1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
|
// Generated file (from: avg_pool_float_4.mod.py). Do not edit
void CreateModel(Model *model) {
OperandType type1(Type::INT32, {});
OperandType type2(Type::TENSOR_FLOAT32, {5, 11, 13, 3});
OperandType type0(Type::TENSOR_FLOAT32, {5, 52, 60, 3});
// Phase 1, operands
auto i0 = model->addOperand(&type0);
auto stride = model->addOperand(&type1);
auto filter = model->addOperand(&type1);
auto padding = model->addOperand(&type1);
auto relu6_activation = model->addOperand(&type1);
auto output = model->addOperand(&type2);
// Phase 2, operations
static int32_t stride_init[] = {5};
model->setOperandValue(stride, stride_init, sizeof(int32_t) * 1);
static int32_t filter_init[] = {100};
model->setOperandValue(filter, filter_init, sizeof(int32_t) * 1);
static int32_t padding_init[] = {50};
model->setOperandValue(padding, padding_init, sizeof(int32_t) * 1);
static int32_t relu6_activation_init[] = {3};
model->setOperandValue(relu6_activation, relu6_activation_init, sizeof(int32_t) * 1);
model->addOperation(ANEURALNETWORKS_AVERAGE_POOL_2D, {i0, padding, padding, padding, padding, stride, stride, filter, filter, relu6_activation}, {output});
// Phase 3, inputs and outputs
model->identifyInputsAndOutputs(
{i0},
{output});
assert(model->isValid());
}
bool is_ignored(int i) {
static std::set<int> ignore = {};
return ignore.find(i) != ignore.end();
}
|
