size_t tensor_size = tensor.get_size();
|
||||
if (0 == tensor_size) {
|
||||
throw std::runtime_error(
|
||||
"Models with dynamic shapes aren't supported. Input tensors must have specific shapes before inference");
|
||||
}
|
||||
|
||||
std::memcpy(tensor.data(), values.data(), sizeof(U) * values.size());
|
||||
|
||||
return tensor;
|
||||
T* data = tensor.data<T>();
|
||||
uniformDistribution<T2> distribution(rand_min, rand_max);
|
||||
for (size_t i = 0; i < tensor_size; i++)
|
||||
data[i] = static_cast<T>(distribution(gen));
|
||||
}
|
||||
|
||||
|
||||
@ -141,26 +141,34 @@ void fillBlobImInfo(InferenceEngine::Blob::Ptr &inputBlob,
|
||||
template<typename T>
|
||||
void fillTensors(ov::InferRequest &infer_request, std::vector<T> &inputs) {
|
||||
for (size_t i = 0; i < inputs.size(); ++i) {
|
||||
ov::Tensor input_tensor;
|
||||
|
||||
if (inputs[i].get_element_type() == ov::element::f32) {
|
||||
input_tensor = fillTensorRandom<T, float>(inputs[i]);
|
||||
} else if (inputs[i].get_element_type() == ov::element::f64) {
|
||||
input_tensor = fillTensorRandom<T, double>(inputs[i]);
|
||||
} else if (inputs[i].get_element_type() == ov::element::f16) {
|
||||
input_tensor = fillTensorRandom<T, short>(inputs[i]);
|
||||
} else if (inputs[i].get_element_type() == ov::element::i32) {
|
||||
input_tensor = fillTensorRandom<T, int32_t>(inputs[i]);
|
||||
} else if (inputs[i].get_element_type() == ov::element::i64) {
|
||||
input_tensor = fillTensorRandom<T, int64_t>(inputs[i]);
|
||||
} else if (inputs[i].get_element_type() == ov::element::u8) {
|
||||
input_tensor = fillTensorRandom<T, uint8_t>(inputs[i]);
|
||||
} else if (inputs[i].get_element_type() == ov::element::i8) {
|
||||
input_tensor = fillTensorRandom<T, int8_t>(inputs[i]);
|
||||
} else if (inputs[i].get_element_type() == ov::element::u16) {
|
||||
input_tensor = fillTensorRandom<T, uint16_t>(inputs[i]);
|
||||
} else if (inputs[i].get_element_type() == ov::element::i16) {
|
||||
input_tensor = fillTensorRandom<T, int16_t>(inputs[i]);
|
||||
auto input_tensor = infer_request.get_tensor(inputs[i]);
|
||||
auto type = inputs[i].get_element_type();
|
||||
if (type == ov::element::f32) {
|
||||
fillTensorRandom<float, float>(input_tensor);
|
||||
} else if (type == ov::element::f64) {
|
||||
fillTensorRandom<double, double>(input_tensor);
|
||||
} else if (type == ov::element::f16) {
|
||||
fillTensorRandom<ov::float16, float>(input_tensor);
|
||||
} else if (type == ov::element::i32) {
|
||||
fillTensorRandom<int32_t, int32_t>(input_tensor);
|
||||
} else if (type == ov::element::i64) {
|
||||
fillTensorRandom<int64_t, int64_t>(input_tensor);
|
||||
} else if ((type == ov::element::u8) || (type == ov::element::boolean)) {
|
||||
// uniform_int_distribution<uint8_t> is not allowed in the C++17
|
||||
// standard and vs2017/19
|
||||
fillTensorRandom<uint8_t, uint32_t>(input_tensor);
|
||||
} else if (type == ov::element::i8) {
|
||||
// uniform_int_distribution<int8_t> is not allowed in the C++17 standard
|
||||
// and vs2017/19
|
||||
fillTensorRandom<int8_t, int32_t>(input_tensor,
|
||||
std::numeric_limits<int8_t>::min(),
|
||||
std::numeric_limits<int8_t>::max());
|
||||
} else if (type == ov::element::u16) {
|
||||
fillTensorRandom<uint16_t, uint16_t>(input_tensor);
|
||||
} else if (type == ov::element::i16) {
|
||||
fillTensorRandom<int16_t, int16_t>(input_tensor);
|
||||
} else if (type == ov::element::boolean) {
|
||||
fillTensorRandom<uint8_t, uint32_t>(input_tensor, 0, 1);
|
||||
} else {
|
||||
throw std::logic_error(
|
||||
"Input precision is not supported for " + inputs[i].get_element_type().get_type_name());
|
||||
|
||||
@ -24,6 +24,14 @@
|
||||
<model name="mtcnn-r" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
<device name="CPU" iterations="1000" processes="1" threads="1">
|
||||
<model name="bert-base-ner" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
<device name="GPU" iterations="1000" processes="1" threads="1">
|
||||
<model name="bert-base-ner" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
<device name="CPU" iterations="1000" processes="1" threads="1">
|
||||
<model name="alexnet" precision="FP32" source="omz" />
|
||||
<model name="mobilenet-ssd" precision="FP32" source="omz" />
|
||||
|
||||
@ -8,4 +8,12 @@
|
||||
<model name="mobilenet-ssd" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
<device name="CPU" iterations="30" processes="1" threads="1">
|
||||
<model name="bert-base-ner" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
<device name="GPU" iterations="30" processes="1" threads="1">
|
||||
<model name="bert-base-ner" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
</cases>
|
||||
|
||||
@ -23,7 +23,15 @@
|
||||
<device name="GPU" iterations="5000" processes="1" threads="1">
|
||||
<model name="mtcnn-r" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
|
||||
<device name="CPU" iterations="5000" processes="1" threads="1">
|
||||
<model name="bert-base-ner" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
<device name="GPU" iterations="5000" processes="1" threads="1">
|
||||
<model name="bert-base-ner" precision="FP32" source="omz" />
|
||||
</device>
|
||||
|
||||
<device name="CPU" iterations="5000" processes="1" threads="1">
|
||||
<model name="alexnet" precision="FP32" source="omz" />
|
||||
<model name="mobilenet-ssd" precision="FP32" source="omz" />
|
||||
|
||||
Loading…
Reference in New Issue
Block a user