Google Git
Sign in
android / platform / packages / modules / NeuralNetworks / c49850065cf80a64f1a99307fdd68f1301fe2582 / . / runtime / test / generated / spec_V1_2 / cts / bidirectional_sequence_lstm_float16_batch_major_aux_input.model.cpp
blob: 028344a33b5ce3f8f019df87cc4e9cc9961fa8d7 [file] [log] [blame]
// Generated from bidirectional_sequence_lstm_float16_batch_major_aux_input.mod.py
// DO NOT EDIT
// clang-format off
#include "TestGenerated.h"
namespace generated_tests::bidirectional_sequence_lstm_float16_batch_major_aux_input {
void CreateModel(Model *model) {
OperandType type0(Type::TENSOR_FLOAT16, {1, 3, 2});
OperandType type1(Type::TENSOR_FLOAT16, {4, 2});
OperandType type2(Type::TENSOR_FLOAT16, {4, 4});
OperandType type3(Type::TENSOR_FLOAT16, {4});
OperandType type4(Type::TENSOR_FLOAT16, {1, 4});
OperandType type5(Type::TENSOR_FLOAT16, {1, 3, 4});
OperandType type6(Type::INT32, {});
OperandType type7(Type::FLOAT16, {});
OperandType type8(Type::BOOL, {});
// Phase 1, operands
auto input = model->addOperand(&type0);
auto fw_input_to_input_weights = model->addOperand(&type1);
auto fw_input_to_forget_weights = model->addOperand(&type1);
auto fw_input_to_cell_weights = model->addOperand(&type1);
auto fw_input_to_output_weights = model->addOperand(&type1);
auto fw_recurrent_to_input_weights = model->addOperand(&type2);
auto fw_recurrent_to_forget_weights = model->addOperand(&type2);
auto fw_recurrent_to_cell_weights = model->addOperand(&type2);
auto fw_recurrent_to_output_weights = model->addOperand(&type2);
auto fw_cell_to_input_weights = model->addOperand(&type3);
auto fw_cell_to_forget_weights = model->addOperand(&type3);
auto fw_cell_to_output_weights = model->addOperand(&type3);
auto fw_input_gate_bias = model->addOperand(&type3);
auto fw_forget_gate_bias = model->addOperand(&type3);
auto fw_cell_bias = model->addOperand(&type3);
auto fw_output_gate_bias = model->addOperand(&type3);
auto fw_projection_weights = model->addOperand(&type2);
auto fw_projection_bias = model->addOperand(&type3);
auto bw_input_to_input_weights = model->addOperand(&type1);
auto bw_input_to_forget_weights = model->addOperand(&type1);
auto bw_input_to_cell_weights = model->addOperand(&type1);
auto bw_input_to_output_weights = model->addOperand(&type1);
auto bw_recurrent_to_input_weights = model->addOperand(&type2);
auto bw_recurrent_to_forget_weights = model->addOperand(&type2);
auto bw_recurrent_to_cell_weights = model->addOperand(&type2);
auto bw_recurrent_to_output_weights = model->addOperand(&type2);
auto bw_cell_to_input_weights = model->addOperand(&type3);
auto bw_cell_to_forget_weights = model->addOperand(&type3);
auto bw_cell_to_output_weights = model->addOperand(&type3);
auto bw_input_gate_bias = model->addOperand(&type3);
auto bw_forget_gate_bias = model->addOperand(&type3);
auto bw_cell_bias = model->addOperand(&type3);
auto bw_output_gate_bias = model->addOperand(&type3);
auto bw_projection_weights = model->addOperand(&type2);
auto bw_projection_bias = model->addOperand(&type3);
auto fw_activatiom_state = model->addOperand(&type4);
auto fw_cell_state = model->addOperand(&type4);
auto bw_activatiom_state = model->addOperand(&type4);
auto bw_cell_state = model->addOperand(&type4);
auto input1 = model->addOperand(&type0);
auto fw_aux_input_to_input_weights = model->addOperand(&type1);
auto fw_input_to_forget_weights1 = model->addOperand(&type1);
auto fw_aux_input_to_cell_weights = model->addOperand(&type1);
auto fw_aux_input_to_output_weights = model->addOperand(&type1);
auto bw_aux_input_to_input_weights = model->addOperand(&type1);
auto bw_input_to_forget_weights1 = model->addOperand(&type1);
auto bw_aux_input_to_cell_weights = model->addOperand(&type1);
auto bw_aux_input_to_output_weights = model->addOperand(&type1);
auto activation = model->addOperand(&type6);
auto cell_clip = model->addOperand(&type7);
auto proj_clip = model->addOperand(&type7);
auto merge_outputs = model->addOperand(&type8);
auto time_major = model->addOperand(&type8);
auto input_layer_norm_weights = model->addOperand(&type3);
auto forget_layer_norm_weights = model->addOperand(&type3);
auto cell_layer_norm_weights = model->addOperand(&type3);
auto output_layer_norm_weights = model->addOperand(&type3);
auto input_layer_norm_weights1 = model->addOperand(&type3);
auto forget_layer_norm_weights1 = model->addOperand(&type3);
auto cell_layer_norm_weights1 = model->addOperand(&type3);
auto output_layer_norm_weights1 = model->addOperand(&type3);
auto fw_output = model->addOperand(&type5);
auto bw_output = model->addOperand(&type5);
// Phase 2, operations
static int32_t activation_init[] = {4};
model->setOperandValue(activation, activation_init, sizeof(int32_t) * 1);
static _Float16 cell_clip_init[] = {0.0f};
model->setOperandValue(cell_clip, cell_clip_init, sizeof(_Float16) * 1);
static _Float16 proj_clip_init[] = {0.0f};
model->setOperandValue(proj_clip, proj_clip_init, sizeof(_Float16) * 1);
static bool8 merge_outputs_init[] = {false};
model->setOperandValue(merge_outputs, merge_outputs_init, sizeof(bool8) * 1);
static bool8 time_major_init[] = {false};
model->setOperandValue(time_major, time_major_init, sizeof(bool8) * 1);
model->addOperation(ANEURALNETWORKS_BIDIRECTIONAL_SEQUENCE_LSTM, {input, fw_input_to_input_weights, fw_input_to_forget_weights, fw_input_to_cell_weights, fw_input_to_output_weights, fw_recurrent_to_input_weights, fw_recurrent_to_forget_weights, fw_recurrent_to_cell_weights, fw_recurrent_to_output_weights, fw_cell_to_input_weights, fw_cell_to_forget_weights, fw_cell_to_output_weights, fw_input_gate_bias, fw_forget_gate_bias, fw_cell_bias, fw_output_gate_bias, fw_projection_weights, fw_projection_bias, bw_input_to_input_weights, bw_input_to_forget_weights, bw_input_to_cell_weights, bw_input_to_output_weights, bw_recurrent_to_input_weights, bw_recurrent_to_forget_weights, bw_recurrent_to_cell_weights, bw_recurrent_to_output_weights, bw_cell_to_input_weights, bw_cell_to_forget_weights, bw_cell_to_output_weights, bw_input_gate_bias, bw_forget_gate_bias, bw_cell_bias, bw_output_gate_bias, bw_projection_weights, bw_projection_bias, fw_activatiom_state, fw_cell_state, bw_activatiom_state, bw_cell_state, input1, fw_aux_input_to_input_weights, fw_input_to_forget_weights1, fw_aux_input_to_cell_weights, fw_aux_input_to_output_weights, bw_aux_input_to_input_weights, bw_input_to_forget_weights1, bw_aux_input_to_cell_weights, bw_aux_input_to_output_weights, activation, cell_clip, proj_clip, merge_outputs, time_major, input_layer_norm_weights, forget_layer_norm_weights, cell_layer_norm_weights, output_layer_norm_weights, input_layer_norm_weights1, forget_layer_norm_weights1, cell_layer_norm_weights1, output_layer_norm_weights1}, {fw_output, bw_output});
// Phase 3, inputs and outputs
model->identifyInputsAndOutputs(
{input, fw_input_to_input_weights, fw_input_to_forget_weights, fw_input_to_cell_weights, fw_input_to_output_weights, fw_recurrent_to_input_weights, fw_recurrent_to_forget_weights, fw_recurrent_to_cell_weights, fw_recurrent_to_output_weights, fw_cell_to_input_weights, fw_cell_to_forget_weights, fw_cell_to_output_weights, fw_input_gate_bias, fw_forget_gate_bias, fw_cell_bias, fw_output_gate_bias, fw_projection_weights, fw_projection_bias, bw_input_to_input_weights, bw_input_to_forget_weights, bw_input_to_cell_weights, bw_input_to_output_weights, bw_recurrent_to_input_weights, bw_recurrent_to_forget_weights, bw_recurrent_to_cell_weights, bw_recurrent_to_output_weights, bw_cell_to_input_weights, bw_cell_to_forget_weights, bw_cell_to_output_weights, bw_input_gate_bias, bw_forget_gate_bias, bw_cell_bias, bw_output_gate_bias, bw_projection_weights, bw_projection_bias, fw_activatiom_state, fw_cell_state, bw_activatiom_state, bw_cell_state, input1, fw_aux_input_to_input_weights, fw_input_to_forget_weights1, fw_aux_input_to_cell_weights, fw_aux_input_to_output_weights, bw_aux_input_to_input_weights, bw_input_to_forget_weights1, bw_aux_input_to_cell_weights, bw_aux_input_to_output_weights, input_layer_norm_weights, forget_layer_norm_weights, cell_layer_norm_weights, output_layer_norm_weights, input_layer_norm_weights1, forget_layer_norm_weights1, cell_layer_norm_weights1, output_layer_norm_weights1},
{fw_output, bw_output});
assert(model->isValid());
}
bool is_ignored(int i) {
static std::set<int> ignore = {};
return ignore.find(i) != ignore.end();
}
} // namespace generated_tests::bidirectional_sequence_lstm_float16_batch_major_aux_input
namespace generated_tests::bidirectional_sequence_lstm_float16_batch_major_aux_input {
void CreateModel_dynamic_output_shape(Model *model) {
OperandType type0(Type::TENSOR_FLOAT16, {1, 3, 2});
OperandType type1(Type::TENSOR_FLOAT16, {4, 2});
OperandType type2(Type::TENSOR_FLOAT16, {4, 4});
OperandType type3(Type::TENSOR_FLOAT16, {4});
OperandType type4(Type::TENSOR_FLOAT16, {1, 4});
OperandType type6(Type::INT32, {});
OperandType type7(Type::FLOAT16, {});
OperandType type8(Type::BOOL, {});
OperandType type9(Type::TENSOR_FLOAT16, {0, 0, 0});
// Phase 1, operands
auto input = model->addOperand(&type0);
auto fw_input_to_input_weights = model->addOperand(&type1);
auto fw_input_to_forget_weights = model->addOperand(&type1);
auto fw_input_to_cell_weights = model->addOperand(&type1);
auto fw_input_to_output_weights = model->addOperand(&type1);
auto fw_recurrent_to_input_weights = model->addOperand(&type2);
auto fw_recurrent_to_forget_weights = model->addOperand(&type2);
auto fw_recurrent_to_cell_weights = model->addOperand(&type2);
auto fw_recurrent_to_output_weights = model->addOperand(&type2);
auto fw_cell_to_input_weights = model->addOperand(&type3);
auto fw_cell_to_forget_weights = model->addOperand(&type3);
auto fw_cell_to_output_weights = model->addOperand(&type3);
auto fw_input_gate_bias = model->addOperand(&type3);
auto fw_forget_gate_bias = model->addOperand(&type3);
auto fw_cell_bias = model->addOperand(&type3);
auto fw_output_gate_bias = model->addOperand(&type3);
auto fw_projection_weights = model->addOperand(&type2);
auto fw_projection_bias = model->addOperand(&type3);
auto bw_input_to_input_weights = model->addOperand(&type1);
auto bw_input_to_forget_weights = model->addOperand(&type1);
auto bw_input_to_cell_weights = model->addOperand(&type1);
auto bw_input_to_output_weights = model->addOperand(&type1);
auto bw_recurrent_to_input_weights = model->addOperand(&type2);
auto bw_recurrent_to_forget_weights = model->addOperand(&type2);
auto bw_recurrent_to_cell_weights = model->addOperand(&type2);
auto bw_recurrent_to_output_weights = model->addOperand(&type2);
auto bw_cell_to_input_weights = model->addOperand(&type3);
auto bw_cell_to_forget_weights = model->addOperand(&type3);
auto bw_cell_to_output_weights = model->addOperand(&type3);
auto bw_input_gate_bias = model->addOperand(&type3);
auto bw_forget_gate_bias = model->addOperand(&type3);
auto bw_cell_bias = model->addOperand(&type3);
auto bw_output_gate_bias = model->addOperand(&type3);
auto bw_projection_weights = model->addOperand(&type2);
auto bw_projection_bias = model->addOperand(&type3);
auto fw_activatiom_state = model->addOperand(&type4);
auto fw_cell_state = model->addOperand(&type4);
auto bw_activatiom_state = model->addOperand(&type4);
auto bw_cell_state = model->addOperand(&type4);
auto input1 = model->addOperand(&type0);
auto fw_aux_input_to_input_weights = model->addOperand(&type1);
auto fw_input_to_forget_weights1 = model->addOperand(&type1);
auto fw_aux_input_to_cell_weights = model->addOperand(&type1);
auto fw_aux_input_to_output_weights = model->addOperand(&type1);
auto bw_aux_input_to_input_weights = model->addOperand(&type1);
auto bw_input_to_forget_weights1 = model->addOperand(&type1);
auto bw_aux_input_to_cell_weights = model->addOperand(&type1);
auto bw_aux_input_to_output_weights = model->addOperand(&type1);
auto activation = model->addOperand(&type6);
auto cell_clip = model->addOperand(&type7);
auto proj_clip = model->addOperand(&type7);
auto merge_outputs = model->addOperand(&type8);
auto time_major = model->addOperand(&type8);
auto input_layer_norm_weights = model->addOperand(&type3);
auto forget_layer_norm_weights = model->addOperand(&type3);
auto cell_layer_norm_weights = model->addOperand(&type3);
auto output_layer_norm_weights = model->addOperand(&type3);
auto input_layer_norm_weights1 = model->addOperand(&type3);
auto forget_layer_norm_weights1 = model->addOperand(&type3);
auto cell_layer_norm_weights1 = model->addOperand(&type3);
auto output_layer_norm_weights1 = model->addOperand(&type3);
auto fw_output = model->addOperand(&type9);
auto bw_output = model->addOperand(&type9);
// Phase 2, operations
static int32_t activation_init[] = {4};
model->setOperandValue(activation, activation_init, sizeof(int32_t) * 1);
static _Float16 cell_clip_init[] = {0.0f};
model->setOperandValue(cell_clip, cell_clip_init, sizeof(_Float16) * 1);
static _Float16 proj_clip_init[] = {0.0f};
model->setOperandValue(proj_clip, proj_clip_init, sizeof(_Float16) * 1);
static bool8 merge_outputs_init[] = {false};
model->setOperandValue(merge_outputs, merge_outputs_init, sizeof(bool8) * 1);
static bool8 time_major_init[] = {false};
model->setOperandValue(time_major, time_major_init, sizeof(bool8) * 1);
model->addOperation(ANEURALNETWORKS_BIDIRECTIONAL_SEQUENCE_LSTM, {input, fw_input_to_input_weights, fw_input_to_forget_weights, fw_input_to_cell_weights, fw_input_to_output_weights, fw_recurrent_to_input_weights, fw_recurrent_to_forget_weights, fw_recurrent_to_cell_weights, fw_recurrent_to_output_weights, fw_cell_to_input_weights, fw_cell_to_forget_weights, fw_cell_to_output_weights, fw_input_gate_bias, fw_forget_gate_bias, fw_cell_bias, fw_output_gate_bias, fw_projection_weights, fw_projection_bias, bw_input_to_input_weights, bw_input_to_forget_weights, bw_input_to_cell_weights, bw_input_to_output_weights, bw_recurrent_to_input_weights, bw_recurrent_to_forget_weights, bw_recurrent_to_cell_weights, bw_recurrent_to_output_weights, bw_cell_to_input_weights, bw_cell_to_forget_weights, bw_cell_to_output_weights, bw_input_gate_bias, bw_forget_gate_bias, bw_cell_bias, bw_output_gate_bias, bw_projection_weights, bw_projection_bias, fw_activatiom_state, fw_cell_state, bw_activatiom_state, bw_cell_state, input1, fw_aux_input_to_input_weights, fw_input_to_forget_weights1, fw_aux_input_to_cell_weights, fw_aux_input_to_output_weights, bw_aux_input_to_input_weights, bw_input_to_forget_weights1, bw_aux_input_to_cell_weights, bw_aux_input_to_output_weights, activation, cell_clip, proj_clip, merge_outputs, time_major, input_layer_norm_weights, forget_layer_norm_weights, cell_layer_norm_weights, output_layer_norm_weights, input_layer_norm_weights1, forget_layer_norm_weights1, cell_layer_norm_weights1, output_layer_norm_weights1}, {fw_output, bw_output});
// Phase 3, inputs and outputs
model->identifyInputsAndOutputs(
{input, fw_input_to_input_weights, fw_input_to_forget_weights, fw_input_to_cell_weights, fw_input_to_output_weights, fw_recurrent_to_input_weights, fw_recurrent_to_forget_weights, fw_recurrent_to_cell_weights, fw_recurrent_to_output_weights, fw_cell_to_input_weights, fw_cell_to_forget_weights, fw_cell_to_output_weights, fw_input_gate_bias, fw_forget_gate_bias, fw_cell_bias, fw_output_gate_bias, fw_projection_weights, fw_projection_bias, bw_input_to_input_weights, bw_input_to_forget_weights, bw_input_to_cell_weights, bw_input_to_output_weights, bw_recurrent_to_input_weights, bw_recurrent_to_forget_weights, bw_recurrent_to_cell_weights, bw_recurrent_to_output_weights, bw_cell_to_input_weights, bw_cell_to_forget_weights, bw_cell_to_output_weights, bw_input_gate_bias, bw_forget_gate_bias, bw_cell_bias, bw_output_gate_bias, bw_projection_weights, bw_projection_bias, fw_activatiom_state, fw_cell_state, bw_activatiom_state, bw_cell_state, input1, fw_aux_input_to_input_weights, fw_input_to_forget_weights1, fw_aux_input_to_cell_weights, fw_aux_input_to_output_weights, bw_aux_input_to_input_weights, bw_input_to_forget_weights1, bw_aux_input_to_cell_weights, bw_aux_input_to_output_weights, input_layer_norm_weights, forget_layer_norm_weights, cell_layer_norm_weights, output_layer_norm_weights, input_layer_norm_weights1, forget_layer_norm_weights1, cell_layer_norm_weights1, output_layer_norm_weights1},
{fw_output, bw_output});
assert(model->isValid());
}
bool is_ignored_dynamic_output_shape(int i) {
static std::set<int> ignore = {};
return ignore.find(i) != ignore.end();
}
} // namespace generated_tests::bidirectional_sequence_lstm_float16_batch_major_aux_input
namespace generated_tests::bidirectional_sequence_lstm_float16_batch_major_aux_input {
void CreateModel_all_inputs_as_internal(Model *model) {
OperandType type0(Type::TENSOR_FLOAT16, {1, 3, 2});
OperandType type1(Type::TENSOR_FLOAT16, {4, 2});
OperandType type10(Type::TENSOR_FLOAT16, {1});
OperandType type2(Type::TENSOR_FLOAT16, {4, 4});
OperandType type3(Type::TENSOR_FLOAT16, {4});
OperandType type4(Type::TENSOR_FLOAT16, {1, 4});
OperandType type5(Type::TENSOR_FLOAT16, {1, 3, 4});
OperandType type6(Type::INT32, {});
OperandType type7(Type::FLOAT16, {});
OperandType type8(Type::BOOL, {});
// Phase 1, operands
auto input = model->addOperand(&type0);
auto fw_input_to_input_weights = model->addOperand(&type1);
auto fw_input_to_forget_weights = model->addOperand(&type1);
auto fw_input_to_cell_weights = model->addOperand(&type1);
auto fw_input_to_output_weights = model->addOperand(&type1);
auto fw_recurrent_to_input_weights = model->addOperand(&type2);
auto fw_recurrent_to_forget_weights = model->addOperand(&type2);
auto fw_recurrent_to_cell_weights = model->addOperand(&type2);
auto fw_recurrent_to_output_weights = model->addOperand(&type2);
auto fw_cell_to_input_weights = model->addOperand(&type3);
auto fw_cell_to_forget_weights = model->addOperand(&type3);
auto fw_cell_to_output_weights = model->addOperand(&type3);
auto fw_input_gate_bias = model->addOperand(&type3);
auto fw_forget_gate_bias = model->addOperand(&type3);
auto fw_cell_bias = model->addOperand(&type3);
auto fw_output_gate_bias = model->addOperand(&type3);
auto fw_projection_weights = model->addOperand(&type2);
auto fw_projection_bias = model->addOperand(&type3);
auto bw_input_to_input_weights = model->addOperand(&type1);
auto bw_input_to_forget_weights = model->addOperand(&type1);
auto bw_input_to_cell_weights = model->addOperand(&type1);
auto bw_input_to_output_weights = model->addOperand(&type1);
auto bw_recurrent_to_input_weights = model->addOperand(&type2);
auto bw_recurrent_to_forget_weights = model->addOperand(&type2);
auto bw_recurrent_to_cell_weights = model->addOperand(&type2);
auto bw_recurrent_to_output_weights = model->addOperand(&type2);
auto bw_cell_to_input_weights = model->addOperand(&type3);
auto bw_cell_to_forget_weights = model->addOperand(&type3);
auto bw_cell_to_output_weights = model->addOperand(&type3);
auto bw_input_gate_bias = model->addOperand(&type3);
auto bw_forget_gate_bias = model->addOperand(&type3);
auto bw_cell_bias = model->addOperand(&type3);
auto bw_output_gate_bias = model->addOperand(&type3);
auto bw_projection_weights = model->addOperand(&type2);
auto bw_projection_bias = model->addOperand(&type3);
auto fw_activatiom_state = model->addOperand(&type4);
auto fw_cell_state = model->addOperand(&type4);
auto bw_activatiom_state = model->addOperand(&type4);
auto bw_cell_state = model->addOperand(&type4);
auto input1 = model->addOperand(&type0);
auto fw_aux_input_to_input_weights = model->addOperand(&type1);
auto fw_input_to_forget_weights1 = model->addOperand(&type1);
auto fw_aux_input_to_cell_weights = model->addOperand(&type1);
auto fw_aux_input_to_output_weights = model->addOperand(&type1);
auto bw_aux_input_to_input_weights = model->addOperand(&type1);
auto bw_input_to_forget_weights1 = model->addOperand(&type1);
auto bw_aux_input_to_cell_weights = model->addOperand(&type1);
auto bw_aux_input_to_output_weights = model->addOperand(&type1);
auto activation = model->addOperand(&type6);
auto cell_clip = model->addOperand(&type7);
auto proj_clip = model->addOperand(&type7);
auto merge_outputs = model->addOperand(&type8);
auto time_major = model->addOperand(&type8);
auto input_layer_norm_weights = model->addOperand(&type3);
auto forget_layer_norm_weights = model->addOperand(&type3);
auto cell_layer_norm_weights = model->addOperand(&type3);
auto output_layer_norm_weights = model->addOperand(&type3);
auto input_layer_norm_weights1 = model->addOperand(&type3);
auto forget_layer_norm_weights1 = model->addOperand(&type3);
auto cell_layer_norm_weights1 = model->addOperand(&type3);
auto output_layer_norm_weights1 = model->addOperand(&type3);
auto fw_output = model->addOperand(&type5);
auto bw_output = model->addOperand(&type5);
auto input_tmp = model->addOperand(&type0);
auto dummy = model->addOperand(&type10);
auto param = model->addOperand(&type6);
auto fw_input_to_input_weights_tmp = model->addOperand(&type1);
auto dummy1 = model->addOperand(&type10);
auto param1 = model->addOperand(&type6);
auto fw_input_to_forget_weights_tmp = model->addOperand(&type1);
auto dummy2 = model->addOperand(&type10);
auto param2 = model->addOperand(&type6);
auto fw_input_to_cell_weights_tmp = model->addOperand(&type1);
auto dummy3 = model->addOperand(&type10);
auto param3 = model->addOperand(&type6);
auto fw_input_to_output_weights_tmp = model->addOperand(&type1);
auto dummy4 = model->addOperand(&type10);
auto param4 = model->addOperand(&type6);
auto fw_recurrent_to_input_weights_tmp = model->addOperand(&type2);
auto dummy5 = model->addOperand(&type10);
auto param5 = model->addOperand(&type6);
auto fw_recurrent_to_forget_weights_tmp = model->addOperand(&type2);
auto dummy6 = model->addOperand(&type10);
auto param6 = model->addOperand(&type6);
auto fw_recurrent_to_cell_weights_tmp = model->addOperand(&type2);
auto dummy7 = model->addOperand(&type10);
auto param7 = model->addOperand(&type6);
auto fw_recurrent_to_output_weights_tmp = model->addOperand(&type2);
auto dummy8 = model->addOperand(&type10);
auto param8 = model->addOperand(&type6);
auto fw_input_gate_bias_tmp = model->addOperand(&type3);
auto dummy9 = model->addOperand(&type10);
auto param9 = model->addOperand(&type6);
auto fw_forget_gate_bias_tmp = model->addOperand(&type3);
auto dummy10 = model->addOperand(&type10);
auto param10 = model->addOperand(&type6);
auto fw_cell_bias_tmp = model->addOperand(&type3);
auto dummy11 = model->addOperand(&type10);
auto param11 = model->addOperand(&type6);
auto fw_output_gate_bias_tmp = model->addOperand(&type3);
auto dummy12 = model->addOperand(&type10);
auto param12 = model->addOperand(&type6);
auto bw_input_to_input_weights_tmp = model->addOperand(&type1);
auto dummy13 = model->addOperand(&type10);
auto param13 = model->addOperand(&type6);
auto bw_input_to_forget_weights_tmp = model->addOperand(&type1);
auto dummy14 = model->addOperand(&type10);
auto param14 = model->addOperand(&type6);
auto bw_input_to_cell_weights_tmp = model->addOperand(&type1);
auto dummy15 = model->addOperand(&type10);
auto param15 = model->addOperand(&type6);
auto bw_input_to_output_weights_tmp = model->addOperand(&type1);
auto dummy16 = model->addOperand(&type10);
auto param16 = model->addOperand(&type6);
auto bw_recurrent_to_input_weights_tmp = model->addOperand(&type2);
auto dummy17 = model->addOperand(&type10);
auto param17 = model->addOperand(&type6);
auto bw_recurrent_to_forget_weights_tmp = model->addOperand(&type2);
auto dummy18 = model->addOperand(&type10);
auto param18 = model->addOperand(&type6);
auto bw_recurrent_to_cell_weights_tmp = model->addOperand(&type2);
auto dummy19 = model->addOperand(&type10);
auto param19 = model->addOperand(&type6);
auto bw_recurrent_to_output_weights_tmp = model->addOperand(&type2);
auto dummy20 = model->addOperand(&type10);
auto param20 = model->addOperand(&type6);
auto bw_input_gate_bias_tmp = model->addOperand(&type3);
auto dummy21 = model->addOperand(&type10);
auto param21 = model->addOperand(&type6);
auto bw_forget_gate_bias_tmp = model->addOperand(&type3);
auto dummy22 = model->addOperand(&type10);
auto param22 = model->addOperand(&type6);
auto bw_cell_bias_tmp = model->addOperand(&type3);
auto dummy23 = model->addOperand(&type10);
auto param23 = model->addOperand(&type6);
auto bw_output_gate_bias_tmp = model->addOperand(&type3);
auto dummy24 = model->addOperand(&type10);
auto param24 = model->addOperand(&type6);
auto fw_activatiom_state_tmp = model->addOperand(&type4);
auto dummy25 = model->addOperand(&type10);
auto param25 = model->addOperand(&type6);
auto fw_cell_state_tmp = model->addOperand(&type4);
auto dummy26 = model->addOperand(&type10);
auto param26 = model->addOperand(&type6);
auto bw_activatiom_state_tmp = model->addOperand(&type4);
auto dummy27 = model->addOperand(&type10);
auto param27 = model->addOperand(&type6);
auto bw_cell_state_tmp = model->addOperand(&type4);
auto dummy28 = model->addOperand(&type10);
auto param28 = model->addOperand(&type6);
auto input1_tmp = model->addOperand(&type0);
auto dummy29 = model->addOperand(&type10);
auto param29 = model->addOperand(&type6);
auto fw_aux_input_to_input_weights_tmp = model->addOperand(&type1);
auto dummy30 = model->addOperand(&type10);
auto param30 = model->addOperand(&type6);
auto fw_input_to_forget_weights1_tmp = model->addOperand(&type1);
auto dummy31 = model->addOperand(&type10);
auto param31 = model->addOperand(&type6);
auto fw_aux_input_to_cell_weights_tmp = model->addOperand(&type1);
auto dummy32 = model->addOperand(&type10);
auto param32 = model->addOperand(&type6);
auto fw_aux_input_to_output_weights_tmp = model->addOperand(&type1);
auto dummy33 = model->addOperand(&type10);
auto param33 = model->addOperand(&type6);
auto bw_aux_input_to_input_weights_tmp = model->addOperand(&type1);
auto dummy34 = model->addOperand(&type10);
auto param34 = model->addOperand(&type6);
auto bw_input_to_forget_weights1_tmp = model->addOperand(&type1);
auto dummy35 = model->addOperand(&type10);
auto param35 = model->addOperand(&type6);
auto bw_aux_input_to_cell_weights_tmp = model->addOperand(&type1);
auto dummy36 = model->addOperand(&type10);
auto param36 = model->addOperand(&type6);
auto bw_aux_input_to_output_weights_tmp = model->addOperand(&type1);
auto dummy37 = model->addOperand(&type10);
auto param37 = model->addOperand(&type6);
// Phase 2, operations
static int32_t activation_init[] = {4};
model->setOperandValue(activation, activation_init, sizeof(int32_t) * 1);
static _Float16 cell_clip_init[] = {0.0f};
model->setOperandValue(cell_clip, cell_clip_init, sizeof(_Float16) * 1);
static _Float16 proj_clip_init[] = {0.0f};
model->setOperandValue(proj_clip, proj_clip_init, sizeof(_Float16) * 1);
static bool8 merge_outputs_init[] = {false};
model->setOperandValue(merge_outputs, merge_outputs_init, sizeof(bool8) * 1);
static bool8 time_major_init[] = {false};
model->setOperandValue(time_major, time_major_init, sizeof(bool8) * 1);
static _Float16 dummy_init[] = {0.0f};
model->setOperandValue(dummy, dummy_init, sizeof(_Float16) * 1);
static int32_t param_init[] = {0};
model->setOperandValue(param, param_init, sizeof(int32_t) * 1);
static _Float16 dummy1_init[] = {0.0f};
model->setOperandValue(dummy1, dummy1_init, sizeof(_Float16) * 1);
static int32_t param1_init[] = {0};
model->setOperandValue(param1, param1_init, sizeof(int32_t) * 1);
static _Float16 dummy2_init[] = {0.0f};
model->setOperandValue(dummy2, dummy2_init, sizeof(_Float16) * 1);
static int32_t param2_init[] = {0};
model->setOperandValue(param2, param2_init, sizeof(int32_t) * 1);
static _Float16 dummy3_init[] = {0.0f};
model->setOperandValue(dummy3, dummy3_init, sizeof(_Float16) * 1);
static int32_t param3_init[] = {0};
model->setOperandValue(param3, param3_init, sizeof(int32_t) * 1);
static _Float16 dummy4_init[] = {0.0f};
model->setOperandValue(dummy4, dummy4_init, sizeof(_Float16) * 1);
static int32_t param4_init[] = {0};
model->setOperandValue(param4, param4_init, sizeof(int32_t) * 1);
static _Float16 dummy5_init[] = {0.0f};
model->setOperandValue(dummy5, dummy5_init, sizeof(_Float16) * 1);
static int32_t param5_init[] = {0};
model->setOperandValue(param5, param5_init, sizeof(int32_t) * 1);
static _Float16 dummy6_init[] = {0.0f};
model->setOperandValue(dummy6, dummy6_init, sizeof(_Float16) * 1);
static int32_t param6_init[] = {0};
model->setOperandValue(param6, param6_init, sizeof(int32_t) * 1);
static _Float16 dummy7_init[] = {0.0f};
model->setOperandValue(dummy7, dummy7_init, sizeof(_Float16) * 1);
static int32_t param7_init[] = {0};
model->setOperandValue(param7, param7_init, sizeof(int32_t) * 1);
static _Float16 dummy8_init[] = {0.0f};
model->setOperandValue(dummy8, dummy8_init, sizeof(_Float16) * 1);
static int32_t param8_init[] = {0};
model->setOperandValue(param8, param8_init, sizeof(int32_t) * 1);
static _Float16 dummy9_init[] = {0.0f};
model->setOperandValue(dummy9, dummy9_init, sizeof(_Float16) * 1);
static int32_t param9_init[] = {0};
model->setOperandValue(param9, param9_init, sizeof(int32_t) * 1);
static _Float16 dummy10_init[] = {0.0f};
model->setOperandValue(dummy10, dummy10_init, sizeof(_Float16) * 1);
static int32_t param10_init[] = {0};
model->setOperandValue(param10, param10_init, sizeof(int32_t) * 1);
static _Float16 dummy11_init[] = {0.0f};
model->setOperandValue(dummy11, dummy11_init, sizeof(_Float16) * 1);
static int32_t param11_init[] = {0};
model->setOperandValue(param11, param11_init, sizeof(int32_t) * 1);
static _Float16 dummy12_init[] = {0.0f};
model->setOperandValue(dummy12, dummy12_init, sizeof(_Float16) * 1);
static int32_t param12_init[] = {0};
model->setOperandValue(param12, param12_init, sizeof(int32_t) * 1);
static _Float16 dummy13_init[] = {0.0f};
model->setOperandValue(dummy13, dummy13_init, sizeof(_Float16) * 1);
static int32_t param13_init[] = {0};
model->setOperandValue(param13, param13_init, sizeof(int32_t) * 1);
static _Float16 dummy14_init[] = {0.0f};
model->setOperandValue(dummy14, dummy14_init, sizeof(_Float16) * 1);
static int32_t param14_init[] = {0};
model->setOperandValue(param14, param14_init, sizeof(int32_t) * 1);
static _Float16 dummy15_init[] = {0.0f};
model->setOperandValue(dummy15, dummy15_init, sizeof(_Float16) * 1);
static int32_t param15_init[] = {0};
model->setOperandValue(param15, param15_init, sizeof(int32_t) * 1);
static _Float16 dummy16_init[] = {0.0f};
model->setOperandValue(dummy16, dummy16_init, sizeof(_Float16) * 1);
static int32_t param16_init[] = {0};
model->setOperandValue(param16, param16_init, sizeof(int32_t) * 1);
static _Float16 dummy17_init[] = {0.0f};
model->setOperandValue(dummy17, dummy17_init, sizeof(_Float16) * 1);
static int32_t param17_init[] = {0};
model->setOperandValue(param17, param17_init, sizeof(int32_t) * 1);
static _Float16 dummy18_init[] = {0.0f};
model->setOperandValue(dummy18, dummy18_init, sizeof(_Float16) * 1);
static int32_t param18_init[] = {0};
model->setOperandValue(param18, param18_init, sizeof(int32_t) * 1);
static _Float16 dummy19_init[] = {0.0f};
model->setOperandValue(dummy19, dummy19_init, sizeof(_Float16) * 1);
static int32_t param19_init[] = {0};
model->setOperandValue(param19, param19_init, sizeof(int32_t) * 1);
static _Float16 dummy20_init[] = {0.0f};
model->setOperandValue(dummy20, dummy20_init, sizeof(_Float16) * 1);
static int32_t param20_init[] = {0};
model->setOperandValue(param20, param20_init, sizeof(int32_t) * 1);
static _Float16 dummy21_init[] = {0.0f};
model->setOperandValue(dummy21, dummy21_init, sizeof(_Float16) * 1);
static int32_t param21_init[] = {0};
model->setOperandValue(param21, param21_init, sizeof(int32_t) * 1);
static _Float16 dummy22_init[] = {0.0f};
model->setOperandValue(dummy22, dummy22_init, sizeof(_Float16) * 1);
static int32_t param22_init[] = {0};
model->setOperandValue(param22, param22_init, sizeof(int32_t) * 1);
static _Float16 dummy23_init[] = {0.0f};
model->setOperandValue(dummy23, dummy23_init, sizeof(_Float16) * 1);
static int32_t param23_init[] = {0};
model->setOperandValue(param23, param23_init, sizeof(int32_t) * 1);
static _Float16 dummy24_init[] = {0.0f};
model->setOperandValue(dummy24, dummy24_init, sizeof(_Float16) * 1);
static int32_t param24_init[] = {0};
model->setOperandValue(param24, param24_init, sizeof(int32_t) * 1);
static _Float16 dummy25_init[] = {0.0f};
model->setOperandValue(dummy25, dummy25_init, sizeof(_Float16) * 1);
static int32_t param25_init[] = {0};
model->setOperandValue(param25, param25_init, sizeof(int32_t) * 1);
static _Float16 dummy26_init[] = {0.0f};
model->setOperandValue(dummy26, dummy26_init, sizeof(_Float16) * 1);
static int32_t param26_init[] = {0};
model->setOperandValue(param26, param26_init, sizeof(int32_t) * 1);
static _Float16 dummy27_init[] = {0.0f};
model->setOperandValue(dummy27, dummy27_init, sizeof(_Float16) * 1);
static int32_t param27_init[] = {0};
model->setOperandValue(param27, param27_init, sizeof(int32_t) * 1);
static _Float16 dummy28_init[] = {0.0f};
model->setOperandValue(dummy28, dummy28_init, sizeof(_Float16) * 1);
static int32_t param28_init[] = {0};
model->setOperandValue(param28, param28_init, sizeof(int32_t) * 1);
static _Float16 dummy29_init[] = {0.0f};
model->setOperandValue(dummy29, dummy29_init, sizeof(_Float16) * 1);
static int32_t param29_init[] = {0};
model->setOperandValue(param29, param29_init, sizeof(int32_t) * 1);
static _Float16 dummy30_init[] = {0.0f};
model->setOperandValue(dummy30, dummy30_init, sizeof(_Float16) * 1);
static int32_t param30_init[] = {0};
model->setOperandValue(param30, param30_init, sizeof(int32_t) * 1);
static _Float16 dummy31_init[] = {0.0f};
model->setOperandValue(dummy31, dummy31_init, sizeof(_Float16) * 1);
static int32_t param31_init[] = {0};
model->setOperandValue(param31, param31_init, sizeof(int32_t) * 1);
static _Float16 dummy32_init[] = {0.0f};
model->setOperandValue(dummy32, dummy32_init, sizeof(_Float16) * 1);
static int32_t param32_init[] = {0};
model->setOperandValue(param32, param32_init, sizeof(int32_t) * 1);
static _Float16 dummy33_init[] = {0.0f};
model->setOperandValue(dummy33, dummy33_init, sizeof(_Float16) * 1);
static int32_t param33_init[] = {0};
model->setOperandValue(param33, param33_init, sizeof(int32_t) * 1);
static _Float16 dummy34_init[] = {0.0f};
model->setOperandValue(dummy34, dummy34_init, sizeof(_Float16) * 1);
static int32_t param34_init[] = {0};
model->setOperandValue(param34, param34_init, sizeof(int32_t) * 1);
static _Float16 dummy35_init[] = {0.0f};
model->setOperandValue(dummy35, dummy35_init, sizeof(_Float16) * 1);
static int32_t param35_init[] = {0};
model->setOperandValue(param35, param35_init, sizeof(int32_t) * 1);
static _Float16 dummy36_init[] = {0.0f};
model->setOperandValue(dummy36, dummy36_init, sizeof(_Float16) * 1);
static int32_t param36_init[] = {0};
model->setOperandValue(param36, param36_init, sizeof(int32_t) * 1);
static _Float16 dummy37_init[] = {0.0f};
model->setOperandValue(dummy37, dummy37_init, sizeof(_Float16) * 1);
static int32_t param37_init[] = {0};
model->setOperandValue(param37, param37_init, sizeof(int32_t) * 1);
model->addOperation(ANEURALNETWORKS_ADD, {input_tmp, dummy, param}, {input});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_input_weights_tmp, dummy1, param1}, {fw_input_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_forget_weights_tmp, dummy2, param2}, {fw_input_to_forget_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_cell_weights_tmp, dummy3, param3}, {fw_input_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_output_weights_tmp, dummy4, param4}, {fw_input_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_recurrent_to_input_weights_tmp, dummy5, param5}, {fw_recurrent_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_recurrent_to_forget_weights_tmp, dummy6, param6}, {fw_recurrent_to_forget_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_recurrent_to_cell_weights_tmp, dummy7, param7}, {fw_recurrent_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_recurrent_to_output_weights_tmp, dummy8, param8}, {fw_recurrent_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_gate_bias_tmp, dummy9, param9}, {fw_input_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {fw_forget_gate_bias_tmp, dummy10, param10}, {fw_forget_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {fw_cell_bias_tmp, dummy11, param11}, {fw_cell_bias});
model->addOperation(ANEURALNETWORKS_ADD, {fw_output_gate_bias_tmp, dummy12, param12}, {fw_output_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_input_weights_tmp, dummy13, param13}, {bw_input_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_forget_weights_tmp, dummy14, param14}, {bw_input_to_forget_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_cell_weights_tmp, dummy15, param15}, {bw_input_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_output_weights_tmp, dummy16, param16}, {bw_input_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_recurrent_to_input_weights_tmp, dummy17, param17}, {bw_recurrent_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_recurrent_to_forget_weights_tmp, dummy18, param18}, {bw_recurrent_to_forget_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_recurrent_to_cell_weights_tmp, dummy19, param19}, {bw_recurrent_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_recurrent_to_output_weights_tmp, dummy20, param20}, {bw_recurrent_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_gate_bias_tmp, dummy21, param21}, {bw_input_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {bw_forget_gate_bias_tmp, dummy22, param22}, {bw_forget_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {bw_cell_bias_tmp, dummy23, param23}, {bw_cell_bias});
model->addOperation(ANEURALNETWORKS_ADD, {bw_output_gate_bias_tmp, dummy24, param24}, {bw_output_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {fw_activatiom_state_tmp, dummy25, param25}, {fw_activatiom_state});
model->addOperation(ANEURALNETWORKS_ADD, {fw_cell_state_tmp, dummy26, param26}, {fw_cell_state});
model->addOperation(ANEURALNETWORKS_ADD, {bw_activatiom_state_tmp, dummy27, param27}, {bw_activatiom_state});
model->addOperation(ANEURALNETWORKS_ADD, {bw_cell_state_tmp, dummy28, param28}, {bw_cell_state});
model->addOperation(ANEURALNETWORKS_ADD, {input1_tmp, dummy29, param29}, {input1});
model->addOperation(ANEURALNETWORKS_ADD, {fw_aux_input_to_input_weights_tmp, dummy30, param30}, {fw_aux_input_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_forget_weights1_tmp, dummy31, param31}, {fw_input_to_forget_weights1});
model->addOperation(ANEURALNETWORKS_ADD, {fw_aux_input_to_cell_weights_tmp, dummy32, param32}, {fw_aux_input_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_aux_input_to_output_weights_tmp, dummy33, param33}, {fw_aux_input_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_aux_input_to_input_weights_tmp, dummy34, param34}, {bw_aux_input_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_forget_weights1_tmp, dummy35, param35}, {bw_input_to_forget_weights1});
model->addOperation(ANEURALNETWORKS_ADD, {bw_aux_input_to_cell_weights_tmp, dummy36, param36}, {bw_aux_input_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_aux_input_to_output_weights_tmp, dummy37, param37}, {bw_aux_input_to_output_weights});
model->addOperation(ANEURALNETWORKS_BIDIRECTIONAL_SEQUENCE_LSTM, {input, fw_input_to_input_weights, fw_input_to_forget_weights, fw_input_to_cell_weights, fw_input_to_output_weights, fw_recurrent_to_input_weights, fw_recurrent_to_forget_weights, fw_recurrent_to_cell_weights, fw_recurrent_to_output_weights, fw_cell_to_input_weights, fw_cell_to_forget_weights, fw_cell_to_output_weights, fw_input_gate_bias, fw_forget_gate_bias, fw_cell_bias, fw_output_gate_bias, fw_projection_weights, fw_projection_bias, bw_input_to_input_weights, bw_input_to_forget_weights, bw_input_to_cell_weights, bw_input_to_output_weights, bw_recurrent_to_input_weights, bw_recurrent_to_forget_weights, bw_recurrent_to_cell_weights, bw_recurrent_to_output_weights, bw_cell_to_input_weights, bw_cell_to_forget_weights, bw_cell_to_output_weights, bw_input_gate_bias, bw_forget_gate_bias, bw_cell_bias, bw_output_gate_bias, bw_projection_weights, bw_projection_bias, fw_activatiom_state, fw_cell_state, bw_activatiom_state, bw_cell_state, input1, fw_aux_input_to_input_weights, fw_input_to_forget_weights1, fw_aux_input_to_cell_weights, fw_aux_input_to_output_weights, bw_aux_input_to_input_weights, bw_input_to_forget_weights1, bw_aux_input_to_cell_weights, bw_aux_input_to_output_weights, activation, cell_clip, proj_clip, merge_outputs, time_major, input_layer_norm_weights, forget_layer_norm_weights, cell_layer_norm_weights, output_layer_norm_weights, input_layer_norm_weights1, forget_layer_norm_weights1, cell_layer_norm_weights1, output_layer_norm_weights1}, {fw_output, bw_output});
// Phase 3, inputs and outputs
model->identifyInputsAndOutputs(
{fw_cell_to_input_weights, fw_cell_to_forget_weights, fw_cell_to_output_weights, fw_projection_weights, fw_projection_bias, bw_cell_to_input_weights, bw_cell_to_forget_weights, bw_cell_to_output_weights, bw_projection_weights, bw_projection_bias, input_layer_norm_weights, forget_layer_norm_weights, cell_layer_norm_weights, output_layer_norm_weights, input_layer_norm_weights1, forget_layer_norm_weights1, cell_layer_norm_weights1, output_layer_norm_weights1, input_tmp, fw_input_to_input_weights_tmp, fw_input_to_forget_weights_tmp, fw_input_to_cell_weights_tmp, fw_input_to_output_weights_tmp, fw_recurrent_to_input_weights_tmp, fw_recurrent_to_forget_weights_tmp, fw_recurrent_to_cell_weights_tmp, fw_recurrent_to_output_weights_tmp, fw_input_gate_bias_tmp, fw_forget_gate_bias_tmp, fw_cell_bias_tmp, fw_output_gate_bias_tmp, bw_input_to_input_weights_tmp, bw_input_to_forget_weights_tmp, bw_input_to_cell_weights_tmp, bw_input_to_output_weights_tmp, bw_recurrent_to_input_weights_tmp, bw_recurrent_to_forget_weights_tmp, bw_recurrent_to_cell_weights_tmp, bw_recurrent_to_output_weights_tmp, bw_input_gate_bias_tmp, bw_forget_gate_bias_tmp, bw_cell_bias_tmp, bw_output_gate_bias_tmp, fw_activatiom_state_tmp, fw_cell_state_tmp, bw_activatiom_state_tmp, bw_cell_state_tmp, input1_tmp, fw_aux_input_to_input_weights_tmp, fw_input_to_forget_weights1_tmp, fw_aux_input_to_cell_weights_tmp, fw_aux_input_to_output_weights_tmp, bw_aux_input_to_input_weights_tmp, bw_input_to_forget_weights1_tmp, bw_aux_input_to_cell_weights_tmp, bw_aux_input_to_output_weights_tmp},
{fw_output, bw_output});
assert(model->isValid());
}
bool is_ignored_all_inputs_as_internal(int i) {
static std::set<int> ignore = {};
return ignore.find(i) != ignore.end();
}
} // namespace generated_tests::bidirectional_sequence_lstm_float16_batch_major_aux_input
namespace generated_tests::bidirectional_sequence_lstm_float16_batch_major_aux_input {
void CreateModel_all_inputs_as_internal_dynamic_output_shape(Model *model) {
OperandType type0(Type::TENSOR_FLOAT16, {1, 3, 2});
OperandType type1(Type::TENSOR_FLOAT16, {4, 2});
OperandType type10(Type::TENSOR_FLOAT16, {1});
OperandType type2(Type::TENSOR_FLOAT16, {4, 4});
OperandType type3(Type::TENSOR_FLOAT16, {4});
OperandType type4(Type::TENSOR_FLOAT16, {1, 4});
OperandType type6(Type::INT32, {});
OperandType type7(Type::FLOAT16, {});
OperandType type8(Type::BOOL, {});
OperandType type9(Type::TENSOR_FLOAT16, {0, 0, 0});
// Phase 1, operands
auto input = model->addOperand(&type0);
auto fw_input_to_input_weights = model->addOperand(&type1);
auto fw_input_to_forget_weights = model->addOperand(&type1);
auto fw_input_to_cell_weights = model->addOperand(&type1);
auto fw_input_to_output_weights = model->addOperand(&type1);
auto fw_recurrent_to_input_weights = model->addOperand(&type2);
auto fw_recurrent_to_forget_weights = model->addOperand(&type2);
auto fw_recurrent_to_cell_weights = model->addOperand(&type2);
auto fw_recurrent_to_output_weights = model->addOperand(&type2);
auto fw_cell_to_input_weights = model->addOperand(&type3);
auto fw_cell_to_forget_weights = model->addOperand(&type3);
auto fw_cell_to_output_weights = model->addOperand(&type3);
auto fw_input_gate_bias = model->addOperand(&type3);
auto fw_forget_gate_bias = model->addOperand(&type3);
auto fw_cell_bias = model->addOperand(&type3);
auto fw_output_gate_bias = model->addOperand(&type3);
auto fw_projection_weights = model->addOperand(&type2);
auto fw_projection_bias = model->addOperand(&type3);
auto bw_input_to_input_weights = model->addOperand(&type1);
auto bw_input_to_forget_weights = model->addOperand(&type1);
auto bw_input_to_cell_weights = model->addOperand(&type1);
auto bw_input_to_output_weights = model->addOperand(&type1);
auto bw_recurrent_to_input_weights = model->addOperand(&type2);
auto bw_recurrent_to_forget_weights = model->addOperand(&type2);
auto bw_recurrent_to_cell_weights = model->addOperand(&type2);
auto bw_recurrent_to_output_weights = model->addOperand(&type2);
auto bw_cell_to_input_weights = model->addOperand(&type3);
auto bw_cell_to_forget_weights = model->addOperand(&type3);
auto bw_cell_to_output_weights = model->addOperand(&type3);
auto bw_input_gate_bias = model->addOperand(&type3);
auto bw_forget_gate_bias = model->addOperand(&type3);
auto bw_cell_bias = model->addOperand(&type3);
auto bw_output_gate_bias = model->addOperand(&type3);
auto bw_projection_weights = model->addOperand(&type2);
auto bw_projection_bias = model->addOperand(&type3);
auto fw_activatiom_state = model->addOperand(&type4);
auto fw_cell_state = model->addOperand(&type4);
auto bw_activatiom_state = model->addOperand(&type4);
auto bw_cell_state = model->addOperand(&type4);
auto input1 = model->addOperand(&type0);
auto fw_aux_input_to_input_weights = model->addOperand(&type1);
auto fw_input_to_forget_weights1 = model->addOperand(&type1);
auto fw_aux_input_to_cell_weights = model->addOperand(&type1);
auto fw_aux_input_to_output_weights = model->addOperand(&type1);
auto bw_aux_input_to_input_weights = model->addOperand(&type1);
auto bw_input_to_forget_weights1 = model->addOperand(&type1);
auto bw_aux_input_to_cell_weights = model->addOperand(&type1);
auto bw_aux_input_to_output_weights = model->addOperand(&type1);
auto activation = model->addOperand(&type6);
auto cell_clip = model->addOperand(&type7);
auto proj_clip = model->addOperand(&type7);
auto merge_outputs = model->addOperand(&type8);
auto time_major = model->addOperand(&type8);
auto input_layer_norm_weights = model->addOperand(&type3);
auto forget_layer_norm_weights = model->addOperand(&type3);
auto cell_layer_norm_weights = model->addOperand(&type3);
auto output_layer_norm_weights = model->addOperand(&type3);
auto input_layer_norm_weights1 = model->addOperand(&type3);
auto forget_layer_norm_weights1 = model->addOperand(&type3);
auto cell_layer_norm_weights1 = model->addOperand(&type3);
auto output_layer_norm_weights1 = model->addOperand(&type3);
auto fw_output = model->addOperand(&type9);
auto bw_output = model->addOperand(&type9);
auto input_tmp = model->addOperand(&type0);
auto dummy38 = model->addOperand(&type10);
auto param38 = model->addOperand(&type6);
auto fw_input_to_input_weights_tmp = model->addOperand(&type1);
auto dummy39 = model->addOperand(&type10);
auto param39 = model->addOperand(&type6);
auto fw_input_to_forget_weights_tmp = model->addOperand(&type1);
auto dummy40 = model->addOperand(&type10);
auto param40 = model->addOperand(&type6);
auto fw_input_to_cell_weights_tmp = model->addOperand(&type1);
auto dummy41 = model->addOperand(&type10);
auto param41 = model->addOperand(&type6);
auto fw_input_to_output_weights_tmp = model->addOperand(&type1);
auto dummy42 = model->addOperand(&type10);
auto param42 = model->addOperand(&type6);
auto fw_recurrent_to_input_weights_tmp = model->addOperand(&type2);
auto dummy43 = model->addOperand(&type10);
auto param43 = model->addOperand(&type6);
auto fw_recurrent_to_forget_weights_tmp = model->addOperand(&type2);
auto dummy44 = model->addOperand(&type10);
auto param44 = model->addOperand(&type6);
auto fw_recurrent_to_cell_weights_tmp = model->addOperand(&type2);
auto dummy45 = model->addOperand(&type10);
auto param45 = model->addOperand(&type6);
auto fw_recurrent_to_output_weights_tmp = model->addOperand(&type2);
auto dummy46 = model->addOperand(&type10);
auto param46 = model->addOperand(&type6);
auto fw_input_gate_bias_tmp = model->addOperand(&type3);
auto dummy47 = model->addOperand(&type10);
auto param47 = model->addOperand(&type6);
auto fw_forget_gate_bias_tmp = model->addOperand(&type3);
auto dummy48 = model->addOperand(&type10);
auto param48 = model->addOperand(&type6);
auto fw_cell_bias_tmp = model->addOperand(&type3);
auto dummy49 = model->addOperand(&type10);
auto param49 = model->addOperand(&type6);
auto fw_output_gate_bias_tmp = model->addOperand(&type3);
auto dummy50 = model->addOperand(&type10);
auto param50 = model->addOperand(&type6);
auto bw_input_to_input_weights_tmp = model->addOperand(&type1);
auto dummy51 = model->addOperand(&type10);
auto param51 = model->addOperand(&type6);
auto bw_input_to_forget_weights_tmp = model->addOperand(&type1);
auto dummy52 = model->addOperand(&type10);
auto param52 = model->addOperand(&type6);
auto bw_input_to_cell_weights_tmp = model->addOperand(&type1);
auto dummy53 = model->addOperand(&type10);
auto param53 = model->addOperand(&type6);
auto bw_input_to_output_weights_tmp = model->addOperand(&type1);
auto dummy54 = model->addOperand(&type10);
auto param54 = model->addOperand(&type6);
auto bw_recurrent_to_input_weights_tmp = model->addOperand(&type2);
auto dummy55 = model->addOperand(&type10);
auto param55 = model->addOperand(&type6);
auto bw_recurrent_to_forget_weights_tmp = model->addOperand(&type2);
auto dummy56 = model->addOperand(&type10);
auto param56 = model->addOperand(&type6);
auto bw_recurrent_to_cell_weights_tmp = model->addOperand(&type2);
auto dummy57 = model->addOperand(&type10);
auto param57 = model->addOperand(&type6);
auto bw_recurrent_to_output_weights_tmp = model->addOperand(&type2);
auto dummy58 = model->addOperand(&type10);
auto param58 = model->addOperand(&type6);
auto bw_input_gate_bias_tmp = model->addOperand(&type3);
auto dummy59 = model->addOperand(&type10);
auto param59 = model->addOperand(&type6);
auto bw_forget_gate_bias_tmp = model->addOperand(&type3);
auto dummy60 = model->addOperand(&type10);
auto param60 = model->addOperand(&type6);
auto bw_cell_bias_tmp = model->addOperand(&type3);
auto dummy61 = model->addOperand(&type10);
auto param61 = model->addOperand(&type6);
auto bw_output_gate_bias_tmp = model->addOperand(&type3);
auto dummy62 = model->addOperand(&type10);
auto param62 = model->addOperand(&type6);
auto fw_activatiom_state_tmp = model->addOperand(&type4);
auto dummy63 = model->addOperand(&type10);
auto param63 = model->addOperand(&type6);
auto fw_cell_state_tmp = model->addOperand(&type4);
auto dummy64 = model->addOperand(&type10);
auto param64 = model->addOperand(&type6);
auto bw_activatiom_state_tmp = model->addOperand(&type4);
auto dummy65 = model->addOperand(&type10);
auto param65 = model->addOperand(&type6);
auto bw_cell_state_tmp = model->addOperand(&type4);
auto dummy66 = model->addOperand(&type10);
auto param66 = model->addOperand(&type6);
auto input1_tmp = model->addOperand(&type0);
auto dummy67 = model->addOperand(&type10);
auto param67 = model->addOperand(&type6);
auto fw_aux_input_to_input_weights_tmp = model->addOperand(&type1);
auto dummy68 = model->addOperand(&type10);
auto param68 = model->addOperand(&type6);
auto fw_input_to_forget_weights1_tmp = model->addOperand(&type1);
auto dummy69 = model->addOperand(&type10);
auto param69 = model->addOperand(&type6);
auto fw_aux_input_to_cell_weights_tmp = model->addOperand(&type1);
auto dummy70 = model->addOperand(&type10);
auto param70 = model->addOperand(&type6);
auto fw_aux_input_to_output_weights_tmp = model->addOperand(&type1);
auto dummy71 = model->addOperand(&type10);
auto param71 = model->addOperand(&type6);
auto bw_aux_input_to_input_weights_tmp = model->addOperand(&type1);
auto dummy72 = model->addOperand(&type10);
auto param72 = model->addOperand(&type6);
auto bw_input_to_forget_weights1_tmp = model->addOperand(&type1);
auto dummy73 = model->addOperand(&type10);
auto param73 = model->addOperand(&type6);
auto bw_aux_input_to_cell_weights_tmp = model->addOperand(&type1);
auto dummy74 = model->addOperand(&type10);
auto param74 = model->addOperand(&type6);
auto bw_aux_input_to_output_weights_tmp = model->addOperand(&type1);
auto dummy75 = model->addOperand(&type10);
auto param75 = model->addOperand(&type6);
// Phase 2, operations
static int32_t activation_init[] = {4};
model->setOperandValue(activation, activation_init, sizeof(int32_t) * 1);
static _Float16 cell_clip_init[] = {0.0f};
model->setOperandValue(cell_clip, cell_clip_init, sizeof(_Float16) * 1);
static _Float16 proj_clip_init[] = {0.0f};
model->setOperandValue(proj_clip, proj_clip_init, sizeof(_Float16) * 1);
static bool8 merge_outputs_init[] = {false};
model->setOperandValue(merge_outputs, merge_outputs_init, sizeof(bool8) * 1);
static bool8 time_major_init[] = {false};
model->setOperandValue(time_major, time_major_init, sizeof(bool8) * 1);
static _Float16 dummy38_init[] = {0.0f};
model->setOperandValue(dummy38, dummy38_init, sizeof(_Float16) * 1);
static int32_t param38_init[] = {0};
model->setOperandValue(param38, param38_init, sizeof(int32_t) * 1);
static _Float16 dummy39_init[] = {0.0f};
model->setOperandValue(dummy39, dummy39_init, sizeof(_Float16) * 1);
static int32_t param39_init[] = {0};
model->setOperandValue(param39, param39_init, sizeof(int32_t) * 1);
static _Float16 dummy40_init[] = {0.0f};
model->setOperandValue(dummy40, dummy40_init, sizeof(_Float16) * 1);
static int32_t param40_init[] = {0};
model->setOperandValue(param40, param40_init, sizeof(int32_t) * 1);
static _Float16 dummy41_init[] = {0.0f};
model->setOperandValue(dummy41, dummy41_init, sizeof(_Float16) * 1);
static int32_t param41_init[] = {0};
model->setOperandValue(param41, param41_init, sizeof(int32_t) * 1);
static _Float16 dummy42_init[] = {0.0f};
model->setOperandValue(dummy42, dummy42_init, sizeof(_Float16) * 1);
static int32_t param42_init[] = {0};
model->setOperandValue(param42, param42_init, sizeof(int32_t) * 1);
static _Float16 dummy43_init[] = {0.0f};
model->setOperandValue(dummy43, dummy43_init, sizeof(_Float16) * 1);
static int32_t param43_init[] = {0};
model->setOperandValue(param43, param43_init, sizeof(int32_t) * 1);
static _Float16 dummy44_init[] = {0.0f};
model->setOperandValue(dummy44, dummy44_init, sizeof(_Float16) * 1);
static int32_t param44_init[] = {0};
model->setOperandValue(param44, param44_init, sizeof(int32_t) * 1);
static _Float16 dummy45_init[] = {0.0f};
model->setOperandValue(dummy45, dummy45_init, sizeof(_Float16) * 1);
static int32_t param45_init[] = {0};
model->setOperandValue(param45, param45_init, sizeof(int32_t) * 1);
static _Float16 dummy46_init[] = {0.0f};
model->setOperandValue(dummy46, dummy46_init, sizeof(_Float16) * 1);
static int32_t param46_init[] = {0};
model->setOperandValue(param46, param46_init, sizeof(int32_t) * 1);
static _Float16 dummy47_init[] = {0.0f};
model->setOperandValue(dummy47, dummy47_init, sizeof(_Float16) * 1);
static int32_t param47_init[] = {0};
model->setOperandValue(param47, param47_init, sizeof(int32_t) * 1);
static _Float16 dummy48_init[] = {0.0f};
model->setOperandValue(dummy48, dummy48_init, sizeof(_Float16) * 1);
static int32_t param48_init[] = {0};
model->setOperandValue(param48, param48_init, sizeof(int32_t) * 1);
static _Float16 dummy49_init[] = {0.0f};
model->setOperandValue(dummy49, dummy49_init, sizeof(_Float16) * 1);
static int32_t param49_init[] = {0};
model->setOperandValue(param49, param49_init, sizeof(int32_t) * 1);
static _Float16 dummy50_init[] = {0.0f};
model->setOperandValue(dummy50, dummy50_init, sizeof(_Float16) * 1);
static int32_t param50_init[] = {0};
model->setOperandValue(param50, param50_init, sizeof(int32_t) * 1);
static _Float16 dummy51_init[] = {0.0f};
model->setOperandValue(dummy51, dummy51_init, sizeof(_Float16) * 1);
static int32_t param51_init[] = {0};
model->setOperandValue(param51, param51_init, sizeof(int32_t) * 1);
static _Float16 dummy52_init[] = {0.0f};
model->setOperandValue(dummy52, dummy52_init, sizeof(_Float16) * 1);
static int32_t param52_init[] = {0};
model->setOperandValue(param52, param52_init, sizeof(int32_t) * 1);
static _Float16 dummy53_init[] = {0.0f};
model->setOperandValue(dummy53, dummy53_init, sizeof(_Float16) * 1);
static int32_t param53_init[] = {0};
model->setOperandValue(param53, param53_init, sizeof(int32_t) * 1);
static _Float16 dummy54_init[] = {0.0f};
model->setOperandValue(dummy54, dummy54_init, sizeof(_Float16) * 1);
static int32_t param54_init[] = {0};
model->setOperandValue(param54, param54_init, sizeof(int32_t) * 1);
static _Float16 dummy55_init[] = {0.0f};
model->setOperandValue(dummy55, dummy55_init, sizeof(_Float16) * 1);
static int32_t param55_init[] = {0};
model->setOperandValue(param55, param55_init, sizeof(int32_t) * 1);
static _Float16 dummy56_init[] = {0.0f};
model->setOperandValue(dummy56, dummy56_init, sizeof(_Float16) * 1);
static int32_t param56_init[] = {0};
model->setOperandValue(param56, param56_init, sizeof(int32_t) * 1);
static _Float16 dummy57_init[] = {0.0f};
model->setOperandValue(dummy57, dummy57_init, sizeof(_Float16) * 1);
static int32_t param57_init[] = {0};
model->setOperandValue(param57, param57_init, sizeof(int32_t) * 1);
static _Float16 dummy58_init[] = {0.0f};
model->setOperandValue(dummy58, dummy58_init, sizeof(_Float16) * 1);
static int32_t param58_init[] = {0};
model->setOperandValue(param58, param58_init, sizeof(int32_t) * 1);
static _Float16 dummy59_init[] = {0.0f};
model->setOperandValue(dummy59, dummy59_init, sizeof(_Float16) * 1);
static int32_t param59_init[] = {0};
model->setOperandValue(param59, param59_init, sizeof(int32_t) * 1);
static _Float16 dummy60_init[] = {0.0f};
model->setOperandValue(dummy60, dummy60_init, sizeof(_Float16) * 1);
static int32_t param60_init[] = {0};
model->setOperandValue(param60, param60_init, sizeof(int32_t) * 1);
static _Float16 dummy61_init[] = {0.0f};
model->setOperandValue(dummy61, dummy61_init, sizeof(_Float16) * 1);
static int32_t param61_init[] = {0};
model->setOperandValue(param61, param61_init, sizeof(int32_t) * 1);
static _Float16 dummy62_init[] = {0.0f};
model->setOperandValue(dummy62, dummy62_init, sizeof(_Float16) * 1);
static int32_t param62_init[] = {0};
model->setOperandValue(param62, param62_init, sizeof(int32_t) * 1);
static _Float16 dummy63_init[] = {0.0f};
model->setOperandValue(dummy63, dummy63_init, sizeof(_Float16) * 1);
static int32_t param63_init[] = {0};
model->setOperandValue(param63, param63_init, sizeof(int32_t) * 1);
static _Float16 dummy64_init[] = {0.0f};
model->setOperandValue(dummy64, dummy64_init, sizeof(_Float16) * 1);
static int32_t param64_init[] = {0};
model->setOperandValue(param64, param64_init, sizeof(int32_t) * 1);
static _Float16 dummy65_init[] = {0.0f};
model->setOperandValue(dummy65, dummy65_init, sizeof(_Float16) * 1);
static int32_t param65_init[] = {0};
model->setOperandValue(param65, param65_init, sizeof(int32_t) * 1);
static _Float16 dummy66_init[] = {0.0f};
model->setOperandValue(dummy66, dummy66_init, sizeof(_Float16) * 1);
static int32_t param66_init[] = {0};
model->setOperandValue(param66, param66_init, sizeof(int32_t) * 1);
static _Float16 dummy67_init[] = {0.0f};
model->setOperandValue(dummy67, dummy67_init, sizeof(_Float16) * 1);
static int32_t param67_init[] = {0};
model->setOperandValue(param67, param67_init, sizeof(int32_t) * 1);
static _Float16 dummy68_init[] = {0.0f};
model->setOperandValue(dummy68, dummy68_init, sizeof(_Float16) * 1);
static int32_t param68_init[] = {0};
model->setOperandValue(param68, param68_init, sizeof(int32_t) * 1);
static _Float16 dummy69_init[] = {0.0f};
model->setOperandValue(dummy69, dummy69_init, sizeof(_Float16) * 1);
static int32_t param69_init[] = {0};
model->setOperandValue(param69, param69_init, sizeof(int32_t) * 1);
static _Float16 dummy70_init[] = {0.0f};
model->setOperandValue(dummy70, dummy70_init, sizeof(_Float16) * 1);
static int32_t param70_init[] = {0};
model->setOperandValue(param70, param70_init, sizeof(int32_t) * 1);
static _Float16 dummy71_init[] = {0.0f};
model->setOperandValue(dummy71, dummy71_init, sizeof(_Float16) * 1);
static int32_t param71_init[] = {0};
model->setOperandValue(param71, param71_init, sizeof(int32_t) * 1);
static _Float16 dummy72_init[] = {0.0f};
model->setOperandValue(dummy72, dummy72_init, sizeof(_Float16) * 1);
static int32_t param72_init[] = {0};
model->setOperandValue(param72, param72_init, sizeof(int32_t) * 1);
static _Float16 dummy73_init[] = {0.0f};
model->setOperandValue(dummy73, dummy73_init, sizeof(_Float16) * 1);
static int32_t param73_init[] = {0};
model->setOperandValue(param73, param73_init, sizeof(int32_t) * 1);
static _Float16 dummy74_init[] = {0.0f};
model->setOperandValue(dummy74, dummy74_init, sizeof(_Float16) * 1);
static int32_t param74_init[] = {0};
model->setOperandValue(param74, param74_init, sizeof(int32_t) * 1);
static _Float16 dummy75_init[] = {0.0f};
model->setOperandValue(dummy75, dummy75_init, sizeof(_Float16) * 1);
static int32_t param75_init[] = {0};
model->setOperandValue(param75, param75_init, sizeof(int32_t) * 1);
model->addOperation(ANEURALNETWORKS_ADD, {input_tmp, dummy38, param38}, {input});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_input_weights_tmp, dummy39, param39}, {fw_input_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_forget_weights_tmp, dummy40, param40}, {fw_input_to_forget_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_cell_weights_tmp, dummy41, param41}, {fw_input_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_output_weights_tmp, dummy42, param42}, {fw_input_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_recurrent_to_input_weights_tmp, dummy43, param43}, {fw_recurrent_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_recurrent_to_forget_weights_tmp, dummy44, param44}, {fw_recurrent_to_forget_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_recurrent_to_cell_weights_tmp, dummy45, param45}, {fw_recurrent_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_recurrent_to_output_weights_tmp, dummy46, param46}, {fw_recurrent_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_gate_bias_tmp, dummy47, param47}, {fw_input_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {fw_forget_gate_bias_tmp, dummy48, param48}, {fw_forget_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {fw_cell_bias_tmp, dummy49, param49}, {fw_cell_bias});
model->addOperation(ANEURALNETWORKS_ADD, {fw_output_gate_bias_tmp, dummy50, param50}, {fw_output_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_input_weights_tmp, dummy51, param51}, {bw_input_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_forget_weights_tmp, dummy52, param52}, {bw_input_to_forget_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_cell_weights_tmp, dummy53, param53}, {bw_input_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_output_weights_tmp, dummy54, param54}, {bw_input_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_recurrent_to_input_weights_tmp, dummy55, param55}, {bw_recurrent_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_recurrent_to_forget_weights_tmp, dummy56, param56}, {bw_recurrent_to_forget_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_recurrent_to_cell_weights_tmp, dummy57, param57}, {bw_recurrent_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_recurrent_to_output_weights_tmp, dummy58, param58}, {bw_recurrent_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_gate_bias_tmp, dummy59, param59}, {bw_input_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {bw_forget_gate_bias_tmp, dummy60, param60}, {bw_forget_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {bw_cell_bias_tmp, dummy61, param61}, {bw_cell_bias});
model->addOperation(ANEURALNETWORKS_ADD, {bw_output_gate_bias_tmp, dummy62, param62}, {bw_output_gate_bias});
model->addOperation(ANEURALNETWORKS_ADD, {fw_activatiom_state_tmp, dummy63, param63}, {fw_activatiom_state});
model->addOperation(ANEURALNETWORKS_ADD, {fw_cell_state_tmp, dummy64, param64}, {fw_cell_state});
model->addOperation(ANEURALNETWORKS_ADD, {bw_activatiom_state_tmp, dummy65, param65}, {bw_activatiom_state});
model->addOperation(ANEURALNETWORKS_ADD, {bw_cell_state_tmp, dummy66, param66}, {bw_cell_state});
model->addOperation(ANEURALNETWORKS_ADD, {input1_tmp, dummy67, param67}, {input1});
model->addOperation(ANEURALNETWORKS_ADD, {fw_aux_input_to_input_weights_tmp, dummy68, param68}, {fw_aux_input_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_input_to_forget_weights1_tmp, dummy69, param69}, {fw_input_to_forget_weights1});
model->addOperation(ANEURALNETWORKS_ADD, {fw_aux_input_to_cell_weights_tmp, dummy70, param70}, {fw_aux_input_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {fw_aux_input_to_output_weights_tmp, dummy71, param71}, {fw_aux_input_to_output_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_aux_input_to_input_weights_tmp, dummy72, param72}, {bw_aux_input_to_input_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_input_to_forget_weights1_tmp, dummy73, param73}, {bw_input_to_forget_weights1});
model->addOperation(ANEURALNETWORKS_ADD, {bw_aux_input_to_cell_weights_tmp, dummy74, param74}, {bw_aux_input_to_cell_weights});
model->addOperation(ANEURALNETWORKS_ADD, {bw_aux_input_to_output_weights_tmp, dummy75, param75}, {bw_aux_input_to_output_weights});
model->addOperation(ANEURALNETWORKS_BIDIRECTIONAL_SEQUENCE_LSTM, {input, fw_input_to_input_weights, fw_input_to_forget_weights, fw_input_to_cell_weights, fw_input_to_output_weights, fw_recurrent_to_input_weights, fw_recurrent_to_forget_weights, fw_recurrent_to_cell_weights, fw_recurrent_to_output_weights, fw_cell_to_input_weights, fw_cell_to_forget_weights, fw_cell_to_output_weights, fw_input_gate_bias, fw_forget_gate_bias, fw_cell_bias, fw_output_gate_bias, fw_projection_weights, fw_projection_bias, bw_input_to_input_weights, bw_input_to_forget_weights, bw_input_to_cell_weights, bw_input_to_output_weights, bw_recurrent_to_input_weights, bw_recurrent_to_forget_weights, bw_recurrent_to_cell_weights, bw_recurrent_to_output_weights, bw_cell_to_input_weights, bw_cell_to_forget_weights, bw_cell_to_output_weights, bw_input_gate_bias, bw_forget_gate_bias, bw_cell_bias, bw_output_gate_bias, bw_projection_weights, bw_projection_bias, fw_activatiom_state, fw_cell_state, bw_activatiom_state, bw_cell_state, input1, fw_aux_input_to_input_weights, fw_input_to_forget_weights1, fw_aux_input_to_cell_weights, fw_aux_input_to_output_weights, bw_aux_input_to_input_weights, bw_input_to_forget_weights1, bw_aux_input_to_cell_weights, bw_aux_input_to_output_weights, activation, cell_clip, proj_clip, merge_outputs, time_major, input_layer_norm_weights, forget_layer_norm_weights, cell_layer_norm_weights, output_layer_norm_weights, input_layer_norm_weights1, forget_layer_norm_weights1, cell_layer_norm_weights1, output_layer_norm_weights1}, {fw_output, bw_output});
// Phase 3, inputs and outputs
model->identifyInputsAndOutputs(
{fw_cell_to_input_weights, fw_cell_to_forget_weights, fw_cell_to_output_weights, fw_projection_weights, fw_projection_bias, bw_cell_to_input_weights, bw_cell_to_forget_weights, bw_cell_to_output_weights, bw_projection_weights, bw_projection_bias, input_layer_norm_weights, forget_layer_norm_weights, cell_layer_norm_weights, output_layer_norm_weights, input_layer_norm_weights1, forget_layer_norm_weights1, cell_layer_norm_weights1, output_layer_norm_weights1, input_tmp, fw_input_to_input_weights_tmp, fw_input_to_forget_weights_tmp, fw_input_to_cell_weights_tmp, fw_input_to_output_weights_tmp, fw_recurrent_to_input_weights_tmp, fw_recurrent_to_forget_weights_tmp, fw_recurrent_to_cell_weights_tmp, fw_recurrent_to_output_weights_tmp, fw_input_gate_bias_tmp, fw_forget_gate_bias_tmp, fw_cell_bias_tmp, fw_output_gate_bias_tmp, bw_input_to_input_weights_tmp, bw_input_to_forget_weights_tmp, bw_input_to_cell_weights_tmp, bw_input_to_output_weights_tmp, bw_recurrent_to_input_weights_tmp, bw_recurrent_to_forget_weights_tmp, bw_recurrent_to_cell_weights_tmp, bw_recurrent_to_output_weights_tmp, bw_input_gate_bias_tmp, bw_forget_gate_bias_tmp, bw_cell_bias_tmp, bw_output_gate_bias_tmp, fw_activatiom_state_tmp, fw_cell_state_tmp, bw_activatiom_state_tmp, bw_cell_state_tmp, input1_tmp, fw_aux_input_to_input_weights_tmp, fw_input_to_forget_weights1_tmp, fw_aux_input_to_cell_weights_tmp, fw_aux_input_to_output_weights_tmp, bw_aux_input_to_input_weights_tmp, bw_input_to_forget_weights1_tmp, bw_aux_input_to_cell_weights_tmp, bw_aux_input_to_output_weights_tmp},
{fw_output, bw_output});
assert(model->isValid());
}
bool is_ignored_all_inputs_as_internal_dynamic_output_shape(int i) {
static std::set<int> ignore = {};
return ignore.find(i) != ignore.end();
}
} // namespace generated_tests::bidirectional_sequence_lstm_float16_batch_major_aux_input