// clang-format off // Generated file (from: softmax_v1_2.mod.py). Do not edit void CreateModel(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dim1_axis0(Model *model) { OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dim3_axis2(Model *model) { OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_relaxed(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_relaxed(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_relaxed_dim1_axis0(Model *model) { OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_relaxed_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_relaxed_dim3_axis2(Model *model) { OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_relaxed_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_float16(Model *model) { OperandType type14(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); OperandType type15(Type::FLOAT16, {}); // Phase 1, operands auto op1 = model->addOperand(&type14); auto param = model->addOperand(&type15); auto op2 = model->addOperand(&type14); // Phase 2, operations static _Float16 param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_float16(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_float16_dim1_axis0(Model *model) { OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param = model->addOperand(&type15); auto op2 = model->addOperand(&type17); // Phase 2, operations static _Float16 param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_float16_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_float16_dim3_axis2(Model *model) { OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param = model->addOperand(&type15); auto op2 = model->addOperand(&type18); // Phase 2, operations static _Float16 param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_float16_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_quant8(Model *model) { OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type20(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type20); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_quant8(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_quant8_dim1_axis0(Model *model) { OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type22(Type::TENSOR_QUANT8_ASYMM, {5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type22); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_quant8_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_quant8_dim3_axis2(Model *model) { OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type24(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type24); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_quant8_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_dim1_axis0(Model *model) { OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_dim3_axis2(Model *model) { OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_relaxed(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_relaxed(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_relaxed_dim1_axis0(Model *model) { OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_relaxed_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_relaxed_dim3_axis2(Model *model) { OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_relaxed_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_float16(Model *model) { OperandType type14(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type14); auto param = model->addOperand(&type15); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_float16(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_float16_dim1_axis0(Model *model) { OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); OperandType type28(Type::TENSOR_FLOAT16, {0}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param = model->addOperand(&type15); auto op2 = model->addOperand(&type28); // Phase 2, operations static _Float16 param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_float16_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_float16_dim3_axis2(Model *model) { OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param = model->addOperand(&type15); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_float16_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_quant8(Model *model) { OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_quant8(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_quant8_dim1_axis0(Model *model) { OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type31(Type::TENSOR_QUANT8_ASYMM, {0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type31); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_quant8_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_quant8_dim3_axis2(Model *model) { OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param = model->addOperand(&type2); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param_init[] = {1.0f}; model->setOperandValue(param, param_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_quant8_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dim1_axis0_2(Model *model) { OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dim3_axis2_2(Model *model) { OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_relaxed_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_relaxed_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_relaxed_dim1_axis0_2(Model *model) { OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_relaxed_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_relaxed_dim3_axis2_2(Model *model) { OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_relaxed_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_float16_2(Model *model) { OperandType type14(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); OperandType type15(Type::FLOAT16, {}); // Phase 1, operands auto op1 = model->addOperand(&type14); auto param1 = model->addOperand(&type15); auto op2 = model->addOperand(&type14); // Phase 2, operations static _Float16 param1_init[] = {9.999999974752427e-07f}; model->setOperandValue(param1, param1_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_float16_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_float16_dim1_axis0_2(Model *model) { OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param1 = model->addOperand(&type15); auto op2 = model->addOperand(&type17); // Phase 2, operations static _Float16 param1_init[] = {9.999999974752427e-07f}; model->setOperandValue(param1, param1_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_float16_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_float16_dim3_axis2_2(Model *model) { OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param1 = model->addOperand(&type15); auto op2 = model->addOperand(&type18); // Phase 2, operations static _Float16 param1_init[] = {9.999999974752427e-07f}; model->setOperandValue(param1, param1_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_float16_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_quant8_2(Model *model) { OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type20(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type20); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_quant8_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_quant8_dim1_axis0_2(Model *model) { OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type22(Type::TENSOR_QUANT8_ASYMM, {5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type22); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_quant8_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_quant8_dim3_axis2_2(Model *model) { OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type24(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type24); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_quant8_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_dim1_axis0_2(Model *model) { OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_dim3_axis2_2(Model *model) { OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_relaxed_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_relaxed_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_relaxed_dim1_axis0_2(Model *model) { OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_relaxed_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_relaxed_dim3_axis2_2(Model *model) { OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_relaxed_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_float16_2(Model *model) { OperandType type14(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type14); auto param1 = model->addOperand(&type15); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param1_init[] = {9.999999974752427e-07f}; model->setOperandValue(param1, param1_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_float16_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_float16_dim1_axis0_2(Model *model) { OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); OperandType type28(Type::TENSOR_FLOAT16, {0}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param1 = model->addOperand(&type15); auto op2 = model->addOperand(&type28); // Phase 2, operations static _Float16 param1_init[] = {9.999999974752427e-07f}; model->setOperandValue(param1, param1_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_float16_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_float16_dim3_axis2_2(Model *model) { OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param1 = model->addOperand(&type15); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param1_init[] = {9.999999974752427e-07f}; model->setOperandValue(param1, param1_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_float16_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_quant8_2(Model *model) { OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_quant8_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_quant8_dim1_axis0_2(Model *model) { OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type31(Type::TENSOR_QUANT8_ASYMM, {0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type31); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_quant8_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_dynamic_output_shape_quant8_dim3_axis2_2(Model *model) { OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param1 = model->addOperand(&type2); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param1_init[] = {1e-06f}; model->setOperandValue(param1, param1_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param1}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_dynamic_output_shape_quant8_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type33); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type33); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type34); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type34); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type35); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type35); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis3(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis3(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis3_neg(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis3_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type36); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type36); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type37); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type37); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim2_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type38); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim2_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim2_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type38); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim2_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim2_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type39); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim2_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim2_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type39); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim2_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim1_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim1_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim1_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type33); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type33); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type34); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type34); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type35); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type35); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis3(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis3(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis3_neg(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis3_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type36); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type36); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type37); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type37); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim2_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type38); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim2_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim2_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type38); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim2_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim2_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type39); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim2_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim2_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type39); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim2_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim1_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim1_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim1_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type40(Type::TENSOR_FLOAT16, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type40); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type40); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type40(Type::TENSOR_FLOAT16, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type40); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type40); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type41(Type::TENSOR_FLOAT16, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type41); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type41); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type41(Type::TENSOR_FLOAT16, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type41); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type41); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type42(Type::TENSOR_FLOAT16, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type42); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type42); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type42(Type::TENSOR_FLOAT16, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type42); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type42); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis3(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type16(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type16); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type16); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis3(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis3_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type16(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type16); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type16); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis3_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type43(Type::TENSOR_FLOAT16, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type43); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type43); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type43(Type::TENSOR_FLOAT16, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type43); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type43); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type44(Type::TENSOR_FLOAT16, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type44); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type44); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type44(Type::TENSOR_FLOAT16, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type44); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type44); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type18); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type18); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim2_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type45(Type::TENSOR_FLOAT16, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type45); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type45); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim2_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim2_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type45(Type::TENSOR_FLOAT16, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type45); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type45); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim2_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim2_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type46(Type::TENSOR_FLOAT16, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type46); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type46); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim2_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim2_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type46(Type::TENSOR_FLOAT16, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type46); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type46); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim2_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim1_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type17); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim1_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type17); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim1_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type47(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.25f, 128); OperandType type48(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type47); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type48); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type47(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.25f, 128); OperandType type48(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type47); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type48); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type49(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.25f, 128); OperandType type50(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type49); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type50); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type49(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.25f, 128); OperandType type50(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type49); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type50); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type51(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.25f, 128); OperandType type52(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type51); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type52); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type51(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.25f, 128); OperandType type52(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type51); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type52); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis3(Model *model) { OperandType type1(Type::INT32, {}); OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type20(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type20); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis3(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis3_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type20(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type20); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis3_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type53(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.25f, 128); OperandType type54(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type53); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type54); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type53(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.25f, 128); OperandType type54(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type53); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type54); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type55(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.25f, 128); OperandType type56(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type55); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type56); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type55(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.25f, 128); OperandType type56(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type55); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type56); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type24(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type24); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type24(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type24); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim2_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type57(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.25f, 128); OperandType type58(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type57); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type58); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim2_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim2_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type57(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.25f, 128); OperandType type58(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type57); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type58); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim2_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim2_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type59(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.25f, 128); OperandType type60(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type59); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type60); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim2_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim2_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type59(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.25f, 128); OperandType type60(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type59); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type60); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim2_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim1_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type22(Type::TENSOR_QUANT8_ASYMM, {5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type22); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim1_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type22(Type::TENSOR_QUANT8_ASYMM, {5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type22); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim1_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis3(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis3(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis3_neg(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis3_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim2_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim2_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim2_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim2_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim2_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim2_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim2_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim2_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim1_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim1_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim1_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis3(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis3(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis3_neg(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis3_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim2_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim2_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim2_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim2_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim2_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim2_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim2_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim2_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim1_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim1_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim1_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type40(Type::TENSOR_FLOAT16, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type40); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type40(Type::TENSOR_FLOAT16, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type40); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type41(Type::TENSOR_FLOAT16, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type41); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type41(Type::TENSOR_FLOAT16, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type41); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type42(Type::TENSOR_FLOAT16, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type42); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type42(Type::TENSOR_FLOAT16, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type42); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis3(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type16(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type16); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis3(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis3_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type16(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type16); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis3_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); OperandType type43(Type::TENSOR_FLOAT16, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type43); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); OperandType type43(Type::TENSOR_FLOAT16, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type43); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); OperandType type44(Type::TENSOR_FLOAT16, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type44); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); OperandType type44(Type::TENSOR_FLOAT16, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type44); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim2_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type45(Type::TENSOR_FLOAT16, {5, 2}); OperandType type62(Type::TENSOR_FLOAT16, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type45); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type62); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim2_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim2_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type45(Type::TENSOR_FLOAT16, {5, 2}); OperandType type62(Type::TENSOR_FLOAT16, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type45); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type62); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim2_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim2_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type46(Type::TENSOR_FLOAT16, {2, 5}); OperandType type62(Type::TENSOR_FLOAT16, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type46); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type62); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim2_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim2_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type46(Type::TENSOR_FLOAT16, {2, 5}); OperandType type62(Type::TENSOR_FLOAT16, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type46); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type62); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim2_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim1_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); OperandType type28(Type::TENSOR_FLOAT16, {0}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type28); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim1_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); OperandType type28(Type::TENSOR_FLOAT16, {0}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param2 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type28); // Phase 2, operations static _Float16 param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim1_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type47(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type47); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type47(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type47); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type49(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type49); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type49(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type49); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type51(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type51); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type51(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type51); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis3(Model *model) { OperandType type1(Type::INT32, {}); OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis3(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis3_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis3_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); OperandType type53(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type53); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); OperandType type53(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type53); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); OperandType type55(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type55); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); OperandType type55(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type55); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis2_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis2_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim2_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type57(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.25f, 128); OperandType type63(Type::TENSOR_QUANT8_ASYMM, {0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type57); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type63); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim2_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim2_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type57(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.25f, 128); OperandType type63(Type::TENSOR_QUANT8_ASYMM, {0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type57); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type63); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim2_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim2_axis1(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type59(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.25f, 128); OperandType type63(Type::TENSOR_QUANT8_ASYMM, {0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type59); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type63); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim2_axis1(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim2_axis1_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type59(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.25f, 128); OperandType type63(Type::TENSOR_QUANT8_ASYMM, {0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type59); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type63); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim2_axis1_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim1_axis0(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type31(Type::TENSOR_QUANT8_ASYMM, {0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type31); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim1_axis0(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim1_axis0_neg(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type31(Type::TENSOR_QUANT8_ASYMM, {0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param2 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type31); // Phase 2, operations static float param2_init[] = {1.0f}; model->setOperandValue(param2, param2_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param2, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim1_axis0_neg(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type33); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type33); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type34); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type34); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type35); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type35); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis3_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis3_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim4_axis3_neg_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim4_axis3_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type36); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type36); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type37); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type37); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim3_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim3_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim2_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type38); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim2_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim2_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type38); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim2_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim2_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type39); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim2_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim2_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type39); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim2_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim1_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dim1_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dim1_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type33); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type33); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type34); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type34); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type35); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type35); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis3_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis3_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim4_axis3_neg_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type0); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim4_axis3_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type36); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type36); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type37); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type37); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim3_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type13); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim3_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim2_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type38); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim2_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim2_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type38); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim2_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim2_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type39); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim2_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim2_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type39); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim2_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim1_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_relaxed_dim1_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type12); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_relaxed_dim1_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type40(Type::TENSOR_FLOAT16, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type40); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type40); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type40(Type::TENSOR_FLOAT16, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type40); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type40); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type41(Type::TENSOR_FLOAT16, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type41); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type41); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type41(Type::TENSOR_FLOAT16, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type41); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type41); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type42(Type::TENSOR_FLOAT16, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type42); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type42); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type42(Type::TENSOR_FLOAT16, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type42); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type42); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis3_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type16(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type16); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type16); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis3_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim4_axis3_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type16(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type16); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type16); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim4_axis3_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type43(Type::TENSOR_FLOAT16, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type43); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type43); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type43(Type::TENSOR_FLOAT16, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type43); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type43); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type44(Type::TENSOR_FLOAT16, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type44); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type44); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type44(Type::TENSOR_FLOAT16, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type44); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type44); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type18); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim3_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type18); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim3_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim2_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type45(Type::TENSOR_FLOAT16, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type45); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type45); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim2_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim2_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type45(Type::TENSOR_FLOAT16, {5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type45); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type45); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim2_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim2_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type46(Type::TENSOR_FLOAT16, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type46); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type46); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim2_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim2_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type46(Type::TENSOR_FLOAT16, {2, 5}); // Phase 1, operands auto op1 = model->addOperand(&type46); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type46); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim2_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim1_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type17); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_float16_dim1_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type17); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_float16_dim1_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type47(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.25f, 128); OperandType type48(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type47); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type48); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type47(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.25f, 128); OperandType type48(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type47); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type48); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type49(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.25f, 128); OperandType type50(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type49); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type50); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type49(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.25f, 128); OperandType type50(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type49); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type50); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type51(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.25f, 128); OperandType type52(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type51); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type52); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type51(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.25f, 128); OperandType type52(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type51); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type52); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis3_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type20(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type20); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis3_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim4_axis3_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type20(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type20); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim4_axis3_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type53(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.25f, 128); OperandType type54(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type53); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type54); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type53(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.25f, 128); OperandType type54(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type53); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type54); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type55(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.25f, 128); OperandType type56(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type55); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type56); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type55(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.25f, 128); OperandType type56(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type55); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type56); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type24(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type24); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim3_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type24(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type24); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim3_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim2_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type57(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.25f, 128); OperandType type58(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type57); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type58); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim2_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim2_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type57(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.25f, 128); OperandType type58(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type57); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type58); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim2_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim2_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type59(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.25f, 128); OperandType type60(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type59); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type60); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim2_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim2_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type59(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.25f, 128); OperandType type60(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type59); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type60); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim2_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim1_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type22(Type::TENSOR_QUANT8_ASYMM, {5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type22); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_quant8_dim1_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type22(Type::TENSOR_QUANT8_ASYMM, {5}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type22); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_quant8_dim1_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis3_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis3_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim4_axis3_neg_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim4_axis3_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim3_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim3_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim2_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim2_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim2_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim2_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim2_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim2_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim2_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim2_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim1_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_dim1_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_dim1_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type33(Type::TENSOR_FLOAT32, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type33); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type34(Type::TENSOR_FLOAT32, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type34); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type35(Type::TENSOR_FLOAT32, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type35); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis3_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis3_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim4_axis3_neg_2(Model *model) { OperandType type0(Type::TENSOR_FLOAT32, {2, 2, 2, 5}); OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type0); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type25); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim4_axis3_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type36(Type::TENSOR_FLOAT32, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type36); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); OperandType type37(Type::TENSOR_FLOAT32, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type37); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim3_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type13(Type::TENSOR_FLOAT32, {2, 2, 5}); OperandType type2(Type::FLOAT32, {}); OperandType type26(Type::TENSOR_FLOAT32, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type13); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type26); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim3_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim2_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim2_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim2_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type38(Type::TENSOR_FLOAT32, {5, 2}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type38); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim2_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim2_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim2_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim2_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type39(Type::TENSOR_FLOAT32, {2, 5}); OperandType type61(Type::TENSOR_FLOAT32, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type39); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type61); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim2_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim1_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_relaxed_dim1_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type12(Type::TENSOR_FLOAT32, {5}); OperandType type2(Type::FLOAT32, {}); OperandType type5(Type::TENSOR_FLOAT32, {0}); // Phase 1, operands auto op1 = model->addOperand(&type12); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type5); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_relaxed_dim1_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type40(Type::TENSOR_FLOAT16, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type40); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type40(Type::TENSOR_FLOAT16, {5, 2, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type40); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type41(Type::TENSOR_FLOAT16, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type41); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type41(Type::TENSOR_FLOAT16, {2, 5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type41); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type42(Type::TENSOR_FLOAT16, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type42); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type42(Type::TENSOR_FLOAT16, {2, 2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type42); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis3_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type16(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type16); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis3_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim4_axis3_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type16(Type::TENSOR_FLOAT16, {2, 2, 2, 5}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type16); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type27); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim4_axis3_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); OperandType type43(Type::TENSOR_FLOAT16, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type43); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); OperandType type43(Type::TENSOR_FLOAT16, {5, 2, 2}); // Phase 1, operands auto op1 = model->addOperand(&type43); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); OperandType type44(Type::TENSOR_FLOAT16, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type44); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); OperandType type44(Type::TENSOR_FLOAT16, {2, 5, 2}); // Phase 1, operands auto op1 = model->addOperand(&type44); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim3_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type18(Type::TENSOR_FLOAT16, {2, 2, 5}); OperandType type29(Type::TENSOR_FLOAT16, {0, 0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type18); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type29); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim3_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim2_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type45(Type::TENSOR_FLOAT16, {5, 2}); OperandType type62(Type::TENSOR_FLOAT16, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type45); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type62); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim2_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim2_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type45(Type::TENSOR_FLOAT16, {5, 2}); OperandType type62(Type::TENSOR_FLOAT16, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type45); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type62); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim2_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim2_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type46(Type::TENSOR_FLOAT16, {2, 5}); OperandType type62(Type::TENSOR_FLOAT16, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type46); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type62); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim2_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim2_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type46(Type::TENSOR_FLOAT16, {2, 5}); OperandType type62(Type::TENSOR_FLOAT16, {0, 0}); // Phase 1, operands auto op1 = model->addOperand(&type46); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type62); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim2_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim1_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); OperandType type28(Type::TENSOR_FLOAT16, {0}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type28); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_float16_dim1_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type17(Type::TENSOR_FLOAT16, {5}); OperandType type28(Type::TENSOR_FLOAT16, {0}); // Phase 1, operands auto op1 = model->addOperand(&type17); auto param3 = model->addOperand(&type15); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type28); // Phase 2, operations static _Float16 param3_init[] = {9.999999974752427e-07f}; model->setOperandValue(param3, param3_init, sizeof(_Float16) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_float16_dim1_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type47(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type47); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type47(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type47); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-4}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type49(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type49); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type49(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type49); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type51(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type51); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type51(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type51); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis3_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis3_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim4_axis3_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type19(Type::TENSOR_QUANT8_ASYMM, {2, 2, 2, 5}, 0.25f, 128); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type19); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type30); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim4_axis3_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); OperandType type53(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type53); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); OperandType type53(Type::TENSOR_QUANT8_ASYMM, {5, 2, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type53); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-3}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); OperandType type55(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type55); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); OperandType type55(Type::TENSOR_QUANT8_ASYMM, {2, 5, 2}, 0.25f, 128); // Phase 1, operands auto op1 = model->addOperand(&type55); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis2_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis2_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim3_axis2_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type23(Type::TENSOR_QUANT8_ASYMM, {2, 2, 5}, 0.25f, 128); OperandType type32(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type23); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type32); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim3_axis2_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim2_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type57(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.25f, 128); OperandType type63(Type::TENSOR_QUANT8_ASYMM, {0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type57); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type63); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim2_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim2_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type57(Type::TENSOR_QUANT8_ASYMM, {5, 2}, 0.25f, 128); OperandType type63(Type::TENSOR_QUANT8_ASYMM, {0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type57); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type63); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-2}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim2_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim2_axis1_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type59(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.25f, 128); OperandType type63(Type::TENSOR_QUANT8_ASYMM, {0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type59); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type63); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim2_axis1_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim2_axis1_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type59(Type::TENSOR_QUANT8_ASYMM, {2, 5}, 0.25f, 128); OperandType type63(Type::TENSOR_QUANT8_ASYMM, {0, 0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type59); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type63); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim2_axis1_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim1_axis0_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type31(Type::TENSOR_QUANT8_ASYMM, {0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type31); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {0}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim1_axis0_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_axis_dynamic_output_shape_quant8_dim1_axis0_neg_2(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type21(Type::TENSOR_QUANT8_ASYMM, {5}, 0.25f, 128); OperandType type31(Type::TENSOR_QUANT8_ASYMM, {0}, 0.00390625f, 0); // Phase 1, operands auto op1 = model->addOperand(&type21); auto param3 = model->addOperand(&type2); auto axis = model->addOperand(&type1); auto op2 = model->addOperand(&type31); // Phase 2, operations static float param3_init[] = {1e-06f}; model->setOperandValue(param3, param3_init, sizeof(float) * 1); static int32_t axis_init[] = {-1}; model->setOperandValue(axis, axis_init, sizeof(int32_t) * 1); model->addOperation(ANEURALNETWORKS_SOFTMAX, {op1, param3, axis}, {op2}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {op1}, {op2}); assert(model->isValid()); } inline bool is_ignored_axis_dynamic_output_shape_quant8_dim1_axis0_neg_2(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_zero_sized(Model *model) { OperandType type1(Type::INT32, {}); OperandType type10(Type::TENSOR_FLOAT32, {1, 1, 1, 1}); OperandType type11(Type::TENSOR_FLOAT32, {0, 2, 2, 1}); OperandType type2(Type::FLOAT32, {}); OperandType type3(Type::TENSOR_FLOAT32, {1, 2}); OperandType type4(Type::TENSOR_FLOAT32, {1, 8}); OperandType type5(Type::TENSOR_FLOAT32, {0}); OperandType type6(Type::TENSOR_INT32, {0}); OperandType type7(Type::TENSOR_FLOAT32, {0, 4}); OperandType type8(Type::TENSOR_INT32, {1}); OperandType type9(Type::BOOL, {}); // Phase 1, operands auto scores = model->addOperand(&type3); auto roi = model->addOperand(&type4); auto param4 = model->addOperand(&type8); auto param5 = model->addOperand(&type2); auto param6 = model->addOperand(&type1); auto param7 = model->addOperand(&type1); auto param8 = model->addOperand(&type2); auto param9 = model->addOperand(&type2); auto param10 = model->addOperand(&type2); auto scoresOut = model->addOperand(&type5); auto roiOut = model->addOperand(&type7); auto classesOut = model->addOperand(&type6); auto batchSplitOut = model->addOperand(&type6); auto in = model->addOperand(&type10); auto param11 = model->addOperand(&type1); auto param12 = model->addOperand(&type1); auto param13 = model->addOperand(&type2); auto param14 = model->addOperand(&type2); auto param15 = model->addOperand(&type1); auto param16 = model->addOperand(&type1); auto layout = model->addOperand(&type9); auto featureMap = model->addOperand(&type11); auto param17 = model->addOperand(&type2); auto out = model->addOperand(&type11); // Phase 2, operations static float scores_init[] = {0.9f, 0.1f}; model->setOperandValue(scores, scores_init, sizeof(float) * 2); static float roi_init[] = {1.0f, 1.0f, 10.0f, 10.0f, 0.0f, 0.0f, 10.0f, 10.0f}; model->setOperandValue(roi, roi_init, sizeof(float) * 8); static int32_t param4_init[] = {0}; model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1); static float param5_init[] = {0.3f}; model->setOperandValue(param5, param5_init, sizeof(float) * 1); static int32_t param6_init[] = {-1}; model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1); static int32_t param7_init[] = {0}; model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1); static float param8_init[] = {0.4f}; model->setOperandValue(param8, param8_init, sizeof(float) * 1); static float param9_init[] = {1.0f}; model->setOperandValue(param9, param9_init, sizeof(float) * 1); static float param10_init[] = {0.3f}; model->setOperandValue(param10, param10_init, sizeof(float) * 1); static int32_t param11_init[] = {2}; model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1); static int32_t param12_init[] = {2}; model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1); static float param13_init[] = {2.0f}; model->setOperandValue(param13, param13_init, sizeof(float) * 1); static float param14_init[] = {2.0f}; model->setOperandValue(param14, param14_init, sizeof(float) * 1); static int32_t param15_init[] = {4}; model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1); static int32_t param16_init[] = {4}; model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1); static bool8 layout_init[] = {false}; model->setOperandValue(layout, layout_init, sizeof(bool8) * 1); static float param17_init[] = {1.0f}; model->setOperandValue(param17, param17_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_BOX_WITH_NMS_LIMIT, {scores, roi, param4, param5, param6, param7, param8, param9, param10}, {scoresOut, roiOut, classesOut, batchSplitOut}); model->addOperation(ANEURALNETWORKS_ROI_ALIGN, {in, roiOut, batchSplitOut, param11, param12, param13, param14, param15, param16, layout}, {featureMap}); model->addOperation(ANEURALNETWORKS_SOFTMAX, {featureMap, param17}, {out}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {in}, {scoresOut, classesOut, out}); assert(model->isValid()); } inline bool is_ignored_zero_sized(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_zero_sized_relaxed(Model *model) { OperandType type1(Type::INT32, {}); OperandType type10(Type::TENSOR_FLOAT32, {1, 1, 1, 1}); OperandType type11(Type::TENSOR_FLOAT32, {0, 2, 2, 1}); OperandType type2(Type::FLOAT32, {}); OperandType type3(Type::TENSOR_FLOAT32, {1, 2}); OperandType type4(Type::TENSOR_FLOAT32, {1, 8}); OperandType type5(Type::TENSOR_FLOAT32, {0}); OperandType type6(Type::TENSOR_INT32, {0}); OperandType type7(Type::TENSOR_FLOAT32, {0, 4}); OperandType type8(Type::TENSOR_INT32, {1}); OperandType type9(Type::BOOL, {}); // Phase 1, operands auto scores = model->addOperand(&type3); auto roi = model->addOperand(&type4); auto param4 = model->addOperand(&type8); auto param5 = model->addOperand(&type2); auto param6 = model->addOperand(&type1); auto param7 = model->addOperand(&type1); auto param8 = model->addOperand(&type2); auto param9 = model->addOperand(&type2); auto param10 = model->addOperand(&type2); auto scoresOut = model->addOperand(&type5); auto roiOut = model->addOperand(&type7); auto classesOut = model->addOperand(&type6); auto batchSplitOut = model->addOperand(&type6); auto in = model->addOperand(&type10); auto param11 = model->addOperand(&type1); auto param12 = model->addOperand(&type1); auto param13 = model->addOperand(&type2); auto param14 = model->addOperand(&type2); auto param15 = model->addOperand(&type1); auto param16 = model->addOperand(&type1); auto layout = model->addOperand(&type9); auto featureMap = model->addOperand(&type11); auto param17 = model->addOperand(&type2); auto out = model->addOperand(&type11); // Phase 2, operations static float scores_init[] = {0.9f, 0.1f}; model->setOperandValue(scores, scores_init, sizeof(float) * 2); static float roi_init[] = {1.0f, 1.0f, 10.0f, 10.0f, 0.0f, 0.0f, 10.0f, 10.0f}; model->setOperandValue(roi, roi_init, sizeof(float) * 8); static int32_t param4_init[] = {0}; model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1); static float param5_init[] = {0.3f}; model->setOperandValue(param5, param5_init, sizeof(float) * 1); static int32_t param6_init[] = {-1}; model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1); static int32_t param7_init[] = {0}; model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1); static float param8_init[] = {0.4f}; model->setOperandValue(param8, param8_init, sizeof(float) * 1); static float param9_init[] = {1.0f}; model->setOperandValue(param9, param9_init, sizeof(float) * 1); static float param10_init[] = {0.3f}; model->setOperandValue(param10, param10_init, sizeof(float) * 1); static int32_t param11_init[] = {2}; model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1); static int32_t param12_init[] = {2}; model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1); static float param13_init[] = {2.0f}; model->setOperandValue(param13, param13_init, sizeof(float) * 1); static float param14_init[] = {2.0f}; model->setOperandValue(param14, param14_init, sizeof(float) * 1); static int32_t param15_init[] = {4}; model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1); static int32_t param16_init[] = {4}; model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1); static bool8 layout_init[] = {false}; model->setOperandValue(layout, layout_init, sizeof(bool8) * 1); static float param17_init[] = {1.0f}; model->setOperandValue(param17, param17_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_BOX_WITH_NMS_LIMIT, {scores, roi, param4, param5, param6, param7, param8, param9, param10}, {scoresOut, roiOut, classesOut, batchSplitOut}); model->addOperation(ANEURALNETWORKS_ROI_ALIGN, {in, roiOut, batchSplitOut, param11, param12, param13, param14, param15, param16, layout}, {featureMap}); model->addOperation(ANEURALNETWORKS_SOFTMAX, {featureMap, param17}, {out}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {in}, {scoresOut, classesOut, out}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_zero_sized_relaxed(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_zero_sized_quant8(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type6(Type::TENSOR_INT32, {0}); OperandType type64(Type::TENSOR_QUANT8_ASYMM, {0, 2, 2, 1}, 0.1f, 128); OperandType type65(Type::TENSOR_QUANT8_ASYMM, {1, 1, 1, 1}, 0.1f, 128); OperandType type66(Type::TENSOR_QUANT8_ASYMM, {0, 2, 2, 1}, 0.00390625f, 0); OperandType type67(Type::TENSOR_QUANT16_ASYMM, {1, 8}, 0.125f, 0); OperandType type68(Type::TENSOR_QUANT16_ASYMM, {0, 4}, 0.125f, 0); OperandType type69(Type::TENSOR_QUANT8_ASYMM, {1, 2}, 0.1f, 128); OperandType type70(Type::TENSOR_QUANT8_ASYMM, {0}, 0.1f, 128); OperandType type8(Type::TENSOR_INT32, {1}); OperandType type9(Type::BOOL, {}); // Phase 1, operands auto scores = model->addOperand(&type69); auto roi = model->addOperand(&type67); auto param4 = model->addOperand(&type8); auto param5 = model->addOperand(&type2); auto param6 = model->addOperand(&type1); auto param7 = model->addOperand(&type1); auto param8 = model->addOperand(&type2); auto param9 = model->addOperand(&type2); auto param10 = model->addOperand(&type2); auto scoresOut = model->addOperand(&type70); auto roiOut = model->addOperand(&type68); auto classesOut = model->addOperand(&type6); auto batchSplitOut = model->addOperand(&type6); auto in = model->addOperand(&type65); auto param11 = model->addOperand(&type1); auto param12 = model->addOperand(&type1); auto param13 = model->addOperand(&type2); auto param14 = model->addOperand(&type2); auto param15 = model->addOperand(&type1); auto param16 = model->addOperand(&type1); auto layout = model->addOperand(&type9); auto featureMap = model->addOperand(&type64); auto param17 = model->addOperand(&type2); auto out = model->addOperand(&type66); // Phase 2, operations static uint8_t scores_init[] = {137, 129}; model->setOperandValue(scores, scores_init, sizeof(uint8_t) * 2); static uint16_t roi_init[] = {8, 8, 80, 80, 0, 0, 80, 80}; model->setOperandValue(roi, roi_init, sizeof(uint16_t) * 8); static int32_t param4_init[] = {0}; model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1); static float param5_init[] = {0.3f}; model->setOperandValue(param5, param5_init, sizeof(float) * 1); static int32_t param6_init[] = {-1}; model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1); static int32_t param7_init[] = {0}; model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1); static float param8_init[] = {0.4f}; model->setOperandValue(param8, param8_init, sizeof(float) * 1); static float param9_init[] = {1.0f}; model->setOperandValue(param9, param9_init, sizeof(float) * 1); static float param10_init[] = {0.3f}; model->setOperandValue(param10, param10_init, sizeof(float) * 1); static int32_t param11_init[] = {2}; model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1); static int32_t param12_init[] = {2}; model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1); static float param13_init[] = {2.0f}; model->setOperandValue(param13, param13_init, sizeof(float) * 1); static float param14_init[] = {2.0f}; model->setOperandValue(param14, param14_init, sizeof(float) * 1); static int32_t param15_init[] = {4}; model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1); static int32_t param16_init[] = {4}; model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1); static bool8 layout_init[] = {false}; model->setOperandValue(layout, layout_init, sizeof(bool8) * 1); static float param17_init[] = {1.0f}; model->setOperandValue(param17, param17_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_BOX_WITH_NMS_LIMIT, {scores, roi, param4, param5, param6, param7, param8, param9, param10}, {scoresOut, roiOut, classesOut, batchSplitOut}); model->addOperation(ANEURALNETWORKS_ROI_ALIGN, {in, roiOut, batchSplitOut, param11, param12, param13, param14, param15, param16, layout}, {featureMap}); model->addOperation(ANEURALNETWORKS_SOFTMAX, {featureMap, param17}, {out}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {in}, {scoresOut, classesOut, out}); assert(model->isValid()); } inline bool is_ignored_zero_sized_quant8(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_zero_sized_float16(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type6(Type::TENSOR_INT32, {0}); OperandType type71(Type::TENSOR_FLOAT16, {0, 2, 2, 1}); OperandType type72(Type::TENSOR_FLOAT16, {1, 1, 1, 1}); OperandType type73(Type::TENSOR_FLOAT16, {1, 8}); OperandType type74(Type::TENSOR_FLOAT16, {0, 4}); OperandType type75(Type::TENSOR_FLOAT16, {1, 2}); OperandType type76(Type::TENSOR_FLOAT16, {0}); OperandType type8(Type::TENSOR_INT32, {1}); OperandType type9(Type::BOOL, {}); // Phase 1, operands auto scores = model->addOperand(&type75); auto roi = model->addOperand(&type73); auto param4 = model->addOperand(&type8); auto param5 = model->addOperand(&type15); auto param6 = model->addOperand(&type1); auto param7 = model->addOperand(&type1); auto param8 = model->addOperand(&type15); auto param9 = model->addOperand(&type15); auto param10 = model->addOperand(&type15); auto scoresOut = model->addOperand(&type76); auto roiOut = model->addOperand(&type74); auto classesOut = model->addOperand(&type6); auto batchSplitOut = model->addOperand(&type6); auto in = model->addOperand(&type72); auto param11 = model->addOperand(&type1); auto param12 = model->addOperand(&type1); auto param13 = model->addOperand(&type15); auto param14 = model->addOperand(&type15); auto param15 = model->addOperand(&type1); auto param16 = model->addOperand(&type1); auto layout = model->addOperand(&type9); auto featureMap = model->addOperand(&type71); auto param17 = model->addOperand(&type15); auto out = model->addOperand(&type71); // Phase 2, operations static _Float16 scores_init[] = {0.8999999761581421f, 0.10000000149011612f}; model->setOperandValue(scores, scores_init, sizeof(_Float16) * 2); static _Float16 roi_init[] = {1.0f, 1.0f, 10.0f, 10.0f, 0.0f, 0.0f, 10.0f, 10.0f}; model->setOperandValue(roi, roi_init, sizeof(_Float16) * 8); static int32_t param4_init[] = {0}; model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1); static _Float16 param5_init[] = {0.30000001192092896f}; model->setOperandValue(param5, param5_init, sizeof(_Float16) * 1); static int32_t param6_init[] = {-1}; model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1); static int32_t param7_init[] = {0}; model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1); static _Float16 param8_init[] = {0.4000000059604645f}; model->setOperandValue(param8, param8_init, sizeof(_Float16) * 1); static _Float16 param9_init[] = {1.0f}; model->setOperandValue(param9, param9_init, sizeof(_Float16) * 1); static _Float16 param10_init[] = {0.30000001192092896f}; model->setOperandValue(param10, param10_init, sizeof(_Float16) * 1); static int32_t param11_init[] = {2}; model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1); static int32_t param12_init[] = {2}; model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1); static _Float16 param13_init[] = {2.0f}; model->setOperandValue(param13, param13_init, sizeof(_Float16) * 1); static _Float16 param14_init[] = {2.0f}; model->setOperandValue(param14, param14_init, sizeof(_Float16) * 1); static int32_t param15_init[] = {4}; model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1); static int32_t param16_init[] = {4}; model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1); static bool8 layout_init[] = {false}; model->setOperandValue(layout, layout_init, sizeof(bool8) * 1); static _Float16 param17_init[] = {1.0f}; model->setOperandValue(param17, param17_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_BOX_WITH_NMS_LIMIT, {scores, roi, param4, param5, param6, param7, param8, param9, param10}, {scoresOut, roiOut, classesOut, batchSplitOut}); model->addOperation(ANEURALNETWORKS_ROI_ALIGN, {in, roiOut, batchSplitOut, param11, param12, param13, param14, param15, param16, layout}, {featureMap}); model->addOperation(ANEURALNETWORKS_SOFTMAX, {featureMap, param17}, {out}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {in}, {scoresOut, classesOut, out}); assert(model->isValid()); } inline bool is_ignored_zero_sized_float16(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_zero_sized_dynamic_output_shape(Model *model) { OperandType type1(Type::INT32, {}); OperandType type10(Type::TENSOR_FLOAT32, {1, 1, 1, 1}); OperandType type11(Type::TENSOR_FLOAT32, {0, 2, 2, 1}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type3(Type::TENSOR_FLOAT32, {1, 2}); OperandType type4(Type::TENSOR_FLOAT32, {1, 8}); OperandType type5(Type::TENSOR_FLOAT32, {0}); OperandType type6(Type::TENSOR_INT32, {0}); OperandType type7(Type::TENSOR_FLOAT32, {0, 4}); OperandType type8(Type::TENSOR_INT32, {1}); OperandType type9(Type::BOOL, {}); // Phase 1, operands auto scores = model->addOperand(&type3); auto roi = model->addOperand(&type4); auto param4 = model->addOperand(&type8); auto param5 = model->addOperand(&type2); auto param6 = model->addOperand(&type1); auto param7 = model->addOperand(&type1); auto param8 = model->addOperand(&type2); auto param9 = model->addOperand(&type2); auto param10 = model->addOperand(&type2); auto scoresOut = model->addOperand(&type5); auto roiOut = model->addOperand(&type7); auto classesOut = model->addOperand(&type6); auto batchSplitOut = model->addOperand(&type6); auto in = model->addOperand(&type10); auto param11 = model->addOperand(&type1); auto param12 = model->addOperand(&type1); auto param13 = model->addOperand(&type2); auto param14 = model->addOperand(&type2); auto param15 = model->addOperand(&type1); auto param16 = model->addOperand(&type1); auto layout = model->addOperand(&type9); auto featureMap = model->addOperand(&type11); auto param17 = model->addOperand(&type2); auto out = model->addOperand(&type25); // Phase 2, operations static float scores_init[] = {0.9f, 0.1f}; model->setOperandValue(scores, scores_init, sizeof(float) * 2); static float roi_init[] = {1.0f, 1.0f, 10.0f, 10.0f, 0.0f, 0.0f, 10.0f, 10.0f}; model->setOperandValue(roi, roi_init, sizeof(float) * 8); static int32_t param4_init[] = {0}; model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1); static float param5_init[] = {0.3f}; model->setOperandValue(param5, param5_init, sizeof(float) * 1); static int32_t param6_init[] = {-1}; model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1); static int32_t param7_init[] = {0}; model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1); static float param8_init[] = {0.4f}; model->setOperandValue(param8, param8_init, sizeof(float) * 1); static float param9_init[] = {1.0f}; model->setOperandValue(param9, param9_init, sizeof(float) * 1); static float param10_init[] = {0.3f}; model->setOperandValue(param10, param10_init, sizeof(float) * 1); static int32_t param11_init[] = {2}; model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1); static int32_t param12_init[] = {2}; model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1); static float param13_init[] = {2.0f}; model->setOperandValue(param13, param13_init, sizeof(float) * 1); static float param14_init[] = {2.0f}; model->setOperandValue(param14, param14_init, sizeof(float) * 1); static int32_t param15_init[] = {4}; model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1); static int32_t param16_init[] = {4}; model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1); static bool8 layout_init[] = {false}; model->setOperandValue(layout, layout_init, sizeof(bool8) * 1); static float param17_init[] = {1.0f}; model->setOperandValue(param17, param17_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_BOX_WITH_NMS_LIMIT, {scores, roi, param4, param5, param6, param7, param8, param9, param10}, {scoresOut, roiOut, classesOut, batchSplitOut}); model->addOperation(ANEURALNETWORKS_ROI_ALIGN, {in, roiOut, batchSplitOut, param11, param12, param13, param14, param15, param16, layout}, {featureMap}); model->addOperation(ANEURALNETWORKS_SOFTMAX, {featureMap, param17}, {out}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {in}, {scoresOut, classesOut, out}); assert(model->isValid()); } inline bool is_ignored_zero_sized_dynamic_output_shape(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_zero_sized_dynamic_output_shape_relaxed(Model *model) { OperandType type1(Type::INT32, {}); OperandType type10(Type::TENSOR_FLOAT32, {1, 1, 1, 1}); OperandType type11(Type::TENSOR_FLOAT32, {0, 2, 2, 1}); OperandType type2(Type::FLOAT32, {}); OperandType type25(Type::TENSOR_FLOAT32, {0, 0, 0, 0}); OperandType type3(Type::TENSOR_FLOAT32, {1, 2}); OperandType type4(Type::TENSOR_FLOAT32, {1, 8}); OperandType type5(Type::TENSOR_FLOAT32, {0}); OperandType type6(Type::TENSOR_INT32, {0}); OperandType type7(Type::TENSOR_FLOAT32, {0, 4}); OperandType type8(Type::TENSOR_INT32, {1}); OperandType type9(Type::BOOL, {}); // Phase 1, operands auto scores = model->addOperand(&type3); auto roi = model->addOperand(&type4); auto param4 = model->addOperand(&type8); auto param5 = model->addOperand(&type2); auto param6 = model->addOperand(&type1); auto param7 = model->addOperand(&type1); auto param8 = model->addOperand(&type2); auto param9 = model->addOperand(&type2); auto param10 = model->addOperand(&type2); auto scoresOut = model->addOperand(&type5); auto roiOut = model->addOperand(&type7); auto classesOut = model->addOperand(&type6); auto batchSplitOut = model->addOperand(&type6); auto in = model->addOperand(&type10); auto param11 = model->addOperand(&type1); auto param12 = model->addOperand(&type1); auto param13 = model->addOperand(&type2); auto param14 = model->addOperand(&type2); auto param15 = model->addOperand(&type1); auto param16 = model->addOperand(&type1); auto layout = model->addOperand(&type9); auto featureMap = model->addOperand(&type11); auto param17 = model->addOperand(&type2); auto out = model->addOperand(&type25); // Phase 2, operations static float scores_init[] = {0.9f, 0.1f}; model->setOperandValue(scores, scores_init, sizeof(float) * 2); static float roi_init[] = {1.0f, 1.0f, 10.0f, 10.0f, 0.0f, 0.0f, 10.0f, 10.0f}; model->setOperandValue(roi, roi_init, sizeof(float) * 8); static int32_t param4_init[] = {0}; model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1); static float param5_init[] = {0.3f}; model->setOperandValue(param5, param5_init, sizeof(float) * 1); static int32_t param6_init[] = {-1}; model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1); static int32_t param7_init[] = {0}; model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1); static float param8_init[] = {0.4f}; model->setOperandValue(param8, param8_init, sizeof(float) * 1); static float param9_init[] = {1.0f}; model->setOperandValue(param9, param9_init, sizeof(float) * 1); static float param10_init[] = {0.3f}; model->setOperandValue(param10, param10_init, sizeof(float) * 1); static int32_t param11_init[] = {2}; model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1); static int32_t param12_init[] = {2}; model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1); static float param13_init[] = {2.0f}; model->setOperandValue(param13, param13_init, sizeof(float) * 1); static float param14_init[] = {2.0f}; model->setOperandValue(param14, param14_init, sizeof(float) * 1); static int32_t param15_init[] = {4}; model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1); static int32_t param16_init[] = {4}; model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1); static bool8 layout_init[] = {false}; model->setOperandValue(layout, layout_init, sizeof(bool8) * 1); static float param17_init[] = {1.0f}; model->setOperandValue(param17, param17_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_BOX_WITH_NMS_LIMIT, {scores, roi, param4, param5, param6, param7, param8, param9, param10}, {scoresOut, roiOut, classesOut, batchSplitOut}); model->addOperation(ANEURALNETWORKS_ROI_ALIGN, {in, roiOut, batchSplitOut, param11, param12, param13, param14, param15, param16, layout}, {featureMap}); model->addOperation(ANEURALNETWORKS_SOFTMAX, {featureMap, param17}, {out}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {in}, {scoresOut, classesOut, out}); // Phase 4: set relaxed execution model->relaxComputationFloat32toFloat16(true); assert(model->isValid()); } inline bool is_ignored_zero_sized_dynamic_output_shape_relaxed(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_zero_sized_dynamic_output_shape_quant8(Model *model) { OperandType type1(Type::INT32, {}); OperandType type2(Type::FLOAT32, {}); OperandType type30(Type::TENSOR_QUANT8_ASYMM, {0, 0, 0, 0}, 0.00390625f, 0); OperandType type6(Type::TENSOR_INT32, {0}); OperandType type64(Type::TENSOR_QUANT8_ASYMM, {0, 2, 2, 1}, 0.1f, 128); OperandType type65(Type::TENSOR_QUANT8_ASYMM, {1, 1, 1, 1}, 0.1f, 128); OperandType type67(Type::TENSOR_QUANT16_ASYMM, {1, 8}, 0.125f, 0); OperandType type68(Type::TENSOR_QUANT16_ASYMM, {0, 4}, 0.125f, 0); OperandType type69(Type::TENSOR_QUANT8_ASYMM, {1, 2}, 0.1f, 128); OperandType type70(Type::TENSOR_QUANT8_ASYMM, {0}, 0.1f, 128); OperandType type8(Type::TENSOR_INT32, {1}); OperandType type9(Type::BOOL, {}); // Phase 1, operands auto scores = model->addOperand(&type69); auto roi = model->addOperand(&type67); auto param4 = model->addOperand(&type8); auto param5 = model->addOperand(&type2); auto param6 = model->addOperand(&type1); auto param7 = model->addOperand(&type1); auto param8 = model->addOperand(&type2); auto param9 = model->addOperand(&type2); auto param10 = model->addOperand(&type2); auto scoresOut = model->addOperand(&type70); auto roiOut = model->addOperand(&type68); auto classesOut = model->addOperand(&type6); auto batchSplitOut = model->addOperand(&type6); auto in = model->addOperand(&type65); auto param11 = model->addOperand(&type1); auto param12 = model->addOperand(&type1); auto param13 = model->addOperand(&type2); auto param14 = model->addOperand(&type2); auto param15 = model->addOperand(&type1); auto param16 = model->addOperand(&type1); auto layout = model->addOperand(&type9); auto featureMap = model->addOperand(&type64); auto param17 = model->addOperand(&type2); auto out = model->addOperand(&type30); // Phase 2, operations static uint8_t scores_init[] = {137, 129}; model->setOperandValue(scores, scores_init, sizeof(uint8_t) * 2); static uint16_t roi_init[] = {8, 8, 80, 80, 0, 0, 80, 80}; model->setOperandValue(roi, roi_init, sizeof(uint16_t) * 8); static int32_t param4_init[] = {0}; model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1); static float param5_init[] = {0.3f}; model->setOperandValue(param5, param5_init, sizeof(float) * 1); static int32_t param6_init[] = {-1}; model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1); static int32_t param7_init[] = {0}; model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1); static float param8_init[] = {0.4f}; model->setOperandValue(param8, param8_init, sizeof(float) * 1); static float param9_init[] = {1.0f}; model->setOperandValue(param9, param9_init, sizeof(float) * 1); static float param10_init[] = {0.3f}; model->setOperandValue(param10, param10_init, sizeof(float) * 1); static int32_t param11_init[] = {2}; model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1); static int32_t param12_init[] = {2}; model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1); static float param13_init[] = {2.0f}; model->setOperandValue(param13, param13_init, sizeof(float) * 1); static float param14_init[] = {2.0f}; model->setOperandValue(param14, param14_init, sizeof(float) * 1); static int32_t param15_init[] = {4}; model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1); static int32_t param16_init[] = {4}; model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1); static bool8 layout_init[] = {false}; model->setOperandValue(layout, layout_init, sizeof(bool8) * 1); static float param17_init[] = {1.0f}; model->setOperandValue(param17, param17_init, sizeof(float) * 1); model->addOperation(ANEURALNETWORKS_BOX_WITH_NMS_LIMIT, {scores, roi, param4, param5, param6, param7, param8, param9, param10}, {scoresOut, roiOut, classesOut, batchSplitOut}); model->addOperation(ANEURALNETWORKS_ROI_ALIGN, {in, roiOut, batchSplitOut, param11, param12, param13, param14, param15, param16, layout}, {featureMap}); model->addOperation(ANEURALNETWORKS_SOFTMAX, {featureMap, param17}, {out}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {in}, {scoresOut, classesOut, out}); assert(model->isValid()); } inline bool is_ignored_zero_sized_dynamic_output_shape_quant8(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); } void CreateModel_zero_sized_dynamic_output_shape_float16(Model *model) { OperandType type1(Type::INT32, {}); OperandType type15(Type::FLOAT16, {}); OperandType type27(Type::TENSOR_FLOAT16, {0, 0, 0, 0}); OperandType type28(Type::TENSOR_FLOAT16, {0}); OperandType type6(Type::TENSOR_INT32, {0}); OperandType type71(Type::TENSOR_FLOAT16, {0, 2, 2, 1}); OperandType type72(Type::TENSOR_FLOAT16, {1, 1, 1, 1}); OperandType type73(Type::TENSOR_FLOAT16, {1, 8}); OperandType type74(Type::TENSOR_FLOAT16, {0, 4}); OperandType type75(Type::TENSOR_FLOAT16, {1, 2}); OperandType type8(Type::TENSOR_INT32, {1}); OperandType type9(Type::BOOL, {}); // Phase 1, operands auto scores = model->addOperand(&type75); auto roi = model->addOperand(&type73); auto param4 = model->addOperand(&type8); auto param5 = model->addOperand(&type15); auto param6 = model->addOperand(&type1); auto param7 = model->addOperand(&type1); auto param8 = model->addOperand(&type15); auto param9 = model->addOperand(&type15); auto param10 = model->addOperand(&type15); auto scoresOut = model->addOperand(&type28); auto roiOut = model->addOperand(&type74); auto classesOut = model->addOperand(&type6); auto batchSplitOut = model->addOperand(&type6); auto in = model->addOperand(&type72); auto param11 = model->addOperand(&type1); auto param12 = model->addOperand(&type1); auto param13 = model->addOperand(&type15); auto param14 = model->addOperand(&type15); auto param15 = model->addOperand(&type1); auto param16 = model->addOperand(&type1); auto layout = model->addOperand(&type9); auto featureMap = model->addOperand(&type71); auto param17 = model->addOperand(&type15); auto out = model->addOperand(&type27); // Phase 2, operations static _Float16 scores_init[] = {0.8999999761581421f, 0.10000000149011612f}; model->setOperandValue(scores, scores_init, sizeof(_Float16) * 2); static _Float16 roi_init[] = {1.0f, 1.0f, 10.0f, 10.0f, 0.0f, 0.0f, 10.0f, 10.0f}; model->setOperandValue(roi, roi_init, sizeof(_Float16) * 8); static int32_t param4_init[] = {0}; model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1); static _Float16 param5_init[] = {0.30000001192092896f}; model->setOperandValue(param5, param5_init, sizeof(_Float16) * 1); static int32_t param6_init[] = {-1}; model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1); static int32_t param7_init[] = {0}; model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1); static _Float16 param8_init[] = {0.4000000059604645f}; model->setOperandValue(param8, param8_init, sizeof(_Float16) * 1); static _Float16 param9_init[] = {1.0f}; model->setOperandValue(param9, param9_init, sizeof(_Float16) * 1); static _Float16 param10_init[] = {0.30000001192092896f}; model->setOperandValue(param10, param10_init, sizeof(_Float16) * 1); static int32_t param11_init[] = {2}; model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1); static int32_t param12_init[] = {2}; model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1); static _Float16 param13_init[] = {2.0f}; model->setOperandValue(param13, param13_init, sizeof(_Float16) * 1); static _Float16 param14_init[] = {2.0f}; model->setOperandValue(param14, param14_init, sizeof(_Float16) * 1); static int32_t param15_init[] = {4}; model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1); static int32_t param16_init[] = {4}; model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1); static bool8 layout_init[] = {false}; model->setOperandValue(layout, layout_init, sizeof(bool8) * 1); static _Float16 param17_init[] = {1.0f}; model->setOperandValue(param17, param17_init, sizeof(_Float16) * 1); model->addOperation(ANEURALNETWORKS_BOX_WITH_NMS_LIMIT, {scores, roi, param4, param5, param6, param7, param8, param9, param10}, {scoresOut, roiOut, classesOut, batchSplitOut}); model->addOperation(ANEURALNETWORKS_ROI_ALIGN, {in, roiOut, batchSplitOut, param11, param12, param13, param14, param15, param16, layout}, {featureMap}); model->addOperation(ANEURALNETWORKS_SOFTMAX, {featureMap, param17}, {out}); // Phase 3, inputs and outputs model->identifyInputsAndOutputs( {in}, {scoresOut, classesOut, out}); assert(model->isValid()); } inline bool is_ignored_zero_sized_dynamic_output_shape_float16(int i) { static std::set ignore = {}; return ignore.find(i) != ignore.end(); }