program(1.0)
[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3510.2.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.7.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
{
    func main<ios17>(tensor<fp32, [1, 4, 2049, 476]> spectrogram) {
            tensor<fp32, [32]> encoder_0_SDlayer_convs_0_bias = const()[name = tensor<string, []>("encoder_0_SDlayer_convs_0_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
            tensor<fp32, [32, 4, 3, 1]> encoder_0_SDlayer_convs_0_weight = const()[name = tensor<string, []>("encoder_0_SDlayer_convs_0_weight"), val = tensor<fp32, [32, 4, 3, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(256)))];
            tensor<fp32, [32]> encoder_0_SDlayer_convs_1_bias = const()[name = tensor<string, []>("encoder_0_SDlayer_convs_1_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1856)))];
            tensor<fp32, [32, 4, 4, 1]> encoder_0_SDlayer_convs_1_weight = const()[name = tensor<string, []>("encoder_0_SDlayer_convs_1_weight"), val = tensor<fp32, [32, 4, 4, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2048)))];
            tensor<fp32, [32]> encoder_0_SDlayer_convs_2_bias = const()[name = tensor<string, []>("encoder_0_SDlayer_convs_2_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4160)))];
            tensor<fp32, [32, 4, 16, 1]> encoder_0_SDlayer_convs_2_weight = const()[name = tensor<string, []>("encoder_0_SDlayer_convs_2_weight"), val = tensor<fp32, [32, 4, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4352)))];
            tensor<fp32, [16]> encoder_0_conv_modules_0_layers_0_1_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_0_1_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12608)))];
            tensor<fp32, [16, 32, 3]> encoder_0_conv_modules_0_layers_0_1_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_0_1_weight"), val = tensor<fp32, [16, 32, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12736)))];
            tensor<fp32, [8]> encoder_0_conv_modules_0_layers_0_3_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_0_3_bias"), val = tensor<fp32, [8]>([-0x1.8188aap-4, -0x1.3c938ep-2, -0x1.157abap-3, -0x1.0fb47ap-2, 0x1.f781p-2, -0x1.c09e8ep-1, 0x1.9bfe8ap-4, 0x1.3ca12p-2])];
            tensor<fp32, [8, 1, 3]> encoder_0_conv_modules_0_layers_0_3_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_0_3_weight"), val = tensor<fp32, [8, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(18944)))];
            tensor<fp32, [32]> encoder_0_conv_modules_0_layers_0_6_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_0_6_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(19136)))];
            tensor<fp32, [32, 8, 1]> encoder_0_conv_modules_0_layers_0_6_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_0_6_weight"), val = tensor<fp32, [32, 8, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(19328)))];
            tensor<fp32, [16]> encoder_0_conv_modules_0_layers_1_1_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_1_1_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(20416)))];
            tensor<fp32, [16, 32, 3]> encoder_0_conv_modules_0_layers_1_1_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_1_1_weight"), val = tensor<fp32, [16, 32, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(20544)))];
            tensor<fp32, [8]> encoder_0_conv_modules_0_layers_1_3_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_1_3_bias"), val = tensor<fp32, [8]>([0x1.a322f2p-2, 0x1.0823c8p-2, 0x1.9caadap-2, 0x1.bf4baep-1, -0x1.52237cp-2, 0x1.13099cp-4, 0x1.371e1ep-2, 0x1.bcb924p-2])];
            tensor<fp32, [8, 1, 3]> encoder_0_conv_modules_0_layers_1_3_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_1_3_weight"), val = tensor<fp32, [8, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26752)))];
            tensor<fp32, [32]> encoder_0_conv_modules_0_layers_1_6_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_1_6_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26944)))];
            tensor<fp32, [32, 8, 1]> encoder_0_conv_modules_0_layers_1_6_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_1_6_weight"), val = tensor<fp32, [32, 8, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(27136)))];
            tensor<fp32, [16]> encoder_0_conv_modules_0_layers_2_1_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_2_1_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(28224)))];
            tensor<fp32, [16, 32, 3]> encoder_0_conv_modules_0_layers_2_1_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_2_1_weight"), val = tensor<fp32, [16, 32, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(28352)))];
            tensor<fp32, [8]> encoder_0_conv_modules_0_layers_2_3_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_2_3_bias"), val = tensor<fp32, [8]>([-0x1.3c7c9ep-2, -0x1.d426d2p-2, 0x1.2d4056p-2, -0x1.10407p-1, 0x1.4e6d18p-3, 0x1.e1c016p-2, 0x1.f4b6fcp-5, 0x1.48b04cp-2])];
            tensor<fp32, [8, 1, 3]> encoder_0_conv_modules_0_layers_2_3_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_2_3_weight"), val = tensor<fp32, [8, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34560)))];
            tensor<fp32, [32]> encoder_0_conv_modules_0_layers_2_6_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_2_6_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34752)))];
            tensor<fp32, [32, 8, 1]> encoder_0_conv_modules_0_layers_2_6_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_0_layers_2_6_weight"), val = tensor<fp32, [32, 8, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34944)))];
            tensor<fp32, [16]> encoder_0_conv_modules_1_layers_0_1_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_0_1_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36032)))];
            tensor<fp32, [16, 32, 3]> encoder_0_conv_modules_1_layers_0_1_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_0_1_weight"), val = tensor<fp32, [16, 32, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36160)))];
            tensor<fp32, [8]> encoder_0_conv_modules_1_layers_0_3_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_0_3_bias"), val = tensor<fp32, [8]>([0x1.55fbeep-2, -0x1.d54cc6p-3, -0x1.7b2fbap-2, 0x1.ddbc7p-3, 0x1.604dc4p-1, -0x1.7d24d2p-6, -0x1.43af54p-2, 0x1.5416ecp-2])];
            tensor<fp32, [8, 1, 3]> encoder_0_conv_modules_1_layers_0_3_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_0_3_weight"), val = tensor<fp32, [8, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(42368)))];
            tensor<fp32, [32]> encoder_0_conv_modules_1_layers_0_6_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_0_6_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(42560)))];
            tensor<fp32, [32, 8, 1]> encoder_0_conv_modules_1_layers_0_6_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_0_6_weight"), val = tensor<fp32, [32, 8, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(42752)))];
            tensor<fp32, [16]> encoder_0_conv_modules_1_layers_1_1_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_1_1_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(43840)))];
            tensor<fp32, [16, 32, 3]> encoder_0_conv_modules_1_layers_1_1_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_1_1_weight"), val = tensor<fp32, [16, 32, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(43968)))];
            tensor<fp32, [8]> encoder_0_conv_modules_1_layers_1_3_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_1_3_bias"), val = tensor<fp32, [8]>([-0x1.0fcc1ap-2, -0x1.c2c0bcp-5, 0x1.71341p-2, 0x1.60e398p-3, -0x1.dda3dcp-7, 0x1.8c918ap-2, -0x1.a20f8p-2, -0x1.eceb2cp-4])];
            tensor<fp32, [8, 1, 3]> encoder_0_conv_modules_1_layers_1_3_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_1_3_weight"), val = tensor<fp32, [8, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(50176)))];
            tensor<fp32, [32]> encoder_0_conv_modules_1_layers_1_6_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_1_6_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(50368)))];
            tensor<fp32, [32, 8, 1]> encoder_0_conv_modules_1_layers_1_6_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_1_layers_1_6_weight"), val = tensor<fp32, [32, 8, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(50560)))];
            tensor<fp32, [16]> encoder_0_conv_modules_2_layers_0_1_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_2_layers_0_1_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(51648)))];
            tensor<fp32, [16, 32, 3]> encoder_0_conv_modules_2_layers_0_1_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_2_layers_0_1_weight"), val = tensor<fp32, [16, 32, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(51776)))];
            tensor<fp32, [8]> encoder_0_conv_modules_2_layers_0_3_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_2_layers_0_3_bias"), val = tensor<fp32, [8]>([0x1.38141ap-1, -0x1.3954c2p-2, 0x1.0ef398p-1, 0x1.0b98fep-2, -0x1.e53718p-2, -0x1.003238p-2, -0x1.83e0f6p-7, -0x1.7a6c94p-6])];
            tensor<fp32, [8, 1, 3]> encoder_0_conv_modules_2_layers_0_3_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_2_layers_0_3_weight"), val = tensor<fp32, [8, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(57984)))];
            tensor<fp32, [32]> encoder_0_conv_modules_2_layers_0_6_bias = const()[name = tensor<string, []>("encoder_0_conv_modules_2_layers_0_6_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(58176)))];
            tensor<fp32, [32, 8, 1]> encoder_0_conv_modules_2_layers_0_6_weight = const()[name = tensor<string, []>("encoder_0_conv_modules_2_layers_0_6_weight"), val = tensor<fp32, [32, 8, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(58368)))];
            tensor<fp32, [32]> encoder_0_globalconv_bias = const()[name = tensor<string, []>("encoder_0_globalconv_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(59456)))];
            tensor<fp32, [32, 32, 3, 3]> encoder_0_globalconv_weight = const()[name = tensor<string, []>("encoder_0_globalconv_weight"), val = tensor<fp32, [32, 32, 3, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(59648)))];
            tensor<fp32, [64]> encoder_1_SDlayer_convs_0_bias = const()[name = tensor<string, []>("encoder_1_SDlayer_convs_0_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(96576)))];
            tensor<fp32, [64, 32, 3, 1]> encoder_1_SDlayer_convs_0_weight = const()[name = tensor<string, []>("encoder_1_SDlayer_convs_0_weight"), val = tensor<fp32, [64, 32, 3, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(96896)))];
            tensor<fp32, [64]> encoder_1_SDlayer_convs_1_bias = const()[name = tensor<string, []>("encoder_1_SDlayer_convs_1_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(121536)))];
            tensor<fp32, [64, 32, 4, 1]> encoder_1_SDlayer_convs_1_weight = const()[name = tensor<string, []>("encoder_1_SDlayer_convs_1_weight"), val = tensor<fp32, [64, 32, 4, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(121856)))];
            tensor<fp32, [64]> encoder_1_SDlayer_convs_2_bias = const()[name = tensor<string, []>("encoder_1_SDlayer_convs_2_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(154688)))];
            tensor<fp32, [64, 32, 16, 1]> encoder_1_SDlayer_convs_2_weight = const()[name = tensor<string, []>("encoder_1_SDlayer_convs_2_weight"), val = tensor<fp32, [64, 32, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(155008)))];
            tensor<fp32, [32]> encoder_1_conv_modules_0_layers_0_1_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_0_1_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(286144)))];
            tensor<fp32, [32, 64, 3]> encoder_1_conv_modules_0_layers_0_1_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_0_1_weight"), val = tensor<fp32, [32, 64, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(286336)))];
            tensor<fp32, [16]> encoder_1_conv_modules_0_layers_0_3_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_0_3_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(310976)))];
            tensor<fp32, [16, 1, 3]> encoder_1_conv_modules_0_layers_0_3_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_0_3_weight"), val = tensor<fp32, [16, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(311104)))];
            tensor<fp32, [64]> encoder_1_conv_modules_0_layers_0_6_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_0_6_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(311360)))];
            tensor<fp32, [64, 16, 1]> encoder_1_conv_modules_0_layers_0_6_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_0_6_weight"), val = tensor<fp32, [64, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(311680)))];
            tensor<fp32, [32]> encoder_1_conv_modules_0_layers_1_1_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_1_1_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(315840)))];
            tensor<fp32, [32, 64, 3]> encoder_1_conv_modules_0_layers_1_1_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_1_1_weight"), val = tensor<fp32, [32, 64, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(316032)))];
            tensor<fp32, [16]> encoder_1_conv_modules_0_layers_1_3_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_1_3_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(340672)))];
            tensor<fp32, [16, 1, 3]> encoder_1_conv_modules_0_layers_1_3_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_1_3_weight"), val = tensor<fp32, [16, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(340800)))];
            tensor<fp32, [64]> encoder_1_conv_modules_0_layers_1_6_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_1_6_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(341056)))];
            tensor<fp32, [64, 16, 1]> encoder_1_conv_modules_0_layers_1_6_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_1_6_weight"), val = tensor<fp32, [64, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(341376)))];
            tensor<fp32, [32]> encoder_1_conv_modules_0_layers_2_1_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_2_1_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(345536)))];
            tensor<fp32, [32, 64, 3]> encoder_1_conv_modules_0_layers_2_1_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_2_1_weight"), val = tensor<fp32, [32, 64, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(345728)))];
            tensor<fp32, [16]> encoder_1_conv_modules_0_layers_2_3_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_2_3_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(370368)))];
            tensor<fp32, [16, 1, 3]> encoder_1_conv_modules_0_layers_2_3_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_2_3_weight"), val = tensor<fp32, [16, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(370496)))];
            tensor<fp32, [64]> encoder_1_conv_modules_0_layers_2_6_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_2_6_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(370752)))];
            tensor<fp32, [64, 16, 1]> encoder_1_conv_modules_0_layers_2_6_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_0_layers_2_6_weight"), val = tensor<fp32, [64, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(371072)))];
            tensor<fp32, [32]> encoder_1_conv_modules_1_layers_0_1_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_0_1_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(375232)))];
            tensor<fp32, [32, 64, 3]> encoder_1_conv_modules_1_layers_0_1_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_0_1_weight"), val = tensor<fp32, [32, 64, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(375424)))];
            tensor<fp32, [16]> encoder_1_conv_modules_1_layers_0_3_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_0_3_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(400064)))];
            tensor<fp32, [16, 1, 3]> encoder_1_conv_modules_1_layers_0_3_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_0_3_weight"), val = tensor<fp32, [16, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(400192)))];
            tensor<fp32, [64]> encoder_1_conv_modules_1_layers_0_6_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_0_6_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(400448)))];
            tensor<fp32, [64, 16, 1]> encoder_1_conv_modules_1_layers_0_6_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_0_6_weight"), val = tensor<fp32, [64, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(400768)))];
            tensor<fp32, [32]> encoder_1_conv_modules_1_layers_1_1_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_1_1_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(404928)))];
            tensor<fp32, [32, 64, 3]> encoder_1_conv_modules_1_layers_1_1_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_1_1_weight"), val = tensor<fp32, [32, 64, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(405120)))];
            tensor<fp32, [16]> encoder_1_conv_modules_1_layers_1_3_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_1_3_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(429760)))];
            tensor<fp32, [16, 1, 3]> encoder_1_conv_modules_1_layers_1_3_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_1_3_weight"), val = tensor<fp32, [16, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(429888)))];
            tensor<fp32, [64]> encoder_1_conv_modules_1_layers_1_6_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_1_6_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(430144)))];
            tensor<fp32, [64, 16, 1]> encoder_1_conv_modules_1_layers_1_6_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_1_layers_1_6_weight"), val = tensor<fp32, [64, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(430464)))];
            tensor<fp32, [32]> encoder_1_conv_modules_2_layers_0_1_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_2_layers_0_1_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(434624)))];
            tensor<fp32, [32, 64, 3]> encoder_1_conv_modules_2_layers_0_1_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_2_layers_0_1_weight"), val = tensor<fp32, [32, 64, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(434816)))];
            tensor<fp32, [16]> encoder_1_conv_modules_2_layers_0_3_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_2_layers_0_3_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(459456)))];
            tensor<fp32, [16, 1, 3]> encoder_1_conv_modules_2_layers_0_3_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_2_layers_0_3_weight"), val = tensor<fp32, [16, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(459584)))];
            tensor<fp32, [64]> encoder_1_conv_modules_2_layers_0_6_bias = const()[name = tensor<string, []>("encoder_1_conv_modules_2_layers_0_6_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(459840)))];
            tensor<fp32, [64, 16, 1]> encoder_1_conv_modules_2_layers_0_6_weight = const()[name = tensor<string, []>("encoder_1_conv_modules_2_layers_0_6_weight"), val = tensor<fp32, [64, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(460160)))];
            tensor<fp32, [64]> encoder_1_globalconv_bias = const()[name = tensor<string, []>("encoder_1_globalconv_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(464320)))];
            tensor<fp32, [64, 64, 3, 3]> encoder_1_globalconv_weight = const()[name = tensor<string, []>("encoder_1_globalconv_weight"), val = tensor<fp32, [64, 64, 3, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(464640)))];
            tensor<fp32, [128]> encoder_2_SDlayer_convs_0_bias = const()[name = tensor<string, []>("encoder_2_SDlayer_convs_0_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(612160)))];
            tensor<fp32, [128, 64, 3, 1]> encoder_2_SDlayer_convs_0_weight = const()[name = tensor<string, []>("encoder_2_SDlayer_convs_0_weight"), val = tensor<fp32, [128, 64, 3, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(612736)))];
            tensor<fp32, [128]> encoder_2_SDlayer_convs_1_bias = const()[name = tensor<string, []>("encoder_2_SDlayer_convs_1_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(711104)))];
            tensor<fp32, [128, 64, 4, 1]> encoder_2_SDlayer_convs_1_weight = const()[name = tensor<string, []>("encoder_2_SDlayer_convs_1_weight"), val = tensor<fp32, [128, 64, 4, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(711680)))];
            tensor<fp32, [128]> encoder_2_SDlayer_convs_2_bias = const()[name = tensor<string, []>("encoder_2_SDlayer_convs_2_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(842816)))];
            tensor<fp32, [128, 64, 16, 1]> encoder_2_SDlayer_convs_2_weight = const()[name = tensor<string, []>("encoder_2_SDlayer_convs_2_weight"), val = tensor<fp32, [128, 64, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(843392)))];
            tensor<fp32, [64]> encoder_2_conv_modules_0_layers_0_1_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_0_1_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1367744)))];
            tensor<fp32, [64, 128, 3]> encoder_2_conv_modules_0_layers_0_1_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_0_1_weight"), val = tensor<fp32, [64, 128, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1368064)))];
            tensor<fp32, [32]> encoder_2_conv_modules_0_layers_0_3_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_0_3_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1466432)))];
            tensor<fp32, [32, 1, 3]> encoder_2_conv_modules_0_layers_0_3_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_0_3_weight"), val = tensor<fp32, [32, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1466624)))];
            tensor<fp32, [128]> encoder_2_conv_modules_0_layers_0_6_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_0_6_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1467072)))];
            tensor<fp32, [128, 32, 1]> encoder_2_conv_modules_0_layers_0_6_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_0_6_weight"), val = tensor<fp32, [128, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1467648)))];
            tensor<fp32, [64]> encoder_2_conv_modules_0_layers_1_1_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_1_1_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1484096)))];
            tensor<fp32, [64, 128, 3]> encoder_2_conv_modules_0_layers_1_1_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_1_1_weight"), val = tensor<fp32, [64, 128, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1484416)))];
            tensor<fp32, [32]> encoder_2_conv_modules_0_layers_1_3_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_1_3_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1582784)))];
            tensor<fp32, [32, 1, 3]> encoder_2_conv_modules_0_layers_1_3_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_1_3_weight"), val = tensor<fp32, [32, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1582976)))];
            tensor<fp32, [128]> encoder_2_conv_modules_0_layers_1_6_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_1_6_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1583424)))];
            tensor<fp32, [128, 32, 1]> encoder_2_conv_modules_0_layers_1_6_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_1_6_weight"), val = tensor<fp32, [128, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1584000)))];
            tensor<fp32, [64]> encoder_2_conv_modules_0_layers_2_1_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_2_1_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1600448)))];
            tensor<fp32, [64, 128, 3]> encoder_2_conv_modules_0_layers_2_1_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_2_1_weight"), val = tensor<fp32, [64, 128, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1600768)))];
            tensor<fp32, [32]> encoder_2_conv_modules_0_layers_2_3_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_2_3_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1699136)))];
            tensor<fp32, [32, 1, 3]> encoder_2_conv_modules_0_layers_2_3_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_2_3_weight"), val = tensor<fp32, [32, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1699328)))];
            tensor<fp32, [128]> encoder_2_conv_modules_0_layers_2_6_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_2_6_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1699776)))];
            tensor<fp32, [128, 32, 1]> encoder_2_conv_modules_0_layers_2_6_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_0_layers_2_6_weight"), val = tensor<fp32, [128, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1700352)))];
            tensor<fp32, [64]> encoder_2_conv_modules_1_layers_0_1_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_0_1_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1716800)))];
            tensor<fp32, [64, 128, 3]> encoder_2_conv_modules_1_layers_0_1_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_0_1_weight"), val = tensor<fp32, [64, 128, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1717120)))];
            tensor<fp32, [32]> encoder_2_conv_modules_1_layers_0_3_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_0_3_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1815488)))];
            tensor<fp32, [32, 1, 3]> encoder_2_conv_modules_1_layers_0_3_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_0_3_weight"), val = tensor<fp32, [32, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1815680)))];
            tensor<fp32, [128]> encoder_2_conv_modules_1_layers_0_6_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_0_6_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1816128)))];
            tensor<fp32, [128, 32, 1]> encoder_2_conv_modules_1_layers_0_6_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_0_6_weight"), val = tensor<fp32, [128, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1816704)))];
            tensor<fp32, [64]> encoder_2_conv_modules_1_layers_1_1_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_1_1_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1833152)))];
            tensor<fp32, [64, 128, 3]> encoder_2_conv_modules_1_layers_1_1_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_1_1_weight"), val = tensor<fp32, [64, 128, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1833472)))];
            tensor<fp32, [32]> encoder_2_conv_modules_1_layers_1_3_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_1_3_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1931840)))];
            tensor<fp32, [32, 1, 3]> encoder_2_conv_modules_1_layers_1_3_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_1_3_weight"), val = tensor<fp32, [32, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1932032)))];
            tensor<fp32, [128]> encoder_2_conv_modules_1_layers_1_6_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_1_6_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1932480)))];
            tensor<fp32, [128, 32, 1]> encoder_2_conv_modules_1_layers_1_6_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_1_layers_1_6_weight"), val = tensor<fp32, [128, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1933056)))];
            tensor<fp32, [64]> encoder_2_conv_modules_2_layers_0_1_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_2_layers_0_1_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1949504)))];
            tensor<fp32, [64, 128, 3]> encoder_2_conv_modules_2_layers_0_1_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_2_layers_0_1_weight"), val = tensor<fp32, [64, 128, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(1949824)))];
            tensor<fp32, [32]> encoder_2_conv_modules_2_layers_0_3_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_2_layers_0_3_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2048192)))];
            tensor<fp32, [32, 1, 3]> encoder_2_conv_modules_2_layers_0_3_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_2_layers_0_3_weight"), val = tensor<fp32, [32, 1, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2048384)))];
            tensor<fp32, [128]> encoder_2_conv_modules_2_layers_0_6_bias = const()[name = tensor<string, []>("encoder_2_conv_modules_2_layers_0_6_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2048832)))];
            tensor<fp32, [128, 32, 1]> encoder_2_conv_modules_2_layers_0_6_weight = const()[name = tensor<string, []>("encoder_2_conv_modules_2_layers_0_6_weight"), val = tensor<fp32, [128, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2049408)))];
            tensor<fp32, [128]> encoder_2_globalconv_bias = const()[name = tensor<string, []>("encoder_2_globalconv_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2065856)))];
            tensor<fp32, [128, 128, 3, 3]> encoder_2_globalconv_weight = const()[name = tensor<string, []>("encoder_2_globalconv_weight"), val = tensor<fp32, [128, 128, 3, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2066432)))];
            tensor<fp32, [128]> separation_net_dp_modules_0_linear_layers_0_bias = const()[name = tensor<string, []>("separation_net_dp_modules_0_linear_layers_0_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2656320)))];
            tensor<fp32, [128, 256]> separation_net_dp_modules_0_linear_layers_0_weight = const()[name = tensor<string, []>("separation_net_dp_modules_0_linear_layers_0_weight"), val = tensor<fp32, [128, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2656896)))];
            tensor<fp32, [128]> separation_net_dp_modules_0_linear_layers_1_bias = const()[name = tensor<string, []>("separation_net_dp_modules_0_linear_layers_1_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2788032)))];
            tensor<fp32, [128, 256]> separation_net_dp_modules_0_linear_layers_1_weight = const()[name = tensor<string, []>("separation_net_dp_modules_0_linear_layers_1_weight"), val = tensor<fp32, [128, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2788608)))];
            tensor<fp32, [256]> separation_net_dp_modules_1_linear_layers_0_bias = const()[name = tensor<string, []>("separation_net_dp_modules_1_linear_layers_0_bias"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2919744)))];
            tensor<fp32, [256, 512]> separation_net_dp_modules_1_linear_layers_0_weight = const()[name = tensor<string, []>("separation_net_dp_modules_1_linear_layers_0_weight"), val = tensor<fp32, [256, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(2920832)))];
            tensor<fp32, [256]> separation_net_dp_modules_1_linear_layers_1_bias = const()[name = tensor<string, []>("separation_net_dp_modules_1_linear_layers_1_bias"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3445184)))];
            tensor<fp32, [256, 512]> separation_net_dp_modules_1_linear_layers_1_weight = const()[name = tensor<string, []>("separation_net_dp_modules_1_linear_layers_1_weight"), val = tensor<fp32, [256, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3446272)))];
            tensor<fp32, [128]> separation_net_dp_modules_2_linear_layers_0_bias = const()[name = tensor<string, []>("separation_net_dp_modules_2_linear_layers_0_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3970624)))];
            tensor<fp32, [128, 256]> separation_net_dp_modules_2_linear_layers_0_weight = const()[name = tensor<string, []>("separation_net_dp_modules_2_linear_layers_0_weight"), val = tensor<fp32, [128, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(3971200)))];
            tensor<fp32, [128]> separation_net_dp_modules_2_linear_layers_1_bias = const()[name = tensor<string, []>("separation_net_dp_modules_2_linear_layers_1_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4102336)))];
            tensor<fp32, [128, 256]> separation_net_dp_modules_2_linear_layers_1_weight = const()[name = tensor<string, []>("separation_net_dp_modules_2_linear_layers_1_weight"), val = tensor<fp32, [128, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4102912)))];
            tensor<fp32, [256]> separation_net_dp_modules_3_linear_layers_0_bias = const()[name = tensor<string, []>("separation_net_dp_modules_3_linear_layers_0_bias"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4234048)))];
            tensor<fp32, [256, 512]> separation_net_dp_modules_3_linear_layers_0_weight = const()[name = tensor<string, []>("separation_net_dp_modules_3_linear_layers_0_weight"), val = tensor<fp32, [256, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4235136)))];
            tensor<fp32, [256]> separation_net_dp_modules_3_linear_layers_1_bias = const()[name = tensor<string, []>("separation_net_dp_modules_3_linear_layers_1_bias"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4759488)))];
            tensor<fp32, [256, 512]> separation_net_dp_modules_3_linear_layers_1_weight = const()[name = tensor<string, []>("separation_net_dp_modules_3_linear_layers_1_weight"), val = tensor<fp32, [256, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(4760576)))];
            tensor<fp32, [128]> separation_net_dp_modules_4_linear_layers_0_bias = const()[name = tensor<string, []>("separation_net_dp_modules_4_linear_layers_0_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5284928)))];
            tensor<fp32, [128, 256]> separation_net_dp_modules_4_linear_layers_0_weight = const()[name = tensor<string, []>("separation_net_dp_modules_4_linear_layers_0_weight"), val = tensor<fp32, [128, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5285504)))];
            tensor<fp32, [128]> separation_net_dp_modules_4_linear_layers_1_bias = const()[name = tensor<string, []>("separation_net_dp_modules_4_linear_layers_1_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5416640)))];
            tensor<fp32, [128, 256]> separation_net_dp_modules_4_linear_layers_1_weight = const()[name = tensor<string, []>("separation_net_dp_modules_4_linear_layers_1_weight"), val = tensor<fp32, [128, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5417216)))];
            tensor<fp32, [256]> separation_net_dp_modules_5_linear_layers_0_bias = const()[name = tensor<string, []>("separation_net_dp_modules_5_linear_layers_0_bias"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5548352)))];
            tensor<fp32, [256, 512]> separation_net_dp_modules_5_linear_layers_0_weight = const()[name = tensor<string, []>("separation_net_dp_modules_5_linear_layers_0_weight"), val = tensor<fp32, [256, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(5549440)))];
            tensor<fp32, [256]> separation_net_dp_modules_5_linear_layers_1_bias = const()[name = tensor<string, []>("separation_net_dp_modules_5_linear_layers_1_bias"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6073792)))];
            tensor<fp32, [256, 512]> separation_net_dp_modules_5_linear_layers_1_weight = const()[name = tensor<string, []>("separation_net_dp_modules_5_linear_layers_1_weight"), val = tensor<fp32, [256, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6074880)))];
            tensor<fp32, [256]> decoder_0_0_conv_bias = const()[name = tensor<string, []>("decoder_0_0_conv_bias"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6599232)))];
            tensor<fp32, [256, 256, 3, 3]> decoder_0_0_conv_weight = const()[name = tensor<string, []>("decoder_0_0_conv_weight"), val = tensor<fp32, [256, 256, 3, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(6600320)))];
            tensor<fp32, [64]> decoder_0_1_convtrs_0_bias = const()[name = tensor<string, []>("decoder_0_1_convtrs_0_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(8959680)))];
            tensor<fp32, [128, 64, 3, 1]> decoder_0_1_convtrs_0_weight = const()[name = tensor<string, []>("decoder_0_1_convtrs_0_weight"), val = tensor<fp32, [128, 64, 3, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(8960000)))];
            tensor<fp32, [64]> decoder_0_1_convtrs_1_bias = const()[name = tensor<string, []>("decoder_0_1_convtrs_1_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9058368)))];
            tensor<fp32, [128, 64, 4, 1]> decoder_0_1_convtrs_1_weight = const()[name = tensor<string, []>("decoder_0_1_convtrs_1_weight"), val = tensor<fp32, [128, 64, 4, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9058688)))];
            tensor<fp32, [64]> decoder_0_1_convtrs_2_bias = const()[name = tensor<string, []>("decoder_0_1_convtrs_2_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9189824)))];
            tensor<fp32, [128, 64, 16, 1]> decoder_0_1_convtrs_2_weight = const()[name = tensor<string, []>("decoder_0_1_convtrs_2_weight"), val = tensor<fp32, [128, 64, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9190144)))];
            tensor<fp32, [128]> decoder_1_0_conv_bias = const()[name = tensor<string, []>("decoder_1_0_conv_bias"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9714496)))];
            tensor<fp32, [128, 128, 3, 3]> decoder_1_0_conv_weight = const()[name = tensor<string, []>("decoder_1_0_conv_weight"), val = tensor<fp32, [128, 128, 3, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(9715072)))];
            tensor<fp32, [32]> decoder_1_1_convtrs_0_bias = const()[name = tensor<string, []>("decoder_1_1_convtrs_0_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10304960)))];
            tensor<fp32, [64, 32, 3, 1]> decoder_1_1_convtrs_0_weight = const()[name = tensor<string, []>("decoder_1_1_convtrs_0_weight"), val = tensor<fp32, [64, 32, 3, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10305152)))];
            tensor<fp32, [32]> decoder_1_1_convtrs_1_bias = const()[name = tensor<string, []>("decoder_1_1_convtrs_1_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10329792)))];
            tensor<fp32, [64, 32, 4, 1]> decoder_1_1_convtrs_1_weight = const()[name = tensor<string, []>("decoder_1_1_convtrs_1_weight"), val = tensor<fp32, [64, 32, 4, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10329984)))];
            tensor<fp32, [32]> decoder_1_1_convtrs_2_bias = const()[name = tensor<string, []>("decoder_1_1_convtrs_2_bias"), val = tensor<fp32, [32]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10362816)))];
            tensor<fp32, [64, 32, 16, 1]> decoder_1_1_convtrs_2_weight = const()[name = tensor<string, []>("decoder_1_1_convtrs_2_weight"), val = tensor<fp32, [64, 32, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10363008)))];
            tensor<fp32, [64]> decoder_2_0_conv_bias = const()[name = tensor<string, []>("decoder_2_0_conv_bias"), val = tensor<fp32, [64]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10494144)))];
            tensor<fp32, [64, 64, 3, 3]> decoder_2_0_conv_weight = const()[name = tensor<string, []>("decoder_2_0_conv_weight"), val = tensor<fp32, [64, 64, 3, 3]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10494464)))];
            tensor<fp32, [16]> decoder_2_1_convtrs_0_bias = const()[name = tensor<string, []>("decoder_2_1_convtrs_0_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10641984)))];
            tensor<fp32, [32, 16, 3, 1]> decoder_2_1_convtrs_0_weight = const()[name = tensor<string, []>("decoder_2_1_convtrs_0_weight"), val = tensor<fp32, [32, 16, 3, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10642112)))];
            tensor<fp32, [16]> decoder_2_1_convtrs_1_bias = const()[name = tensor<string, []>("decoder_2_1_convtrs_1_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10648320)))];
            tensor<fp32, [32, 16, 4, 1]> decoder_2_1_convtrs_1_weight = const()[name = tensor<string, []>("decoder_2_1_convtrs_1_weight"), val = tensor<fp32, [32, 16, 4, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10648448)))];
            tensor<fp32, [16]> decoder_2_1_convtrs_2_bias = const()[name = tensor<string, []>("decoder_2_1_convtrs_2_bias"), val = tensor<fp32, [16]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10656704)))];
            tensor<fp32, [32, 16, 16, 1]> decoder_2_1_convtrs_2_weight = const()[name = tensor<string, []>("decoder_2_1_convtrs_2_weight"), val = tensor<fp32, [32, 16, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10656832)))];
            tensor<int32, [4]> var_64_begin_0 = const()[name = tensor<string, []>("op_64_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_64_end_0 = const()[name = tensor<string, []>("op_64_end_0"), val = tensor<int32, [4]>([1, 4, 359, 476])];
            tensor<bool, [4]> var_64_end_mask_0 = const()[name = tensor<string, []>("op_64_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 4, 359, 476]> var_64 = slice_by_index(begin = var_64_begin_0, end = var_64_end_0, end_mask = var_64_end_mask_0, x = spectrogram)[name = tensor<string, []>("op_64")];
            tensor<fp32, []> const_1 = const()[name = tensor<string, []>("const_1"), val = tensor<fp32, []>(0x0p+0)];
            tensor<int32, [8]> input_1_pad_0 = const()[name = tensor<string, []>("input_1_pad_0"), val = tensor<int32, [8]>([0, 0, 0, 0, 1, 1, 0, 0])];
            tensor<string, []> input_1_mode_0 = const()[name = tensor<string, []>("input_1_mode_0"), val = tensor<string, []>("constant")];
            tensor<fp32, [1, 4, 361, 476]> input_1 = pad(constant_val = const_1, mode = input_1_mode_0, pad = input_1_pad_0, x = var_64)[name = tensor<string, []>("input_1")];
            tensor<string, []> band_1_pad_type_0 = const()[name = tensor<string, []>("band_1_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_1_strides_0 = const()[name = tensor<string, []>("band_1_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [4]> band_1_pad_0 = const()[name = tensor<string, []>("band_1_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_1_dilations_0 = const()[name = tensor<string, []>("band_1_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_1_groups_0 = const()[name = tensor<string, []>("band_1_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 32, 359, 476]> band_1 = conv(bias = encoder_0_SDlayer_convs_0_bias, dilations = band_1_dilations_0, groups = band_1_groups_0, pad = band_1_pad_0, pad_type = band_1_pad_type_0, strides = band_1_strides_0, weight = encoder_0_SDlayer_convs_0_weight, x = input_1)[name = tensor<string, []>("band_1")];
            tensor<int32, [4]> var_77_begin_0 = const()[name = tensor<string, []>("op_77_begin_0"), val = tensor<int32, [4]>([0, 0, 359, 0])];
            tensor<int32, [4]> var_77_end_0 = const()[name = tensor<string, []>("op_77_end_0"), val = tensor<int32, [4]>([1, 4, 1162, 476])];
            tensor<bool, [4]> var_77_end_mask_0 = const()[name = tensor<string, []>("op_77_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 4, 803, 476]> var_77 = slice_by_index(begin = var_77_begin_0, end = var_77_end_0, end_mask = var_77_end_mask_0, x = spectrogram)[name = tensor<string, []>("op_77")];
            tensor<fp32, []> const_3 = const()[name = tensor<string, []>("const_3"), val = tensor<fp32, []>(0x0p+0)];
            tensor<int32, [8]> input_3_pad_0 = const()[name = tensor<string, []>("input_3_pad_0"), val = tensor<int32, [8]>([0, 0, 0, 0, 0, 1, 0, 0])];
            tensor<string, []> input_3_mode_0 = const()[name = tensor<string, []>("input_3_mode_0"), val = tensor<string, []>("constant")];
            tensor<fp32, [1, 4, 804, 476]> input_3 = pad(constant_val = const_3, mode = input_3_mode_0, pad = input_3_pad_0, x = var_77)[name = tensor<string, []>("input_3")];
            tensor<string, []> band_3_pad_type_0 = const()[name = tensor<string, []>("band_3_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_3_strides_0 = const()[name = tensor<string, []>("band_3_strides_0"), val = tensor<int32, [2]>([4, 1])];
            tensor<int32, [4]> band_3_pad_0 = const()[name = tensor<string, []>("band_3_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_3_dilations_0 = const()[name = tensor<string, []>("band_3_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_3_groups_0 = const()[name = tensor<string, []>("band_3_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 32, 201, 476]> band_3 = conv(bias = encoder_0_SDlayer_convs_1_bias, dilations = band_3_dilations_0, groups = band_3_groups_0, pad = band_3_pad_0, pad_type = band_3_pad_type_0, strides = band_3_strides_0, weight = encoder_0_SDlayer_convs_1_weight, x = input_3)[name = tensor<string, []>("band_3")];
            tensor<int32, [4]> var_99_begin_0 = const()[name = tensor<string, []>("op_99_begin_0"), val = tensor<int32, [4]>([0, 0, 1162, 0])];
            tensor<int32, [4]> var_99_end_0 = const()[name = tensor<string, []>("op_99_end_0"), val = tensor<int32, [4]>([1, 4, 1, 476])];
            tensor<bool, [4]> var_99_end_mask_0 = const()[name = tensor<string, []>("op_99_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 4, 887, 476]> var_99 = slice_by_index(begin = var_99_begin_0, end = var_99_end_0, end_mask = var_99_end_mask_0, x = spectrogram)[name = tensor<string, []>("op_99")];
            tensor<fp32, []> const_5 = const()[name = tensor<string, []>("const_5"), val = tensor<fp32, []>(0x0p+0)];
            tensor<int32, [8]> input_5_pad_0 = const()[name = tensor<string, []>("input_5_pad_0"), val = tensor<int32, [8]>([0, 0, 0, 0, 4, 5, 0, 0])];
            tensor<string, []> input_5_mode_0 = const()[name = tensor<string, []>("input_5_mode_0"), val = tensor<string, []>("constant")];
            tensor<fp32, [1, 4, 896, 476]> input_5 = pad(constant_val = const_5, mode = input_5_mode_0, pad = input_5_pad_0, x = var_99)[name = tensor<string, []>("input_5")];
            tensor<string, []> band_5_pad_type_0 = const()[name = tensor<string, []>("band_5_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_5_strides_0 = const()[name = tensor<string, []>("band_5_strides_0"), val = tensor<int32, [2]>([16, 1])];
            tensor<int32, [4]> band_5_pad_0 = const()[name = tensor<string, []>("band_5_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_5_dilations_0 = const()[name = tensor<string, []>("band_5_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_5_groups_0 = const()[name = tensor<string, []>("band_5_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 32, 56, 476]> band_5 = conv(bias = encoder_0_SDlayer_convs_2_bias, dilations = band_5_dilations_0, groups = band_5_groups_0, pad = band_5_pad_0, pad_type = band_5_pad_type_0, strides = band_5_strides_0, weight = encoder_0_SDlayer_convs_2_weight, x = input_5)[name = tensor<string, []>("band_5")];
            tensor<int32, [4]> var_126 = const()[name = tensor<string, []>("op_126"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_130 = const()[name = tensor<string, []>("op_130"), val = tensor<int32, [3]>([-1, 32, 476])];
            tensor<fp32, [1, 359, 32, 476]> var_127 = transpose(perm = var_126, x = band_1)[name = tensor<string, []>("transpose_122")];
            tensor<fp32, [359, 32, 476]> input_7 = reshape(shape = var_130, x = var_127)[name = tensor<string, []>("input_7")];
            tensor<int32, [4]> reshape_0_shape_0 = const()[name = tensor<string, []>("reshape_0_shape_0"), val = tensor<int32, [4]>([359, 1, 32, 476])];
            tensor<fp32, [359, 1, 32, 476]> reshape_0 = reshape(shape = reshape_0_shape_0, x = input_7)[name = tensor<string, []>("reshape_0")];
            tensor<int32, [2]> reduce_mean_0_axes_0 = const()[name = tensor<string, []>("reduce_mean_0_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_0_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_0_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_0 = reduce_mean(axes = reduce_mean_0_axes_0, keep_dims = reduce_mean_0_keep_dims_0, x = reshape_0)[name = tensor<string, []>("reduce_mean_0")];
            tensor<fp32, [359, 1, 32, 476]> sub_0 = sub(x = reshape_0, y = reduce_mean_0)[name = tensor<string, []>("sub_0")];
            tensor<fp32, [359, 1, 32, 476]> square_0 = square(x = sub_0)[name = tensor<string, []>("square_0")];
            tensor<int32, [2]> reduce_mean_2_axes_0 = const()[name = tensor<string, []>("reduce_mean_2_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_2_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_2_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_2 = reduce_mean(axes = reduce_mean_2_axes_0, keep_dims = reduce_mean_2_keep_dims_0, x = square_0)[name = tensor<string, []>("reduce_mean_2")];
            tensor<fp32, []> add_0_y_0 = const()[name = tensor<string, []>("add_0_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [359, 1, 1, 1]> add_0 = add(x = reduce_mean_2, y = add_0_y_0)[name = tensor<string, []>("add_0")];
            tensor<fp32, [359, 1, 1, 1]> sqrt_0 = sqrt(x = add_0)[name = tensor<string, []>("sqrt_0")];
            tensor<fp32, [359, 1, 32, 476]> real_div_0 = real_div(x = sub_0, y = sqrt_0)[name = tensor<string, []>("real_div_0")];
            tensor<int32, [3]> reshape_1_shape_0 = const()[name = tensor<string, []>("reshape_1_shape_0"), val = tensor<int32, [3]>([359, 32, 476])];
            tensor<fp32, [359, 32, 476]> reshape_1 = reshape(shape = reshape_1_shape_0, x = real_div_0)[name = tensor<string, []>("reshape_1")];
            tensor<fp32, [1, 32, 1]> reshape_2 = const()[name = tensor<string, []>("reshape_2"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10689664)))];
            tensor<fp32, [359, 32, 476]> mul_0 = mul(x = reshape_1, y = reshape_2)[name = tensor<string, []>("mul_0")];
            tensor<fp32, [1, 32, 1]> reshape_3 = const()[name = tensor<string, []>("reshape_3"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10689856)))];
            tensor<fp32, [359, 32, 476]> add_1 = add(x = mul_0, y = reshape_3)[name = tensor<string, []>("add_1")];
            tensor<string, []> input_11_pad_type_0 = const()[name = tensor<string, []>("input_11_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_11_pad_0 = const()[name = tensor<string, []>("input_11_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_11_strides_0 = const()[name = tensor<string, []>("input_11_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_11_dilations_0 = const()[name = tensor<string, []>("input_11_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_11_groups_0 = const()[name = tensor<string, []>("input_11_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 16, 476]> input_11 = conv(bias = encoder_0_conv_modules_0_layers_0_1_bias, dilations = input_11_dilations_0, groups = input_11_groups_0, pad = input_11_pad_0, pad_type = input_11_pad_type_0, strides = input_11_strides_0, weight = encoder_0_conv_modules_0_layers_0_1_weight, x = add_1)[name = tensor<string, []>("input_11")];
            tensor<int32, []> input_13_split_num_splits_0 = const()[name = tensor<string, []>("input_13_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_13_split_axis_0 = const()[name = tensor<string, []>("input_13_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 8, 476]> input_13_split_0, tensor<fp32, [359, 8, 476]> input_13_split_1 = split(axis = input_13_split_axis_0, num_splits = input_13_split_num_splits_0, x = input_11)[name = tensor<string, []>("input_13_split")];
            tensor<fp32, [359, 8, 476]> input_13_split_1_sigmoid = sigmoid(x = input_13_split_1)[name = tensor<string, []>("input_13_split_1_sigmoid")];
            tensor<fp32, [359, 8, 476]> input_13 = mul(x = input_13_split_0, y = input_13_split_1_sigmoid)[name = tensor<string, []>("input_13")];
            tensor<string, []> input_15_pad_type_0 = const()[name = tensor<string, []>("input_15_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_15_pad_0 = const()[name = tensor<string, []>("input_15_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_15_groups_0 = const()[name = tensor<string, []>("input_15_groups_0"), val = tensor<int32, []>(8)];
            tensor<int32, [1]> input_15_strides_0 = const()[name = tensor<string, []>("input_15_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_15_dilations_0 = const()[name = tensor<string, []>("input_15_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [359, 8, 476]> input_15 = conv(bias = encoder_0_conv_modules_0_layers_0_3_bias, dilations = input_15_dilations_0, groups = input_15_groups_0, pad = input_15_pad_0, pad_type = input_15_pad_type_0, strides = input_15_strides_0, weight = encoder_0_conv_modules_0_layers_0_3_weight, x = input_13)[name = tensor<string, []>("input_15")];
            tensor<int32, [4]> reshape_4_shape_0 = const()[name = tensor<string, []>("reshape_4_shape_0"), val = tensor<int32, [4]>([359, 1, 8, 476])];
            tensor<fp32, [359, 1, 8, 476]> reshape_4 = reshape(shape = reshape_4_shape_0, x = input_15)[name = tensor<string, []>("reshape_4")];
            tensor<int32, [2]> reduce_mean_3_axes_0 = const()[name = tensor<string, []>("reduce_mean_3_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_3_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_3_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_3 = reduce_mean(axes = reduce_mean_3_axes_0, keep_dims = reduce_mean_3_keep_dims_0, x = reshape_4)[name = tensor<string, []>("reduce_mean_3")];
            tensor<fp32, [359, 1, 8, 476]> sub_2 = sub(x = reshape_4, y = reduce_mean_3)[name = tensor<string, []>("sub_2")];
            tensor<fp32, [359, 1, 8, 476]> square_1 = square(x = sub_2)[name = tensor<string, []>("square_1")];
            tensor<int32, [2]> reduce_mean_5_axes_0 = const()[name = tensor<string, []>("reduce_mean_5_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_5_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_5_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_5 = reduce_mean(axes = reduce_mean_5_axes_0, keep_dims = reduce_mean_5_keep_dims_0, x = square_1)[name = tensor<string, []>("reduce_mean_5")];
            tensor<fp32, []> add_2_y_0 = const()[name = tensor<string, []>("add_2_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [359, 1, 1, 1]> add_2 = add(x = reduce_mean_5, y = add_2_y_0)[name = tensor<string, []>("add_2")];
            tensor<fp32, [359, 1, 1, 1]> sqrt_1 = sqrt(x = add_2)[name = tensor<string, []>("sqrt_1")];
            tensor<fp32, [359, 1, 8, 476]> real_div_1 = real_div(x = sub_2, y = sqrt_1)[name = tensor<string, []>("real_div_1")];
            tensor<int32, [3]> reshape_5_shape_0 = const()[name = tensor<string, []>("reshape_5_shape_0"), val = tensor<int32, [3]>([359, 8, 476])];
            tensor<fp32, [359, 8, 476]> reshape_5 = reshape(shape = reshape_5_shape_0, x = real_div_1)[name = tensor<string, []>("reshape_5")];
            tensor<fp32, [1, 8, 1]> reshape_6 = const()[name = tensor<string, []>("reshape_6"), val = tensor<fp32, [1, 8, 1]>([[[0x1.d749e4p-1], [0x1.a170d6p-1], [0x1.8c3592p-1], [0x1.a00808p-1], [0x1.9279aap-1], [0x1.464208p-1], [0x1.798c02p-1], [0x1.6897a4p-1]]])];
            tensor<fp32, [359, 8, 476]> mul_1 = mul(x = reshape_5, y = reshape_6)[name = tensor<string, []>("mul_1")];
            tensor<fp32, [1, 8, 1]> reshape_7 = const()[name = tensor<string, []>("reshape_7"), val = tensor<fp32, [1, 8, 1]>([[[-0x1.79a6d4p-2], [0x1.9debdep-6], [-0x1.5937dep-3], [-0x1.8033d6p-4], [-0x1.70316ap-3], [-0x1.77a928p-2], [-0x1.e9893ep-3], [-0x1.d0af9ep-3]]])];
            tensor<fp32, [359, 8, 476]> add_3 = add(x = mul_1, y = reshape_7)[name = tensor<string, []>("add_3")];
            tensor<fp32, [359, 8, 476]> input_17 = silu(x = add_3)[name = tensor<string, []>("input_17")];
            tensor<string, []> var_172_pad_type_0 = const()[name = tensor<string, []>("op_172_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_172_strides_0 = const()[name = tensor<string, []>("op_172_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_172_pad_0 = const()[name = tensor<string, []>("op_172_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_172_dilations_0 = const()[name = tensor<string, []>("op_172_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_172_groups_0 = const()[name = tensor<string, []>("op_172_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 32, 476]> var_172 = conv(bias = encoder_0_conv_modules_0_layers_0_6_bias, dilations = var_172_dilations_0, groups = var_172_groups_0, pad = var_172_pad_0, pad_type = var_172_pad_type_0, strides = var_172_strides_0, weight = encoder_0_conv_modules_0_layers_0_6_weight, x = input_17)[name = tensor<string, []>("op_172")];
            tensor<fp32, [359, 32, 476]> input_19 = add(x = input_7, y = var_172)[name = tensor<string, []>("input_19")];
            tensor<int32, [4]> reshape_8_shape_0 = const()[name = tensor<string, []>("reshape_8_shape_0"), val = tensor<int32, [4]>([359, 1, 32, 476])];
            tensor<fp32, [359, 1, 32, 476]> reshape_8 = reshape(shape = reshape_8_shape_0, x = input_19)[name = tensor<string, []>("reshape_8")];
            tensor<int32, [2]> reduce_mean_6_axes_0 = const()[name = tensor<string, []>("reduce_mean_6_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_6_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_6_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_6 = reduce_mean(axes = reduce_mean_6_axes_0, keep_dims = reduce_mean_6_keep_dims_0, x = reshape_8)[name = tensor<string, []>("reduce_mean_6")];
            tensor<fp32, [359, 1, 32, 476]> sub_4 = sub(x = reshape_8, y = reduce_mean_6)[name = tensor<string, []>("sub_4")];
            tensor<fp32, [359, 1, 32, 476]> square_2 = square(x = sub_4)[name = tensor<string, []>("square_2")];
            tensor<int32, [2]> reduce_mean_8_axes_0 = const()[name = tensor<string, []>("reduce_mean_8_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_8_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_8_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_8 = reduce_mean(axes = reduce_mean_8_axes_0, keep_dims = reduce_mean_8_keep_dims_0, x = square_2)[name = tensor<string, []>("reduce_mean_8")];
            tensor<fp32, []> add_4_y_0 = const()[name = tensor<string, []>("add_4_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [359, 1, 1, 1]> add_4 = add(x = reduce_mean_8, y = add_4_y_0)[name = tensor<string, []>("add_4")];
            tensor<fp32, [359, 1, 1, 1]> sqrt_2 = sqrt(x = add_4)[name = tensor<string, []>("sqrt_2")];
            tensor<fp32, [359, 1, 32, 476]> real_div_2 = real_div(x = sub_4, y = sqrt_2)[name = tensor<string, []>("real_div_2")];
            tensor<int32, [3]> reshape_9_shape_0 = const()[name = tensor<string, []>("reshape_9_shape_0"), val = tensor<int32, [3]>([359, 32, 476])];
            tensor<fp32, [359, 32, 476]> reshape_9 = reshape(shape = reshape_9_shape_0, x = real_div_2)[name = tensor<string, []>("reshape_9")];
            tensor<fp32, [1, 32, 1]> reshape_10 = const()[name = tensor<string, []>("reshape_10"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10690048)))];
            tensor<fp32, [359, 32, 476]> mul_2 = mul(x = reshape_9, y = reshape_10)[name = tensor<string, []>("mul_2")];
            tensor<fp32, [1, 32, 1]> reshape_11 = const()[name = tensor<string, []>("reshape_11"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10690240)))];
            tensor<fp32, [359, 32, 476]> add_5 = add(x = mul_2, y = reshape_11)[name = tensor<string, []>("add_5")];
            tensor<string, []> input_23_pad_type_0 = const()[name = tensor<string, []>("input_23_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_23_pad_0 = const()[name = tensor<string, []>("input_23_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_23_strides_0 = const()[name = tensor<string, []>("input_23_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_23_dilations_0 = const()[name = tensor<string, []>("input_23_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_23_groups_0 = const()[name = tensor<string, []>("input_23_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 16, 476]> input_23 = conv(bias = encoder_0_conv_modules_0_layers_1_1_bias, dilations = input_23_dilations_0, groups = input_23_groups_0, pad = input_23_pad_0, pad_type = input_23_pad_type_0, strides = input_23_strides_0, weight = encoder_0_conv_modules_0_layers_1_1_weight, x = add_5)[name = tensor<string, []>("input_23")];
            tensor<int32, []> input_25_split_num_splits_0 = const()[name = tensor<string, []>("input_25_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_25_split_axis_0 = const()[name = tensor<string, []>("input_25_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 8, 476]> input_25_split_0, tensor<fp32, [359, 8, 476]> input_25_split_1 = split(axis = input_25_split_axis_0, num_splits = input_25_split_num_splits_0, x = input_23)[name = tensor<string, []>("input_25_split")];
            tensor<fp32, [359, 8, 476]> input_25_split_1_sigmoid = sigmoid(x = input_25_split_1)[name = tensor<string, []>("input_25_split_1_sigmoid")];
            tensor<fp32, [359, 8, 476]> input_25 = mul(x = input_25_split_0, y = input_25_split_1_sigmoid)[name = tensor<string, []>("input_25")];
            tensor<string, []> input_27_pad_type_0 = const()[name = tensor<string, []>("input_27_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_27_pad_0 = const()[name = tensor<string, []>("input_27_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_27_groups_0 = const()[name = tensor<string, []>("input_27_groups_0"), val = tensor<int32, []>(8)];
            tensor<int32, [1]> input_27_strides_0 = const()[name = tensor<string, []>("input_27_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_27_dilations_0 = const()[name = tensor<string, []>("input_27_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [359, 8, 476]> input_27 = conv(bias = encoder_0_conv_modules_0_layers_1_3_bias, dilations = input_27_dilations_0, groups = input_27_groups_0, pad = input_27_pad_0, pad_type = input_27_pad_type_0, strides = input_27_strides_0, weight = encoder_0_conv_modules_0_layers_1_3_weight, x = input_25)[name = tensor<string, []>("input_27")];
            tensor<int32, [4]> reshape_12_shape_0 = const()[name = tensor<string, []>("reshape_12_shape_0"), val = tensor<int32, [4]>([359, 1, 8, 476])];
            tensor<fp32, [359, 1, 8, 476]> reshape_12 = reshape(shape = reshape_12_shape_0, x = input_27)[name = tensor<string, []>("reshape_12")];
            tensor<int32, [2]> reduce_mean_9_axes_0 = const()[name = tensor<string, []>("reduce_mean_9_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_9_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_9_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_9 = reduce_mean(axes = reduce_mean_9_axes_0, keep_dims = reduce_mean_9_keep_dims_0, x = reshape_12)[name = tensor<string, []>("reduce_mean_9")];
            tensor<fp32, [359, 1, 8, 476]> sub_6 = sub(x = reshape_12, y = reduce_mean_9)[name = tensor<string, []>("sub_6")];
            tensor<fp32, [359, 1, 8, 476]> square_3 = square(x = sub_6)[name = tensor<string, []>("square_3")];
            tensor<int32, [2]> reduce_mean_11_axes_0 = const()[name = tensor<string, []>("reduce_mean_11_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_11_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_11_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_11 = reduce_mean(axes = reduce_mean_11_axes_0, keep_dims = reduce_mean_11_keep_dims_0, x = square_3)[name = tensor<string, []>("reduce_mean_11")];
            tensor<fp32, []> add_6_y_0 = const()[name = tensor<string, []>("add_6_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [359, 1, 1, 1]> add_6 = add(x = reduce_mean_11, y = add_6_y_0)[name = tensor<string, []>("add_6")];
            tensor<fp32, [359, 1, 1, 1]> sqrt_3 = sqrt(x = add_6)[name = tensor<string, []>("sqrt_3")];
            tensor<fp32, [359, 1, 8, 476]> real_div_3 = real_div(x = sub_6, y = sqrt_3)[name = tensor<string, []>("real_div_3")];
            tensor<int32, [3]> reshape_13_shape_0 = const()[name = tensor<string, []>("reshape_13_shape_0"), val = tensor<int32, [3]>([359, 8, 476])];
            tensor<fp32, [359, 8, 476]> reshape_13 = reshape(shape = reshape_13_shape_0, x = real_div_3)[name = tensor<string, []>("reshape_13")];
            tensor<fp32, [1, 8, 1]> reshape_14 = const()[name = tensor<string, []>("reshape_14"), val = tensor<fp32, [1, 8, 1]>([[[0x1.15e904p+0], [0x1.a85accp-1], [0x1.e218acp-1], [0x1.ab9fc6p-1], [0x1.1e523p-1], [0x1.dbf454p-1], [0x1.4792d6p+0], [0x1.571724p-1]]])];
            tensor<fp32, [359, 8, 476]> mul_3 = mul(x = reshape_13, y = reshape_14)[name = tensor<string, []>("mul_3")];
            tensor<fp32, [1, 8, 1]> reshape_15 = const()[name = tensor<string, []>("reshape_15"), val = tensor<fp32, [1, 8, 1]>([[[-0x1.80317ap-2], [-0x1.595584p-2], [-0x1.3ad58ep-2], [-0x1.135bf6p+1], [-0x1.18971cp-3], [-0x1.3db6eap-3], [-0x1.dc0498p-2], [-0x1.31a3f8p-4]]])];
            tensor<fp32, [359, 8, 476]> add_7 = add(x = mul_3, y = reshape_15)[name = tensor<string, []>("add_7")];
            tensor<fp32, [359, 8, 476]> input_29 = silu(x = add_7)[name = tensor<string, []>("input_29")];
            tensor<string, []> var_208_pad_type_0 = const()[name = tensor<string, []>("op_208_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_208_strides_0 = const()[name = tensor<string, []>("op_208_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_208_pad_0 = const()[name = tensor<string, []>("op_208_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_208_dilations_0 = const()[name = tensor<string, []>("op_208_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_208_groups_0 = const()[name = tensor<string, []>("op_208_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 32, 476]> var_208 = conv(bias = encoder_0_conv_modules_0_layers_1_6_bias, dilations = var_208_dilations_0, groups = var_208_groups_0, pad = var_208_pad_0, pad_type = var_208_pad_type_0, strides = var_208_strides_0, weight = encoder_0_conv_modules_0_layers_1_6_weight, x = input_29)[name = tensor<string, []>("op_208")];
            tensor<fp32, [359, 32, 476]> input_31 = add(x = input_19, y = var_208)[name = tensor<string, []>("input_31")];
            tensor<int32, [4]> reshape_16_shape_0 = const()[name = tensor<string, []>("reshape_16_shape_0"), val = tensor<int32, [4]>([359, 1, 32, 476])];
            tensor<fp32, [359, 1, 32, 476]> reshape_16 = reshape(shape = reshape_16_shape_0, x = input_31)[name = tensor<string, []>("reshape_16")];
            tensor<int32, [2]> reduce_mean_12_axes_0 = const()[name = tensor<string, []>("reduce_mean_12_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_12_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_12_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_12 = reduce_mean(axes = reduce_mean_12_axes_0, keep_dims = reduce_mean_12_keep_dims_0, x = reshape_16)[name = tensor<string, []>("reduce_mean_12")];
            tensor<fp32, [359, 1, 32, 476]> sub_8 = sub(x = reshape_16, y = reduce_mean_12)[name = tensor<string, []>("sub_8")];
            tensor<fp32, [359, 1, 32, 476]> square_4 = square(x = sub_8)[name = tensor<string, []>("square_4")];
            tensor<int32, [2]> reduce_mean_14_axes_0 = const()[name = tensor<string, []>("reduce_mean_14_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_14_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_14_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_14 = reduce_mean(axes = reduce_mean_14_axes_0, keep_dims = reduce_mean_14_keep_dims_0, x = square_4)[name = tensor<string, []>("reduce_mean_14")];
            tensor<fp32, []> add_8_y_0 = const()[name = tensor<string, []>("add_8_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [359, 1, 1, 1]> add_8 = add(x = reduce_mean_14, y = add_8_y_0)[name = tensor<string, []>("add_8")];
            tensor<fp32, [359, 1, 1, 1]> sqrt_4 = sqrt(x = add_8)[name = tensor<string, []>("sqrt_4")];
            tensor<fp32, [359, 1, 32, 476]> real_div_4 = real_div(x = sub_8, y = sqrt_4)[name = tensor<string, []>("real_div_4")];
            tensor<int32, [3]> reshape_17_shape_0 = const()[name = tensor<string, []>("reshape_17_shape_0"), val = tensor<int32, [3]>([359, 32, 476])];
            tensor<fp32, [359, 32, 476]> reshape_17 = reshape(shape = reshape_17_shape_0, x = real_div_4)[name = tensor<string, []>("reshape_17")];
            tensor<fp32, [1, 32, 1]> reshape_18 = const()[name = tensor<string, []>("reshape_18"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10690432)))];
            tensor<fp32, [359, 32, 476]> mul_4 = mul(x = reshape_17, y = reshape_18)[name = tensor<string, []>("mul_4")];
            tensor<fp32, [1, 32, 1]> reshape_19 = const()[name = tensor<string, []>("reshape_19"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10690624)))];
            tensor<fp32, [359, 32, 476]> add_9 = add(x = mul_4, y = reshape_19)[name = tensor<string, []>("add_9")];
            tensor<string, []> input_35_pad_type_0 = const()[name = tensor<string, []>("input_35_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_35_pad_0 = const()[name = tensor<string, []>("input_35_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_35_strides_0 = const()[name = tensor<string, []>("input_35_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_35_dilations_0 = const()[name = tensor<string, []>("input_35_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_35_groups_0 = const()[name = tensor<string, []>("input_35_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 16, 476]> input_35 = conv(bias = encoder_0_conv_modules_0_layers_2_1_bias, dilations = input_35_dilations_0, groups = input_35_groups_0, pad = input_35_pad_0, pad_type = input_35_pad_type_0, strides = input_35_strides_0, weight = encoder_0_conv_modules_0_layers_2_1_weight, x = add_9)[name = tensor<string, []>("input_35")];
            tensor<int32, []> input_37_split_num_splits_0 = const()[name = tensor<string, []>("input_37_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_37_split_axis_0 = const()[name = tensor<string, []>("input_37_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 8, 476]> input_37_split_0, tensor<fp32, [359, 8, 476]> input_37_split_1 = split(axis = input_37_split_axis_0, num_splits = input_37_split_num_splits_0, x = input_35)[name = tensor<string, []>("input_37_split")];
            tensor<fp32, [359, 8, 476]> input_37_split_1_sigmoid = sigmoid(x = input_37_split_1)[name = tensor<string, []>("input_37_split_1_sigmoid")];
            tensor<fp32, [359, 8, 476]> input_37 = mul(x = input_37_split_0, y = input_37_split_1_sigmoid)[name = tensor<string, []>("input_37")];
            tensor<string, []> input_39_pad_type_0 = const()[name = tensor<string, []>("input_39_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_39_pad_0 = const()[name = tensor<string, []>("input_39_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_39_groups_0 = const()[name = tensor<string, []>("input_39_groups_0"), val = tensor<int32, []>(8)];
            tensor<int32, [1]> input_39_strides_0 = const()[name = tensor<string, []>("input_39_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_39_dilations_0 = const()[name = tensor<string, []>("input_39_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [359, 8, 476]> input_39 = conv(bias = encoder_0_conv_modules_0_layers_2_3_bias, dilations = input_39_dilations_0, groups = input_39_groups_0, pad = input_39_pad_0, pad_type = input_39_pad_type_0, strides = input_39_strides_0, weight = encoder_0_conv_modules_0_layers_2_3_weight, x = input_37)[name = tensor<string, []>("input_39")];
            tensor<int32, [4]> reshape_20_shape_0 = const()[name = tensor<string, []>("reshape_20_shape_0"), val = tensor<int32, [4]>([359, 1, 8, 476])];
            tensor<fp32, [359, 1, 8, 476]> reshape_20 = reshape(shape = reshape_20_shape_0, x = input_39)[name = tensor<string, []>("reshape_20")];
            tensor<int32, [2]> reduce_mean_15_axes_0 = const()[name = tensor<string, []>("reduce_mean_15_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_15_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_15_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_15 = reduce_mean(axes = reduce_mean_15_axes_0, keep_dims = reduce_mean_15_keep_dims_0, x = reshape_20)[name = tensor<string, []>("reduce_mean_15")];
            tensor<fp32, [359, 1, 8, 476]> sub_10 = sub(x = reshape_20, y = reduce_mean_15)[name = tensor<string, []>("sub_10")];
            tensor<fp32, [359, 1, 8, 476]> square_5 = square(x = sub_10)[name = tensor<string, []>("square_5")];
            tensor<int32, [2]> reduce_mean_17_axes_0 = const()[name = tensor<string, []>("reduce_mean_17_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_17_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_17_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [359, 1, 1, 1]> reduce_mean_17 = reduce_mean(axes = reduce_mean_17_axes_0, keep_dims = reduce_mean_17_keep_dims_0, x = square_5)[name = tensor<string, []>("reduce_mean_17")];
            tensor<fp32, []> add_10_y_0 = const()[name = tensor<string, []>("add_10_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [359, 1, 1, 1]> add_10 = add(x = reduce_mean_17, y = add_10_y_0)[name = tensor<string, []>("add_10")];
            tensor<fp32, [359, 1, 1, 1]> sqrt_5 = sqrt(x = add_10)[name = tensor<string, []>("sqrt_5")];
            tensor<fp32, [359, 1, 8, 476]> real_div_5 = real_div(x = sub_10, y = sqrt_5)[name = tensor<string, []>("real_div_5")];
            tensor<int32, [3]> reshape_21_shape_0 = const()[name = tensor<string, []>("reshape_21_shape_0"), val = tensor<int32, [3]>([359, 8, 476])];
            tensor<fp32, [359, 8, 476]> reshape_21 = reshape(shape = reshape_21_shape_0, x = real_div_5)[name = tensor<string, []>("reshape_21")];
            tensor<fp32, [1, 8, 1]> reshape_22 = const()[name = tensor<string, []>("reshape_22"), val = tensor<fp32, [1, 8, 1]>([[[0x1.327c06p+0], [0x1.a9a414p-1], [0x1.b269cp-1], [0x1.426d8cp+0], [0x1.6fbf44p-1], [0x1.2482eep+0], [0x1.78498p-1], [0x1.3c737ep-1]]])];
            tensor<fp32, [359, 8, 476]> mul_5 = mul(x = reshape_21, y = reshape_22)[name = tensor<string, []>("mul_5")];
            tensor<fp32, [1, 8, 1]> reshape_23 = const()[name = tensor<string, []>("reshape_23"), val = tensor<fp32, [1, 8, 1]>([[[-0x1.1c4bb4p-2], [-0x1.4f59b6p-3], [-0x1.c8e2b8p-2], [-0x1.6c1f34p+0], [-0x1.15a7cp-2], [-0x1.799fe6p-1], [-0x1.291302p-4], [-0x1.1dbc7p-3]]])];
            tensor<fp32, [359, 8, 476]> add_11 = add(x = mul_5, y = reshape_23)[name = tensor<string, []>("add_11")];
            tensor<fp32, [359, 8, 476]> input_41 = silu(x = add_11)[name = tensor<string, []>("input_41")];
            tensor<string, []> var_244_pad_type_0 = const()[name = tensor<string, []>("op_244_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_244_strides_0 = const()[name = tensor<string, []>("op_244_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_244_pad_0 = const()[name = tensor<string, []>("op_244_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_244_dilations_0 = const()[name = tensor<string, []>("op_244_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_244_groups_0 = const()[name = tensor<string, []>("op_244_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [359, 32, 476]> var_244 = conv(bias = encoder_0_conv_modules_0_layers_2_6_bias, dilations = var_244_dilations_0, groups = var_244_groups_0, pad = var_244_pad_0, pad_type = var_244_pad_type_0, strides = var_244_strides_0, weight = encoder_0_conv_modules_0_layers_2_6_weight, x = input_41)[name = tensor<string, []>("op_244")];
            tensor<fp32, [359, 32, 476]> var_245 = add(x = input_31, y = var_244)[name = tensor<string, []>("op_245")];
            tensor<int32, [4]> var_250 = const()[name = tensor<string, []>("op_250"), val = tensor<int32, [4]>([1, 359, 32, 476])];
            tensor<fp32, [1, 359, 32, 476]> var_251 = reshape(shape = var_250, x = var_245)[name = tensor<string, []>("op_251")];
            tensor<string, []> band_7_mode_0 = const()[name = tensor<string, []>("band_7_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 359, 32, 476]> band_7 = gelu(mode = band_7_mode_0, x = var_251)[name = tensor<string, []>("band_7")];
            tensor<int32, [4]> var_255 = const()[name = tensor<string, []>("op_255"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_259 = const()[name = tensor<string, []>("op_259"), val = tensor<int32, [3]>([-1, 32, 476])];
            tensor<fp32, [1, 201, 32, 476]> var_256 = transpose(perm = var_255, x = band_3)[name = tensor<string, []>("transpose_121")];
            tensor<fp32, [201, 32, 476]> input_43 = reshape(shape = var_259, x = var_256)[name = tensor<string, []>("input_43")];
            tensor<int32, [4]> reshape_24_shape_0 = const()[name = tensor<string, []>("reshape_24_shape_0"), val = tensor<int32, [4]>([201, 1, 32, 476])];
            tensor<fp32, [201, 1, 32, 476]> reshape_24 = reshape(shape = reshape_24_shape_0, x = input_43)[name = tensor<string, []>("reshape_24")];
            tensor<int32, [2]> reduce_mean_18_axes_0 = const()[name = tensor<string, []>("reduce_mean_18_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_18_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_18_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [201, 1, 1, 1]> reduce_mean_18 = reduce_mean(axes = reduce_mean_18_axes_0, keep_dims = reduce_mean_18_keep_dims_0, x = reshape_24)[name = tensor<string, []>("reduce_mean_18")];
            tensor<fp32, [201, 1, 32, 476]> sub_12 = sub(x = reshape_24, y = reduce_mean_18)[name = tensor<string, []>("sub_12")];
            tensor<fp32, [201, 1, 32, 476]> square_6 = square(x = sub_12)[name = tensor<string, []>("square_6")];
            tensor<int32, [2]> reduce_mean_20_axes_0 = const()[name = tensor<string, []>("reduce_mean_20_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_20_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_20_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [201, 1, 1, 1]> reduce_mean_20 = reduce_mean(axes = reduce_mean_20_axes_0, keep_dims = reduce_mean_20_keep_dims_0, x = square_6)[name = tensor<string, []>("reduce_mean_20")];
            tensor<fp32, []> add_12_y_0 = const()[name = tensor<string, []>("add_12_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [201, 1, 1, 1]> add_12 = add(x = reduce_mean_20, y = add_12_y_0)[name = tensor<string, []>("add_12")];
            tensor<fp32, [201, 1, 1, 1]> sqrt_6 = sqrt(x = add_12)[name = tensor<string, []>("sqrt_6")];
            tensor<fp32, [201, 1, 32, 476]> real_div_6 = real_div(x = sub_12, y = sqrt_6)[name = tensor<string, []>("real_div_6")];
            tensor<int32, [3]> reshape_25_shape_0 = const()[name = tensor<string, []>("reshape_25_shape_0"), val = tensor<int32, [3]>([201, 32, 476])];
            tensor<fp32, [201, 32, 476]> reshape_25 = reshape(shape = reshape_25_shape_0, x = real_div_6)[name = tensor<string, []>("reshape_25")];
            tensor<fp32, [1, 32, 1]> reshape_26 = const()[name = tensor<string, []>("reshape_26"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10690816)))];
            tensor<fp32, [201, 32, 476]> mul_6 = mul(x = reshape_25, y = reshape_26)[name = tensor<string, []>("mul_6")];
            tensor<fp32, [1, 32, 1]> reshape_27 = const()[name = tensor<string, []>("reshape_27"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10691008)))];
            tensor<fp32, [201, 32, 476]> add_13 = add(x = mul_6, y = reshape_27)[name = tensor<string, []>("add_13")];
            tensor<string, []> input_47_pad_type_0 = const()[name = tensor<string, []>("input_47_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_47_pad_0 = const()[name = tensor<string, []>("input_47_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_47_strides_0 = const()[name = tensor<string, []>("input_47_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_47_dilations_0 = const()[name = tensor<string, []>("input_47_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_47_groups_0 = const()[name = tensor<string, []>("input_47_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [201, 16, 476]> input_47 = conv(bias = encoder_0_conv_modules_1_layers_0_1_bias, dilations = input_47_dilations_0, groups = input_47_groups_0, pad = input_47_pad_0, pad_type = input_47_pad_type_0, strides = input_47_strides_0, weight = encoder_0_conv_modules_1_layers_0_1_weight, x = add_13)[name = tensor<string, []>("input_47")];
            tensor<int32, []> input_49_split_num_splits_0 = const()[name = tensor<string, []>("input_49_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_49_split_axis_0 = const()[name = tensor<string, []>("input_49_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [201, 8, 476]> input_49_split_0, tensor<fp32, [201, 8, 476]> input_49_split_1 = split(axis = input_49_split_axis_0, num_splits = input_49_split_num_splits_0, x = input_47)[name = tensor<string, []>("input_49_split")];
            tensor<fp32, [201, 8, 476]> input_49_split_1_sigmoid = sigmoid(x = input_49_split_1)[name = tensor<string, []>("input_49_split_1_sigmoid")];
            tensor<fp32, [201, 8, 476]> input_49 = mul(x = input_49_split_0, y = input_49_split_1_sigmoid)[name = tensor<string, []>("input_49")];
            tensor<string, []> input_51_pad_type_0 = const()[name = tensor<string, []>("input_51_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_51_pad_0 = const()[name = tensor<string, []>("input_51_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_51_groups_0 = const()[name = tensor<string, []>("input_51_groups_0"), val = tensor<int32, []>(8)];
            tensor<int32, [1]> input_51_strides_0 = const()[name = tensor<string, []>("input_51_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_51_dilations_0 = const()[name = tensor<string, []>("input_51_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [201, 8, 476]> input_51 = conv(bias = encoder_0_conv_modules_1_layers_0_3_bias, dilations = input_51_dilations_0, groups = input_51_groups_0, pad = input_51_pad_0, pad_type = input_51_pad_type_0, strides = input_51_strides_0, weight = encoder_0_conv_modules_1_layers_0_3_weight, x = input_49)[name = tensor<string, []>("input_51")];
            tensor<int32, [4]> reshape_28_shape_0 = const()[name = tensor<string, []>("reshape_28_shape_0"), val = tensor<int32, [4]>([201, 1, 8, 476])];
            tensor<fp32, [201, 1, 8, 476]> reshape_28 = reshape(shape = reshape_28_shape_0, x = input_51)[name = tensor<string, []>("reshape_28")];
            tensor<int32, [2]> reduce_mean_21_axes_0 = const()[name = tensor<string, []>("reduce_mean_21_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_21_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_21_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [201, 1, 1, 1]> reduce_mean_21 = reduce_mean(axes = reduce_mean_21_axes_0, keep_dims = reduce_mean_21_keep_dims_0, x = reshape_28)[name = tensor<string, []>("reduce_mean_21")];
            tensor<fp32, [201, 1, 8, 476]> sub_14 = sub(x = reshape_28, y = reduce_mean_21)[name = tensor<string, []>("sub_14")];
            tensor<fp32, [201, 1, 8, 476]> square_7 = square(x = sub_14)[name = tensor<string, []>("square_7")];
            tensor<int32, [2]> reduce_mean_23_axes_0 = const()[name = tensor<string, []>("reduce_mean_23_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_23_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_23_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [201, 1, 1, 1]> reduce_mean_23 = reduce_mean(axes = reduce_mean_23_axes_0, keep_dims = reduce_mean_23_keep_dims_0, x = square_7)[name = tensor<string, []>("reduce_mean_23")];
            tensor<fp32, []> add_14_y_0 = const()[name = tensor<string, []>("add_14_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [201, 1, 1, 1]> add_14 = add(x = reduce_mean_23, y = add_14_y_0)[name = tensor<string, []>("add_14")];
            tensor<fp32, [201, 1, 1, 1]> sqrt_7 = sqrt(x = add_14)[name = tensor<string, []>("sqrt_7")];
            tensor<fp32, [201, 1, 8, 476]> real_div_7 = real_div(x = sub_14, y = sqrt_7)[name = tensor<string, []>("real_div_7")];
            tensor<int32, [3]> reshape_29_shape_0 = const()[name = tensor<string, []>("reshape_29_shape_0"), val = tensor<int32, [3]>([201, 8, 476])];
            tensor<fp32, [201, 8, 476]> reshape_29 = reshape(shape = reshape_29_shape_0, x = real_div_7)[name = tensor<string, []>("reshape_29")];
            tensor<fp32, [1, 8, 1]> reshape_30 = const()[name = tensor<string, []>("reshape_30"), val = tensor<fp32, [1, 8, 1]>([[[0x1.b7a4ap+0], [0x1.370ea4p+0], [0x1.1d7cap+0], [0x1.696ad8p+0], [0x1.5fa726p+0], [0x1.630bcap+0], [0x1.a0ee66p-1], [0x1.f6654ep-1]]])];
            tensor<fp32, [201, 8, 476]> mul_7 = mul(x = reshape_29, y = reshape_30)[name = tensor<string, []>("mul_7")];
            tensor<fp32, [1, 8, 1]> reshape_31 = const()[name = tensor<string, []>("reshape_31"), val = tensor<fp32, [1, 8, 1]>([[[0x1.945caap-2], [0x1.e79b96p-4], [0x1.1a9aep-6], [0x1.26dbf2p-2], [-0x1.f69aacp-5], [0x1.951d88p-6], [0x1.8a74d6p-4], [0x1.fd1982p-4]]])];
            tensor<fp32, [201, 8, 476]> add_15 = add(x = mul_7, y = reshape_31)[name = tensor<string, []>("add_15")];
            tensor<fp32, [201, 8, 476]> input_53 = silu(x = add_15)[name = tensor<string, []>("input_53")];
            tensor<string, []> var_299_pad_type_0 = const()[name = tensor<string, []>("op_299_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_299_strides_0 = const()[name = tensor<string, []>("op_299_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_299_pad_0 = const()[name = tensor<string, []>("op_299_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_299_dilations_0 = const()[name = tensor<string, []>("op_299_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_299_groups_0 = const()[name = tensor<string, []>("op_299_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [201, 32, 476]> var_299 = conv(bias = encoder_0_conv_modules_1_layers_0_6_bias, dilations = var_299_dilations_0, groups = var_299_groups_0, pad = var_299_pad_0, pad_type = var_299_pad_type_0, strides = var_299_strides_0, weight = encoder_0_conv_modules_1_layers_0_6_weight, x = input_53)[name = tensor<string, []>("op_299")];
            tensor<fp32, [201, 32, 476]> input_55 = add(x = input_43, y = var_299)[name = tensor<string, []>("input_55")];
            tensor<int32, [4]> reshape_32_shape_0 = const()[name = tensor<string, []>("reshape_32_shape_0"), val = tensor<int32, [4]>([201, 1, 32, 476])];
            tensor<fp32, [201, 1, 32, 476]> reshape_32 = reshape(shape = reshape_32_shape_0, x = input_55)[name = tensor<string, []>("reshape_32")];
            tensor<int32, [2]> reduce_mean_24_axes_0 = const()[name = tensor<string, []>("reduce_mean_24_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_24_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_24_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [201, 1, 1, 1]> reduce_mean_24 = reduce_mean(axes = reduce_mean_24_axes_0, keep_dims = reduce_mean_24_keep_dims_0, x = reshape_32)[name = tensor<string, []>("reduce_mean_24")];
            tensor<fp32, [201, 1, 32, 476]> sub_16 = sub(x = reshape_32, y = reduce_mean_24)[name = tensor<string, []>("sub_16")];
            tensor<fp32, [201, 1, 32, 476]> square_8 = square(x = sub_16)[name = tensor<string, []>("square_8")];
            tensor<int32, [2]> reduce_mean_26_axes_0 = const()[name = tensor<string, []>("reduce_mean_26_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_26_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_26_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [201, 1, 1, 1]> reduce_mean_26 = reduce_mean(axes = reduce_mean_26_axes_0, keep_dims = reduce_mean_26_keep_dims_0, x = square_8)[name = tensor<string, []>("reduce_mean_26")];
            tensor<fp32, []> add_16_y_0 = const()[name = tensor<string, []>("add_16_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [201, 1, 1, 1]> add_16 = add(x = reduce_mean_26, y = add_16_y_0)[name = tensor<string, []>("add_16")];
            tensor<fp32, [201, 1, 1, 1]> sqrt_8 = sqrt(x = add_16)[name = tensor<string, []>("sqrt_8")];
            tensor<fp32, [201, 1, 32, 476]> real_div_8 = real_div(x = sub_16, y = sqrt_8)[name = tensor<string, []>("real_div_8")];
            tensor<int32, [3]> reshape_33_shape_0 = const()[name = tensor<string, []>("reshape_33_shape_0"), val = tensor<int32, [3]>([201, 32, 476])];
            tensor<fp32, [201, 32, 476]> reshape_33 = reshape(shape = reshape_33_shape_0, x = real_div_8)[name = tensor<string, []>("reshape_33")];
            tensor<fp32, [1, 32, 1]> reshape_34 = const()[name = tensor<string, []>("reshape_34"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10691200)))];
            tensor<fp32, [201, 32, 476]> mul_8 = mul(x = reshape_33, y = reshape_34)[name = tensor<string, []>("mul_8")];
            tensor<fp32, [1, 32, 1]> reshape_35 = const()[name = tensor<string, []>("reshape_35"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10691392)))];
            tensor<fp32, [201, 32, 476]> add_17 = add(x = mul_8, y = reshape_35)[name = tensor<string, []>("add_17")];
            tensor<string, []> input_59_pad_type_0 = const()[name = tensor<string, []>("input_59_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_59_pad_0 = const()[name = tensor<string, []>("input_59_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_59_strides_0 = const()[name = tensor<string, []>("input_59_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_59_dilations_0 = const()[name = tensor<string, []>("input_59_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_59_groups_0 = const()[name = tensor<string, []>("input_59_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [201, 16, 476]> input_59 = conv(bias = encoder_0_conv_modules_1_layers_1_1_bias, dilations = input_59_dilations_0, groups = input_59_groups_0, pad = input_59_pad_0, pad_type = input_59_pad_type_0, strides = input_59_strides_0, weight = encoder_0_conv_modules_1_layers_1_1_weight, x = add_17)[name = tensor<string, []>("input_59")];
            tensor<int32, []> input_61_split_num_splits_0 = const()[name = tensor<string, []>("input_61_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_61_split_axis_0 = const()[name = tensor<string, []>("input_61_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [201, 8, 476]> input_61_split_0, tensor<fp32, [201, 8, 476]> input_61_split_1 = split(axis = input_61_split_axis_0, num_splits = input_61_split_num_splits_0, x = input_59)[name = tensor<string, []>("input_61_split")];
            tensor<fp32, [201, 8, 476]> input_61_split_1_sigmoid = sigmoid(x = input_61_split_1)[name = tensor<string, []>("input_61_split_1_sigmoid")];
            tensor<fp32, [201, 8, 476]> input_61 = mul(x = input_61_split_0, y = input_61_split_1_sigmoid)[name = tensor<string, []>("input_61")];
            tensor<string, []> input_63_pad_type_0 = const()[name = tensor<string, []>("input_63_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_63_pad_0 = const()[name = tensor<string, []>("input_63_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_63_groups_0 = const()[name = tensor<string, []>("input_63_groups_0"), val = tensor<int32, []>(8)];
            tensor<int32, [1]> input_63_strides_0 = const()[name = tensor<string, []>("input_63_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_63_dilations_0 = const()[name = tensor<string, []>("input_63_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [201, 8, 476]> input_63 = conv(bias = encoder_0_conv_modules_1_layers_1_3_bias, dilations = input_63_dilations_0, groups = input_63_groups_0, pad = input_63_pad_0, pad_type = input_63_pad_type_0, strides = input_63_strides_0, weight = encoder_0_conv_modules_1_layers_1_3_weight, x = input_61)[name = tensor<string, []>("input_63")];
            tensor<int32, [4]> reshape_36_shape_0 = const()[name = tensor<string, []>("reshape_36_shape_0"), val = tensor<int32, [4]>([201, 1, 8, 476])];
            tensor<fp32, [201, 1, 8, 476]> reshape_36 = reshape(shape = reshape_36_shape_0, x = input_63)[name = tensor<string, []>("reshape_36")];
            tensor<int32, [2]> reduce_mean_27_axes_0 = const()[name = tensor<string, []>("reduce_mean_27_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_27_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_27_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [201, 1, 1, 1]> reduce_mean_27 = reduce_mean(axes = reduce_mean_27_axes_0, keep_dims = reduce_mean_27_keep_dims_0, x = reshape_36)[name = tensor<string, []>("reduce_mean_27")];
            tensor<fp32, [201, 1, 8, 476]> sub_18 = sub(x = reshape_36, y = reduce_mean_27)[name = tensor<string, []>("sub_18")];
            tensor<fp32, [201, 1, 8, 476]> square_9 = square(x = sub_18)[name = tensor<string, []>("square_9")];
            tensor<int32, [2]> reduce_mean_29_axes_0 = const()[name = tensor<string, []>("reduce_mean_29_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_29_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_29_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [201, 1, 1, 1]> reduce_mean_29 = reduce_mean(axes = reduce_mean_29_axes_0, keep_dims = reduce_mean_29_keep_dims_0, x = square_9)[name = tensor<string, []>("reduce_mean_29")];
            tensor<fp32, []> add_18_y_0 = const()[name = tensor<string, []>("add_18_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [201, 1, 1, 1]> add_18 = add(x = reduce_mean_29, y = add_18_y_0)[name = tensor<string, []>("add_18")];
            tensor<fp32, [201, 1, 1, 1]> sqrt_9 = sqrt(x = add_18)[name = tensor<string, []>("sqrt_9")];
            tensor<fp32, [201, 1, 8, 476]> real_div_9 = real_div(x = sub_18, y = sqrt_9)[name = tensor<string, []>("real_div_9")];
            tensor<int32, [3]> reshape_37_shape_0 = const()[name = tensor<string, []>("reshape_37_shape_0"), val = tensor<int32, [3]>([201, 8, 476])];
            tensor<fp32, [201, 8, 476]> reshape_37 = reshape(shape = reshape_37_shape_0, x = real_div_9)[name = tensor<string, []>("reshape_37")];
            tensor<fp32, [1, 8, 1]> reshape_38 = const()[name = tensor<string, []>("reshape_38"), val = tensor<fp32, [1, 8, 1]>([[[0x1.944b56p+0], [0x1.f129d6p-1], [0x1.e0497p+0], [0x1.b7c6a4p+0], [0x1.33e068p+0], [0x1.1a6cfep+0], [0x1.64584cp+0], [0x1.8b734ep-1]]])];
            tensor<fp32, [201, 8, 476]> mul_9 = mul(x = reshape_37, y = reshape_38)[name = tensor<string, []>("mul_9")];
            tensor<fp32, [1, 8, 1]> reshape_39 = const()[name = tensor<string, []>("reshape_39"), val = tensor<fp32, [1, 8, 1]>([[[-0x1.06ecdp-1], [0x1.6ef122p-4], [0x1.5752aap-4], [0x1.62885p-2], [0x1.0c423ep-2], [0x1.f2773cp-5], [0x1.df8f2ap-4], [-0x1.eb1174p-3]]])];
            tensor<fp32, [201, 8, 476]> add_19 = add(x = mul_9, y = reshape_39)[name = tensor<string, []>("add_19")];
            tensor<fp32, [201, 8, 476]> input_65 = silu(x = add_19)[name = tensor<string, []>("input_65")];
            tensor<string, []> var_335_pad_type_0 = const()[name = tensor<string, []>("op_335_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_335_strides_0 = const()[name = tensor<string, []>("op_335_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_335_pad_0 = const()[name = tensor<string, []>("op_335_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_335_dilations_0 = const()[name = tensor<string, []>("op_335_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_335_groups_0 = const()[name = tensor<string, []>("op_335_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [201, 32, 476]> var_335 = conv(bias = encoder_0_conv_modules_1_layers_1_6_bias, dilations = var_335_dilations_0, groups = var_335_groups_0, pad = var_335_pad_0, pad_type = var_335_pad_type_0, strides = var_335_strides_0, weight = encoder_0_conv_modules_1_layers_1_6_weight, x = input_65)[name = tensor<string, []>("op_335")];
            tensor<fp32, [201, 32, 476]> var_336 = add(x = input_55, y = var_335)[name = tensor<string, []>("op_336")];
            tensor<int32, [4]> var_341 = const()[name = tensor<string, []>("op_341"), val = tensor<int32, [4]>([1, 201, 32, 476])];
            tensor<fp32, [1, 201, 32, 476]> var_342 = reshape(shape = var_341, x = var_336)[name = tensor<string, []>("op_342")];
            tensor<string, []> band_9_mode_0 = const()[name = tensor<string, []>("band_9_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 201, 32, 476]> band_9 = gelu(mode = band_9_mode_0, x = var_342)[name = tensor<string, []>("band_9")];
            tensor<int32, [4]> var_346 = const()[name = tensor<string, []>("op_346"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_350 = const()[name = tensor<string, []>("op_350"), val = tensor<int32, [3]>([-1, 32, 476])];
            tensor<fp32, [1, 56, 32, 476]> var_347 = transpose(perm = var_346, x = band_5)[name = tensor<string, []>("transpose_120")];
            tensor<fp32, [56, 32, 476]> input_67 = reshape(shape = var_350, x = var_347)[name = tensor<string, []>("input_67")];
            tensor<int32, [4]> reshape_40_shape_0 = const()[name = tensor<string, []>("reshape_40_shape_0"), val = tensor<int32, [4]>([56, 1, 32, 476])];
            tensor<fp32, [56, 1, 32, 476]> reshape_40 = reshape(shape = reshape_40_shape_0, x = input_67)[name = tensor<string, []>("reshape_40")];
            tensor<int32, [2]> reduce_mean_30_axes_0 = const()[name = tensor<string, []>("reduce_mean_30_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_30_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_30_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [56, 1, 1, 1]> reduce_mean_30 = reduce_mean(axes = reduce_mean_30_axes_0, keep_dims = reduce_mean_30_keep_dims_0, x = reshape_40)[name = tensor<string, []>("reduce_mean_30")];
            tensor<fp32, [56, 1, 32, 476]> sub_20 = sub(x = reshape_40, y = reduce_mean_30)[name = tensor<string, []>("sub_20")];
            tensor<fp32, [56, 1, 32, 476]> square_10 = square(x = sub_20)[name = tensor<string, []>("square_10")];
            tensor<int32, [2]> reduce_mean_32_axes_0 = const()[name = tensor<string, []>("reduce_mean_32_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_32_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_32_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [56, 1, 1, 1]> reduce_mean_32 = reduce_mean(axes = reduce_mean_32_axes_0, keep_dims = reduce_mean_32_keep_dims_0, x = square_10)[name = tensor<string, []>("reduce_mean_32")];
            tensor<fp32, []> add_20_y_0 = const()[name = tensor<string, []>("add_20_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [56, 1, 1, 1]> add_20 = add(x = reduce_mean_32, y = add_20_y_0)[name = tensor<string, []>("add_20")];
            tensor<fp32, [56, 1, 1, 1]> sqrt_10 = sqrt(x = add_20)[name = tensor<string, []>("sqrt_10")];
            tensor<fp32, [56, 1, 32, 476]> real_div_10 = real_div(x = sub_20, y = sqrt_10)[name = tensor<string, []>("real_div_10")];
            tensor<int32, [3]> reshape_41_shape_0 = const()[name = tensor<string, []>("reshape_41_shape_0"), val = tensor<int32, [3]>([56, 32, 476])];
            tensor<fp32, [56, 32, 476]> reshape_41 = reshape(shape = reshape_41_shape_0, x = real_div_10)[name = tensor<string, []>("reshape_41")];
            tensor<fp32, [1, 32, 1]> reshape_42 = const()[name = tensor<string, []>("reshape_42"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10691584)))];
            tensor<fp32, [56, 32, 476]> mul_10 = mul(x = reshape_41, y = reshape_42)[name = tensor<string, []>("mul_10")];
            tensor<fp32, [1, 32, 1]> reshape_43 = const()[name = tensor<string, []>("reshape_43"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10691776)))];
            tensor<fp32, [56, 32, 476]> add_21 = add(x = mul_10, y = reshape_43)[name = tensor<string, []>("add_21")];
            tensor<string, []> input_71_pad_type_0 = const()[name = tensor<string, []>("input_71_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_71_pad_0 = const()[name = tensor<string, []>("input_71_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_71_strides_0 = const()[name = tensor<string, []>("input_71_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_71_dilations_0 = const()[name = tensor<string, []>("input_71_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_71_groups_0 = const()[name = tensor<string, []>("input_71_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [56, 16, 476]> input_71 = conv(bias = encoder_0_conv_modules_2_layers_0_1_bias, dilations = input_71_dilations_0, groups = input_71_groups_0, pad = input_71_pad_0, pad_type = input_71_pad_type_0, strides = input_71_strides_0, weight = encoder_0_conv_modules_2_layers_0_1_weight, x = add_21)[name = tensor<string, []>("input_71")];
            tensor<int32, []> input_73_split_num_splits_0 = const()[name = tensor<string, []>("input_73_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_73_split_axis_0 = const()[name = tensor<string, []>("input_73_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [56, 8, 476]> input_73_split_0, tensor<fp32, [56, 8, 476]> input_73_split_1 = split(axis = input_73_split_axis_0, num_splits = input_73_split_num_splits_0, x = input_71)[name = tensor<string, []>("input_73_split")];
            tensor<fp32, [56, 8, 476]> input_73_split_1_sigmoid = sigmoid(x = input_73_split_1)[name = tensor<string, []>("input_73_split_1_sigmoid")];
            tensor<fp32, [56, 8, 476]> input_73 = mul(x = input_73_split_0, y = input_73_split_1_sigmoid)[name = tensor<string, []>("input_73")];
            tensor<string, []> input_75_pad_type_0 = const()[name = tensor<string, []>("input_75_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_75_pad_0 = const()[name = tensor<string, []>("input_75_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_75_groups_0 = const()[name = tensor<string, []>("input_75_groups_0"), val = tensor<int32, []>(8)];
            tensor<int32, [1]> input_75_strides_0 = const()[name = tensor<string, []>("input_75_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_75_dilations_0 = const()[name = tensor<string, []>("input_75_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [56, 8, 476]> input_75 = conv(bias = encoder_0_conv_modules_2_layers_0_3_bias, dilations = input_75_dilations_0, groups = input_75_groups_0, pad = input_75_pad_0, pad_type = input_75_pad_type_0, strides = input_75_strides_0, weight = encoder_0_conv_modules_2_layers_0_3_weight, x = input_73)[name = tensor<string, []>("input_75")];
            tensor<int32, [4]> reshape_44_shape_0 = const()[name = tensor<string, []>("reshape_44_shape_0"), val = tensor<int32, [4]>([56, 1, 8, 476])];
            tensor<fp32, [56, 1, 8, 476]> reshape_44 = reshape(shape = reshape_44_shape_0, x = input_75)[name = tensor<string, []>("reshape_44")];
            tensor<int32, [2]> reduce_mean_33_axes_0 = const()[name = tensor<string, []>("reduce_mean_33_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_33_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_33_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [56, 1, 1, 1]> reduce_mean_33 = reduce_mean(axes = reduce_mean_33_axes_0, keep_dims = reduce_mean_33_keep_dims_0, x = reshape_44)[name = tensor<string, []>("reduce_mean_33")];
            tensor<fp32, [56, 1, 8, 476]> sub_22 = sub(x = reshape_44, y = reduce_mean_33)[name = tensor<string, []>("sub_22")];
            tensor<fp32, [56, 1, 8, 476]> square_11 = square(x = sub_22)[name = tensor<string, []>("square_11")];
            tensor<int32, [2]> reduce_mean_35_axes_0 = const()[name = tensor<string, []>("reduce_mean_35_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_35_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_35_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [56, 1, 1, 1]> reduce_mean_35 = reduce_mean(axes = reduce_mean_35_axes_0, keep_dims = reduce_mean_35_keep_dims_0, x = square_11)[name = tensor<string, []>("reduce_mean_35")];
            tensor<fp32, []> add_22_y_0 = const()[name = tensor<string, []>("add_22_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [56, 1, 1, 1]> add_22 = add(x = reduce_mean_35, y = add_22_y_0)[name = tensor<string, []>("add_22")];
            tensor<fp32, [56, 1, 1, 1]> sqrt_11 = sqrt(x = add_22)[name = tensor<string, []>("sqrt_11")];
            tensor<fp32, [56, 1, 8, 476]> real_div_11 = real_div(x = sub_22, y = sqrt_11)[name = tensor<string, []>("real_div_11")];
            tensor<int32, [3]> reshape_45_shape_0 = const()[name = tensor<string, []>("reshape_45_shape_0"), val = tensor<int32, [3]>([56, 8, 476])];
            tensor<fp32, [56, 8, 476]> reshape_45 = reshape(shape = reshape_45_shape_0, x = real_div_11)[name = tensor<string, []>("reshape_45")];
            tensor<fp32, [1, 8, 1]> reshape_46 = const()[name = tensor<string, []>("reshape_46"), val = tensor<fp32, [1, 8, 1]>([[[0x1.f8876p+0], [0x1.c38bfcp-1], [0x1.139e7p+1], [0x1.08c908p+1], [0x1.80fd78p+0], [0x1.49032p-1], [0x1.a10cfcp-1], [0x1.257298p+0]]])];
            tensor<fp32, [56, 8, 476]> mul_11 = mul(x = reshape_45, y = reshape_46)[name = tensor<string, []>("mul_11")];
            tensor<fp32, [1, 8, 1]> reshape_47 = const()[name = tensor<string, []>("reshape_47"), val = tensor<fp32, [1, 8, 1]>([[[-0x1.0f224ap-4], [-0x1.cf2bfp-5], [0x1.21714cp-1], [0x1.65f2eep-1], [0x1.bbb08ep-3], [0x1.49baep-3], [-0x1.3ece0ap-1], [0x1.0da09cp-1]]])];
            tensor<fp32, [56, 8, 476]> add_23 = add(x = mul_11, y = reshape_47)[name = tensor<string, []>("add_23")];
            tensor<fp32, [56, 8, 476]> input_77 = silu(x = add_23)[name = tensor<string, []>("input_77")];
            tensor<string, []> var_388_pad_type_0 = const()[name = tensor<string, []>("op_388_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_388_strides_0 = const()[name = tensor<string, []>("op_388_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_388_pad_0 = const()[name = tensor<string, []>("op_388_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_388_dilations_0 = const()[name = tensor<string, []>("op_388_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_388_groups_0 = const()[name = tensor<string, []>("op_388_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [56, 32, 476]> var_388 = conv(bias = encoder_0_conv_modules_2_layers_0_6_bias, dilations = var_388_dilations_0, groups = var_388_groups_0, pad = var_388_pad_0, pad_type = var_388_pad_type_0, strides = var_388_strides_0, weight = encoder_0_conv_modules_2_layers_0_6_weight, x = input_77)[name = tensor<string, []>("op_388")];
            tensor<fp32, [56, 32, 476]> var_389 = add(x = input_67, y = var_388)[name = tensor<string, []>("op_389")];
            tensor<int32, [4]> var_394 = const()[name = tensor<string, []>("op_394"), val = tensor<int32, [4]>([1, 56, 32, 476])];
            tensor<fp32, [1, 56, 32, 476]> var_395 = reshape(shape = var_394, x = var_389)[name = tensor<string, []>("op_395")];
            tensor<string, []> band_11_mode_0 = const()[name = tensor<string, []>("band_11_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 56, 32, 476]> band_11 = gelu(mode = band_11_mode_0, x = var_395)[name = tensor<string, []>("band_11")];
            tensor<bool, []> input_79_interleave_0 = const()[name = tensor<string, []>("input_79_interleave_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> const_129 = const()[name = tensor<string, []>("const_129"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 616, 32, 476]> input_79 = concat(axis = const_129, interleave = input_79_interleave_0, values = (band_7, band_9, band_11))[name = tensor<string, []>("input_79")];
            tensor<string, []> x_15_pad_type_0 = const()[name = tensor<string, []>("x_15_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [4]> x_15_pad_0 = const()[name = tensor<string, []>("x_15_pad_0"), val = tensor<int32, [4]>([1, 1, 1, 1])];
            tensor<int32, [2]> x_15_strides_0 = const()[name = tensor<string, []>("x_15_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> x_15_dilations_0 = const()[name = tensor<string, []>("x_15_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> x_15_groups_0 = const()[name = tensor<string, []>("x_15_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> transpose_24_perm_0 = const()[name = tensor<string, []>("transpose_24_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 32, 616, 476]> transpose_24 = transpose(perm = transpose_24_perm_0, x = input_79)[name = tensor<string, []>("transpose_119")];
            tensor<fp32, [1, 32, 616, 476]> x_15 = conv(bias = encoder_0_globalconv_bias, dilations = x_15_dilations_0, groups = x_15_groups_0, pad = x_15_pad_0, pad_type = x_15_pad_type_0, strides = x_15_strides_0, weight = encoder_0_globalconv_weight, x = transpose_24)[name = tensor<string, []>("x_15")];
            tensor<int32, [4]> var_456_begin_0 = const()[name = tensor<string, []>("op_456_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_456_end_0 = const()[name = tensor<string, []>("op_456_end_0"), val = tensor<int32, [4]>([1, 32, 108, 476])];
            tensor<bool, [4]> var_456_end_mask_0 = const()[name = tensor<string, []>("op_456_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 32, 108, 476]> var_456 = slice_by_index(begin = var_456_begin_0, end = var_456_end_0, end_mask = var_456_end_mask_0, x = x_15)[name = tensor<string, []>("op_456")];
            tensor<fp32, []> const_27 = const()[name = tensor<string, []>("const_27"), val = tensor<fp32, []>(0x0p+0)];
            tensor<int32, [8]> input_81_pad_0 = const()[name = tensor<string, []>("input_81_pad_0"), val = tensor<int32, [8]>([0, 0, 0, 0, 1, 1, 0, 0])];
            tensor<string, []> input_81_mode_0 = const()[name = tensor<string, []>("input_81_mode_0"), val = tensor<string, []>("constant")];
            tensor<fp32, [1, 32, 110, 476]> input_81 = pad(constant_val = const_27, mode = input_81_mode_0, pad = input_81_pad_0, x = var_456)[name = tensor<string, []>("input_81")];
            tensor<string, []> band_13_pad_type_0 = const()[name = tensor<string, []>("band_13_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_13_strides_0 = const()[name = tensor<string, []>("band_13_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [4]> band_13_pad_0 = const()[name = tensor<string, []>("band_13_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_13_dilations_0 = const()[name = tensor<string, []>("band_13_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_13_groups_0 = const()[name = tensor<string, []>("band_13_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 64, 108, 476]> band_13 = conv(bias = encoder_1_SDlayer_convs_0_bias, dilations = band_13_dilations_0, groups = band_13_groups_0, pad = band_13_pad_0, pad_type = band_13_pad_type_0, strides = band_13_strides_0, weight = encoder_1_SDlayer_convs_0_weight, x = input_81)[name = tensor<string, []>("band_13")];
            tensor<int32, [4]> var_469_begin_0 = const()[name = tensor<string, []>("op_469_begin_0"), val = tensor<int32, [4]>([0, 0, 108, 0])];
            tensor<int32, [4]> var_469_end_0 = const()[name = tensor<string, []>("op_469_end_0"), val = tensor<int32, [4]>([1, 32, 350, 476])];
            tensor<bool, [4]> var_469_end_mask_0 = const()[name = tensor<string, []>("op_469_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 32, 242, 476]> var_469 = slice_by_index(begin = var_469_begin_0, end = var_469_end_0, end_mask = var_469_end_mask_0, x = x_15)[name = tensor<string, []>("op_469")];
            tensor<fp32, []> const_29 = const()[name = tensor<string, []>("const_29"), val = tensor<fp32, []>(0x0p+0)];
            tensor<int32, [8]> input_83_pad_0 = const()[name = tensor<string, []>("input_83_pad_0"), val = tensor<int32, [8]>([0, 0, 0, 0, 1, 1, 0, 0])];
            tensor<string, []> input_83_mode_0 = const()[name = tensor<string, []>("input_83_mode_0"), val = tensor<string, []>("constant")];
            tensor<fp32, [1, 32, 244, 476]> input_83 = pad(constant_val = const_29, mode = input_83_mode_0, pad = input_83_pad_0, x = var_469)[name = tensor<string, []>("input_83")];
            tensor<string, []> band_15_pad_type_0 = const()[name = tensor<string, []>("band_15_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_15_strides_0 = const()[name = tensor<string, []>("band_15_strides_0"), val = tensor<int32, [2]>([4, 1])];
            tensor<int32, [4]> band_15_pad_0 = const()[name = tensor<string, []>("band_15_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_15_dilations_0 = const()[name = tensor<string, []>("band_15_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_15_groups_0 = const()[name = tensor<string, []>("band_15_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 64, 61, 476]> band_15 = conv(bias = encoder_1_SDlayer_convs_1_bias, dilations = band_15_dilations_0, groups = band_15_groups_0, pad = band_15_pad_0, pad_type = band_15_pad_type_0, strides = band_15_strides_0, weight = encoder_1_SDlayer_convs_1_weight, x = input_83)[name = tensor<string, []>("band_15")];
            tensor<int32, [4]> var_491_begin_0 = const()[name = tensor<string, []>("op_491_begin_0"), val = tensor<int32, [4]>([0, 0, 350, 0])];
            tensor<int32, [4]> var_491_end_0 = const()[name = tensor<string, []>("op_491_end_0"), val = tensor<int32, [4]>([1, 32, 1, 476])];
            tensor<bool, [4]> var_491_end_mask_0 = const()[name = tensor<string, []>("op_491_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 32, 266, 476]> var_491 = slice_by_index(begin = var_491_begin_0, end = var_491_end_0, end_mask = var_491_end_mask_0, x = x_15)[name = tensor<string, []>("op_491")];
            tensor<fp32, []> const_31 = const()[name = tensor<string, []>("const_31"), val = tensor<fp32, []>(0x0p+0)];
            tensor<int32, [8]> input_85_pad_0 = const()[name = tensor<string, []>("input_85_pad_0"), val = tensor<int32, [8]>([0, 0, 0, 0, 3, 3, 0, 0])];
            tensor<string, []> input_85_mode_0 = const()[name = tensor<string, []>("input_85_mode_0"), val = tensor<string, []>("constant")];
            tensor<fp32, [1, 32, 272, 476]> input_85 = pad(constant_val = const_31, mode = input_85_mode_0, pad = input_85_pad_0, x = var_491)[name = tensor<string, []>("input_85")];
            tensor<string, []> band_17_pad_type_0 = const()[name = tensor<string, []>("band_17_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_17_strides_0 = const()[name = tensor<string, []>("band_17_strides_0"), val = tensor<int32, [2]>([16, 1])];
            tensor<int32, [4]> band_17_pad_0 = const()[name = tensor<string, []>("band_17_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_17_dilations_0 = const()[name = tensor<string, []>("band_17_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_17_groups_0 = const()[name = tensor<string, []>("band_17_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 64, 17, 476]> band_17 = conv(bias = encoder_1_SDlayer_convs_2_bias, dilations = band_17_dilations_0, groups = band_17_groups_0, pad = band_17_pad_0, pad_type = band_17_pad_type_0, strides = band_17_strides_0, weight = encoder_1_SDlayer_convs_2_weight, x = input_85)[name = tensor<string, []>("band_17")];
            tensor<int32, [4]> var_518 = const()[name = tensor<string, []>("op_518"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_522 = const()[name = tensor<string, []>("op_522"), val = tensor<int32, [3]>([-1, 64, 476])];
            tensor<fp32, [1, 108, 64, 476]> var_519 = transpose(perm = var_518, x = band_13)[name = tensor<string, []>("transpose_118")];
            tensor<fp32, [108, 64, 476]> input_87 = reshape(shape = var_522, x = var_519)[name = tensor<string, []>("input_87")];
            tensor<int32, [4]> reshape_48_shape_0 = const()[name = tensor<string, []>("reshape_48_shape_0"), val = tensor<int32, [4]>([108, 1, 64, 476])];
            tensor<fp32, [108, 1, 64, 476]> reshape_48 = reshape(shape = reshape_48_shape_0, x = input_87)[name = tensor<string, []>("reshape_48")];
            tensor<int32, [2]> reduce_mean_36_axes_0 = const()[name = tensor<string, []>("reduce_mean_36_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_36_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_36_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_36 = reduce_mean(axes = reduce_mean_36_axes_0, keep_dims = reduce_mean_36_keep_dims_0, x = reshape_48)[name = tensor<string, []>("reduce_mean_36")];
            tensor<fp32, [108, 1, 64, 476]> sub_24 = sub(x = reshape_48, y = reduce_mean_36)[name = tensor<string, []>("sub_24")];
            tensor<fp32, [108, 1, 64, 476]> square_12 = square(x = sub_24)[name = tensor<string, []>("square_12")];
            tensor<int32, [2]> reduce_mean_38_axes_0 = const()[name = tensor<string, []>("reduce_mean_38_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_38_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_38_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_38 = reduce_mean(axes = reduce_mean_38_axes_0, keep_dims = reduce_mean_38_keep_dims_0, x = square_12)[name = tensor<string, []>("reduce_mean_38")];
            tensor<fp32, []> add_24_y_0 = const()[name = tensor<string, []>("add_24_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [108, 1, 1, 1]> add_24 = add(x = reduce_mean_38, y = add_24_y_0)[name = tensor<string, []>("add_24")];
            tensor<fp32, [108, 1, 1, 1]> sqrt_12 = sqrt(x = add_24)[name = tensor<string, []>("sqrt_12")];
            tensor<fp32, [108, 1, 64, 476]> real_div_12 = real_div(x = sub_24, y = sqrt_12)[name = tensor<string, []>("real_div_12")];
            tensor<int32, [3]> reshape_49_shape_0 = const()[name = tensor<string, []>("reshape_49_shape_0"), val = tensor<int32, [3]>([108, 64, 476])];
            tensor<fp32, [108, 64, 476]> reshape_49 = reshape(shape = reshape_49_shape_0, x = real_div_12)[name = tensor<string, []>("reshape_49")];
            tensor<fp32, [1, 64, 1]> reshape_50 = const()[name = tensor<string, []>("reshape_50"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10691968)))];
            tensor<fp32, [108, 64, 476]> mul_12 = mul(x = reshape_49, y = reshape_50)[name = tensor<string, []>("mul_12")];
            tensor<fp32, [1, 64, 1]> reshape_51 = const()[name = tensor<string, []>("reshape_51"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10692288)))];
            tensor<fp32, [108, 64, 476]> add_25 = add(x = mul_12, y = reshape_51)[name = tensor<string, []>("add_25")];
            tensor<string, []> input_91_pad_type_0 = const()[name = tensor<string, []>("input_91_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_91_pad_0 = const()[name = tensor<string, []>("input_91_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_91_strides_0 = const()[name = tensor<string, []>("input_91_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_91_dilations_0 = const()[name = tensor<string, []>("input_91_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_91_groups_0 = const()[name = tensor<string, []>("input_91_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 32, 476]> input_91 = conv(bias = encoder_1_conv_modules_0_layers_0_1_bias, dilations = input_91_dilations_0, groups = input_91_groups_0, pad = input_91_pad_0, pad_type = input_91_pad_type_0, strides = input_91_strides_0, weight = encoder_1_conv_modules_0_layers_0_1_weight, x = add_25)[name = tensor<string, []>("input_91")];
            tensor<int32, []> input_93_split_num_splits_0 = const()[name = tensor<string, []>("input_93_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_93_split_axis_0 = const()[name = tensor<string, []>("input_93_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 16, 476]> input_93_split_0, tensor<fp32, [108, 16, 476]> input_93_split_1 = split(axis = input_93_split_axis_0, num_splits = input_93_split_num_splits_0, x = input_91)[name = tensor<string, []>("input_93_split")];
            tensor<fp32, [108, 16, 476]> input_93_split_1_sigmoid = sigmoid(x = input_93_split_1)[name = tensor<string, []>("input_93_split_1_sigmoid")];
            tensor<fp32, [108, 16, 476]> input_93 = mul(x = input_93_split_0, y = input_93_split_1_sigmoid)[name = tensor<string, []>("input_93")];
            tensor<string, []> input_95_pad_type_0 = const()[name = tensor<string, []>("input_95_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_95_pad_0 = const()[name = tensor<string, []>("input_95_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_95_groups_0 = const()[name = tensor<string, []>("input_95_groups_0"), val = tensor<int32, []>(16)];
            tensor<int32, [1]> input_95_strides_0 = const()[name = tensor<string, []>("input_95_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_95_dilations_0 = const()[name = tensor<string, []>("input_95_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [108, 16, 476]> input_95 = conv(bias = encoder_1_conv_modules_0_layers_0_3_bias, dilations = input_95_dilations_0, groups = input_95_groups_0, pad = input_95_pad_0, pad_type = input_95_pad_type_0, strides = input_95_strides_0, weight = encoder_1_conv_modules_0_layers_0_3_weight, x = input_93)[name = tensor<string, []>("input_95")];
            tensor<int32, [4]> reshape_52_shape_0 = const()[name = tensor<string, []>("reshape_52_shape_0"), val = tensor<int32, [4]>([108, 1, 16, 476])];
            tensor<fp32, [108, 1, 16, 476]> reshape_52 = reshape(shape = reshape_52_shape_0, x = input_95)[name = tensor<string, []>("reshape_52")];
            tensor<int32, [2]> reduce_mean_39_axes_0 = const()[name = tensor<string, []>("reduce_mean_39_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_39_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_39_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_39 = reduce_mean(axes = reduce_mean_39_axes_0, keep_dims = reduce_mean_39_keep_dims_0, x = reshape_52)[name = tensor<string, []>("reduce_mean_39")];
            tensor<fp32, [108, 1, 16, 476]> sub_26 = sub(x = reshape_52, y = reduce_mean_39)[name = tensor<string, []>("sub_26")];
            tensor<fp32, [108, 1, 16, 476]> square_13 = square(x = sub_26)[name = tensor<string, []>("square_13")];
            tensor<int32, [2]> reduce_mean_41_axes_0 = const()[name = tensor<string, []>("reduce_mean_41_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_41_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_41_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_41 = reduce_mean(axes = reduce_mean_41_axes_0, keep_dims = reduce_mean_41_keep_dims_0, x = square_13)[name = tensor<string, []>("reduce_mean_41")];
            tensor<fp32, []> add_26_y_0 = const()[name = tensor<string, []>("add_26_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [108, 1, 1, 1]> add_26 = add(x = reduce_mean_41, y = add_26_y_0)[name = tensor<string, []>("add_26")];
            tensor<fp32, [108, 1, 1, 1]> sqrt_13 = sqrt(x = add_26)[name = tensor<string, []>("sqrt_13")];
            tensor<fp32, [108, 1, 16, 476]> real_div_13 = real_div(x = sub_26, y = sqrt_13)[name = tensor<string, []>("real_div_13")];
            tensor<int32, [3]> reshape_53_shape_0 = const()[name = tensor<string, []>("reshape_53_shape_0"), val = tensor<int32, [3]>([108, 16, 476])];
            tensor<fp32, [108, 16, 476]> reshape_53 = reshape(shape = reshape_53_shape_0, x = real_div_13)[name = tensor<string, []>("reshape_53")];
            tensor<fp32, [1, 16, 1]> reshape_54 = const()[name = tensor<string, []>("reshape_54"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10692608)))];
            tensor<fp32, [108, 16, 476]> mul_13 = mul(x = reshape_53, y = reshape_54)[name = tensor<string, []>("mul_13")];
            tensor<fp32, [1, 16, 1]> reshape_55 = const()[name = tensor<string, []>("reshape_55"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10692736)))];
            tensor<fp32, [108, 16, 476]> add_27 = add(x = mul_13, y = reshape_55)[name = tensor<string, []>("add_27")];
            tensor<fp32, [108, 16, 476]> input_97 = silu(x = add_27)[name = tensor<string, []>("input_97")];
            tensor<string, []> var_564_pad_type_0 = const()[name = tensor<string, []>("op_564_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_564_strides_0 = const()[name = tensor<string, []>("op_564_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_564_pad_0 = const()[name = tensor<string, []>("op_564_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_564_dilations_0 = const()[name = tensor<string, []>("op_564_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_564_groups_0 = const()[name = tensor<string, []>("op_564_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 64, 476]> var_564 = conv(bias = encoder_1_conv_modules_0_layers_0_6_bias, dilations = var_564_dilations_0, groups = var_564_groups_0, pad = var_564_pad_0, pad_type = var_564_pad_type_0, strides = var_564_strides_0, weight = encoder_1_conv_modules_0_layers_0_6_weight, x = input_97)[name = tensor<string, []>("op_564")];
            tensor<fp32, [108, 64, 476]> input_99 = add(x = input_87, y = var_564)[name = tensor<string, []>("input_99")];
            tensor<int32, [4]> reshape_56_shape_0 = const()[name = tensor<string, []>("reshape_56_shape_0"), val = tensor<int32, [4]>([108, 1, 64, 476])];
            tensor<fp32, [108, 1, 64, 476]> reshape_56 = reshape(shape = reshape_56_shape_0, x = input_99)[name = tensor<string, []>("reshape_56")];
            tensor<int32, [2]> reduce_mean_42_axes_0 = const()[name = tensor<string, []>("reduce_mean_42_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_42_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_42_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_42 = reduce_mean(axes = reduce_mean_42_axes_0, keep_dims = reduce_mean_42_keep_dims_0, x = reshape_56)[name = tensor<string, []>("reduce_mean_42")];
            tensor<fp32, [108, 1, 64, 476]> sub_28 = sub(x = reshape_56, y = reduce_mean_42)[name = tensor<string, []>("sub_28")];
            tensor<fp32, [108, 1, 64, 476]> square_14 = square(x = sub_28)[name = tensor<string, []>("square_14")];
            tensor<int32, [2]> reduce_mean_44_axes_0 = const()[name = tensor<string, []>("reduce_mean_44_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_44_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_44_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_44 = reduce_mean(axes = reduce_mean_44_axes_0, keep_dims = reduce_mean_44_keep_dims_0, x = square_14)[name = tensor<string, []>("reduce_mean_44")];
            tensor<fp32, []> add_28_y_0 = const()[name = tensor<string, []>("add_28_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [108, 1, 1, 1]> add_28 = add(x = reduce_mean_44, y = add_28_y_0)[name = tensor<string, []>("add_28")];
            tensor<fp32, [108, 1, 1, 1]> sqrt_14 = sqrt(x = add_28)[name = tensor<string, []>("sqrt_14")];
            tensor<fp32, [108, 1, 64, 476]> real_div_14 = real_div(x = sub_28, y = sqrt_14)[name = tensor<string, []>("real_div_14")];
            tensor<int32, [3]> reshape_57_shape_0 = const()[name = tensor<string, []>("reshape_57_shape_0"), val = tensor<int32, [3]>([108, 64, 476])];
            tensor<fp32, [108, 64, 476]> reshape_57 = reshape(shape = reshape_57_shape_0, x = real_div_14)[name = tensor<string, []>("reshape_57")];
            tensor<fp32, [1, 64, 1]> reshape_58 = const()[name = tensor<string, []>("reshape_58"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10692864)))];
            tensor<fp32, [108, 64, 476]> mul_14 = mul(x = reshape_57, y = reshape_58)[name = tensor<string, []>("mul_14")];
            tensor<fp32, [1, 64, 1]> reshape_59 = const()[name = tensor<string, []>("reshape_59"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10693184)))];
            tensor<fp32, [108, 64, 476]> add_29 = add(x = mul_14, y = reshape_59)[name = tensor<string, []>("add_29")];
            tensor<string, []> input_103_pad_type_0 = const()[name = tensor<string, []>("input_103_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_103_pad_0 = const()[name = tensor<string, []>("input_103_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_103_strides_0 = const()[name = tensor<string, []>("input_103_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_103_dilations_0 = const()[name = tensor<string, []>("input_103_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_103_groups_0 = const()[name = tensor<string, []>("input_103_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 32, 476]> input_103 = conv(bias = encoder_1_conv_modules_0_layers_1_1_bias, dilations = input_103_dilations_0, groups = input_103_groups_0, pad = input_103_pad_0, pad_type = input_103_pad_type_0, strides = input_103_strides_0, weight = encoder_1_conv_modules_0_layers_1_1_weight, x = add_29)[name = tensor<string, []>("input_103")];
            tensor<int32, []> input_105_split_num_splits_0 = const()[name = tensor<string, []>("input_105_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_105_split_axis_0 = const()[name = tensor<string, []>("input_105_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 16, 476]> input_105_split_0, tensor<fp32, [108, 16, 476]> input_105_split_1 = split(axis = input_105_split_axis_0, num_splits = input_105_split_num_splits_0, x = input_103)[name = tensor<string, []>("input_105_split")];
            tensor<fp32, [108, 16, 476]> input_105_split_1_sigmoid = sigmoid(x = input_105_split_1)[name = tensor<string, []>("input_105_split_1_sigmoid")];
            tensor<fp32, [108, 16, 476]> input_105 = mul(x = input_105_split_0, y = input_105_split_1_sigmoid)[name = tensor<string, []>("input_105")];
            tensor<string, []> input_107_pad_type_0 = const()[name = tensor<string, []>("input_107_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_107_pad_0 = const()[name = tensor<string, []>("input_107_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_107_groups_0 = const()[name = tensor<string, []>("input_107_groups_0"), val = tensor<int32, []>(16)];
            tensor<int32, [1]> input_107_strides_0 = const()[name = tensor<string, []>("input_107_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_107_dilations_0 = const()[name = tensor<string, []>("input_107_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [108, 16, 476]> input_107 = conv(bias = encoder_1_conv_modules_0_layers_1_3_bias, dilations = input_107_dilations_0, groups = input_107_groups_0, pad = input_107_pad_0, pad_type = input_107_pad_type_0, strides = input_107_strides_0, weight = encoder_1_conv_modules_0_layers_1_3_weight, x = input_105)[name = tensor<string, []>("input_107")];
            tensor<int32, [4]> reshape_60_shape_0 = const()[name = tensor<string, []>("reshape_60_shape_0"), val = tensor<int32, [4]>([108, 1, 16, 476])];
            tensor<fp32, [108, 1, 16, 476]> reshape_60 = reshape(shape = reshape_60_shape_0, x = input_107)[name = tensor<string, []>("reshape_60")];
            tensor<int32, [2]> reduce_mean_45_axes_0 = const()[name = tensor<string, []>("reduce_mean_45_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_45_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_45_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_45 = reduce_mean(axes = reduce_mean_45_axes_0, keep_dims = reduce_mean_45_keep_dims_0, x = reshape_60)[name = tensor<string, []>("reduce_mean_45")];
            tensor<fp32, [108, 1, 16, 476]> sub_30 = sub(x = reshape_60, y = reduce_mean_45)[name = tensor<string, []>("sub_30")];
            tensor<fp32, [108, 1, 16, 476]> square_15 = square(x = sub_30)[name = tensor<string, []>("square_15")];
            tensor<int32, [2]> reduce_mean_47_axes_0 = const()[name = tensor<string, []>("reduce_mean_47_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_47_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_47_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_47 = reduce_mean(axes = reduce_mean_47_axes_0, keep_dims = reduce_mean_47_keep_dims_0, x = square_15)[name = tensor<string, []>("reduce_mean_47")];
            tensor<fp32, []> add_30_y_0 = const()[name = tensor<string, []>("add_30_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [108, 1, 1, 1]> add_30 = add(x = reduce_mean_47, y = add_30_y_0)[name = tensor<string, []>("add_30")];
            tensor<fp32, [108, 1, 1, 1]> sqrt_15 = sqrt(x = add_30)[name = tensor<string, []>("sqrt_15")];
            tensor<fp32, [108, 1, 16, 476]> real_div_15 = real_div(x = sub_30, y = sqrt_15)[name = tensor<string, []>("real_div_15")];
            tensor<int32, [3]> reshape_61_shape_0 = const()[name = tensor<string, []>("reshape_61_shape_0"), val = tensor<int32, [3]>([108, 16, 476])];
            tensor<fp32, [108, 16, 476]> reshape_61 = reshape(shape = reshape_61_shape_0, x = real_div_15)[name = tensor<string, []>("reshape_61")];
            tensor<fp32, [1, 16, 1]> reshape_62 = const()[name = tensor<string, []>("reshape_62"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10693504)))];
            tensor<fp32, [108, 16, 476]> mul_15 = mul(x = reshape_61, y = reshape_62)[name = tensor<string, []>("mul_15")];
            tensor<fp32, [1, 16, 1]> reshape_63 = const()[name = tensor<string, []>("reshape_63"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10693632)))];
            tensor<fp32, [108, 16, 476]> add_31 = add(x = mul_15, y = reshape_63)[name = tensor<string, []>("add_31")];
            tensor<fp32, [108, 16, 476]> input_109 = silu(x = add_31)[name = tensor<string, []>("input_109")];
            tensor<string, []> var_600_pad_type_0 = const()[name = tensor<string, []>("op_600_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_600_strides_0 = const()[name = tensor<string, []>("op_600_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_600_pad_0 = const()[name = tensor<string, []>("op_600_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_600_dilations_0 = const()[name = tensor<string, []>("op_600_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_600_groups_0 = const()[name = tensor<string, []>("op_600_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 64, 476]> var_600 = conv(bias = encoder_1_conv_modules_0_layers_1_6_bias, dilations = var_600_dilations_0, groups = var_600_groups_0, pad = var_600_pad_0, pad_type = var_600_pad_type_0, strides = var_600_strides_0, weight = encoder_1_conv_modules_0_layers_1_6_weight, x = input_109)[name = tensor<string, []>("op_600")];
            tensor<fp32, [108, 64, 476]> input_111 = add(x = input_99, y = var_600)[name = tensor<string, []>("input_111")];
            tensor<int32, [4]> reshape_64_shape_0 = const()[name = tensor<string, []>("reshape_64_shape_0"), val = tensor<int32, [4]>([108, 1, 64, 476])];
            tensor<fp32, [108, 1, 64, 476]> reshape_64 = reshape(shape = reshape_64_shape_0, x = input_111)[name = tensor<string, []>("reshape_64")];
            tensor<int32, [2]> reduce_mean_48_axes_0 = const()[name = tensor<string, []>("reduce_mean_48_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_48_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_48_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_48 = reduce_mean(axes = reduce_mean_48_axes_0, keep_dims = reduce_mean_48_keep_dims_0, x = reshape_64)[name = tensor<string, []>("reduce_mean_48")];
            tensor<fp32, [108, 1, 64, 476]> sub_32 = sub(x = reshape_64, y = reduce_mean_48)[name = tensor<string, []>("sub_32")];
            tensor<fp32, [108, 1, 64, 476]> square_16 = square(x = sub_32)[name = tensor<string, []>("square_16")];
            tensor<int32, [2]> reduce_mean_50_axes_0 = const()[name = tensor<string, []>("reduce_mean_50_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_50_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_50_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_50 = reduce_mean(axes = reduce_mean_50_axes_0, keep_dims = reduce_mean_50_keep_dims_0, x = square_16)[name = tensor<string, []>("reduce_mean_50")];
            tensor<fp32, []> add_32_y_0 = const()[name = tensor<string, []>("add_32_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [108, 1, 1, 1]> add_32 = add(x = reduce_mean_50, y = add_32_y_0)[name = tensor<string, []>("add_32")];
            tensor<fp32, [108, 1, 1, 1]> sqrt_16 = sqrt(x = add_32)[name = tensor<string, []>("sqrt_16")];
            tensor<fp32, [108, 1, 64, 476]> real_div_16 = real_div(x = sub_32, y = sqrt_16)[name = tensor<string, []>("real_div_16")];
            tensor<int32, [3]> reshape_65_shape_0 = const()[name = tensor<string, []>("reshape_65_shape_0"), val = tensor<int32, [3]>([108, 64, 476])];
            tensor<fp32, [108, 64, 476]> reshape_65 = reshape(shape = reshape_65_shape_0, x = real_div_16)[name = tensor<string, []>("reshape_65")];
            tensor<fp32, [1, 64, 1]> reshape_66 = const()[name = tensor<string, []>("reshape_66"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10693760)))];
            tensor<fp32, [108, 64, 476]> mul_16 = mul(x = reshape_65, y = reshape_66)[name = tensor<string, []>("mul_16")];
            tensor<fp32, [1, 64, 1]> reshape_67 = const()[name = tensor<string, []>("reshape_67"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10694080)))];
            tensor<fp32, [108, 64, 476]> add_33 = add(x = mul_16, y = reshape_67)[name = tensor<string, []>("add_33")];
            tensor<string, []> input_115_pad_type_0 = const()[name = tensor<string, []>("input_115_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_115_pad_0 = const()[name = tensor<string, []>("input_115_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_115_strides_0 = const()[name = tensor<string, []>("input_115_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_115_dilations_0 = const()[name = tensor<string, []>("input_115_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_115_groups_0 = const()[name = tensor<string, []>("input_115_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 32, 476]> input_115 = conv(bias = encoder_1_conv_modules_0_layers_2_1_bias, dilations = input_115_dilations_0, groups = input_115_groups_0, pad = input_115_pad_0, pad_type = input_115_pad_type_0, strides = input_115_strides_0, weight = encoder_1_conv_modules_0_layers_2_1_weight, x = add_33)[name = tensor<string, []>("input_115")];
            tensor<int32, []> input_117_split_num_splits_0 = const()[name = tensor<string, []>("input_117_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_117_split_axis_0 = const()[name = tensor<string, []>("input_117_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 16, 476]> input_117_split_0, tensor<fp32, [108, 16, 476]> input_117_split_1 = split(axis = input_117_split_axis_0, num_splits = input_117_split_num_splits_0, x = input_115)[name = tensor<string, []>("input_117_split")];
            tensor<fp32, [108, 16, 476]> input_117_split_1_sigmoid = sigmoid(x = input_117_split_1)[name = tensor<string, []>("input_117_split_1_sigmoid")];
            tensor<fp32, [108, 16, 476]> input_117 = mul(x = input_117_split_0, y = input_117_split_1_sigmoid)[name = tensor<string, []>("input_117")];
            tensor<string, []> input_119_pad_type_0 = const()[name = tensor<string, []>("input_119_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_119_pad_0 = const()[name = tensor<string, []>("input_119_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_119_groups_0 = const()[name = tensor<string, []>("input_119_groups_0"), val = tensor<int32, []>(16)];
            tensor<int32, [1]> input_119_strides_0 = const()[name = tensor<string, []>("input_119_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_119_dilations_0 = const()[name = tensor<string, []>("input_119_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [108, 16, 476]> input_119 = conv(bias = encoder_1_conv_modules_0_layers_2_3_bias, dilations = input_119_dilations_0, groups = input_119_groups_0, pad = input_119_pad_0, pad_type = input_119_pad_type_0, strides = input_119_strides_0, weight = encoder_1_conv_modules_0_layers_2_3_weight, x = input_117)[name = tensor<string, []>("input_119")];
            tensor<int32, [4]> reshape_68_shape_0 = const()[name = tensor<string, []>("reshape_68_shape_0"), val = tensor<int32, [4]>([108, 1, 16, 476])];
            tensor<fp32, [108, 1, 16, 476]> reshape_68 = reshape(shape = reshape_68_shape_0, x = input_119)[name = tensor<string, []>("reshape_68")];
            tensor<int32, [2]> reduce_mean_51_axes_0 = const()[name = tensor<string, []>("reduce_mean_51_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_51_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_51_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_51 = reduce_mean(axes = reduce_mean_51_axes_0, keep_dims = reduce_mean_51_keep_dims_0, x = reshape_68)[name = tensor<string, []>("reduce_mean_51")];
            tensor<fp32, [108, 1, 16, 476]> sub_34 = sub(x = reshape_68, y = reduce_mean_51)[name = tensor<string, []>("sub_34")];
            tensor<fp32, [108, 1, 16, 476]> square_17 = square(x = sub_34)[name = tensor<string, []>("square_17")];
            tensor<int32, [2]> reduce_mean_53_axes_0 = const()[name = tensor<string, []>("reduce_mean_53_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_53_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_53_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [108, 1, 1, 1]> reduce_mean_53 = reduce_mean(axes = reduce_mean_53_axes_0, keep_dims = reduce_mean_53_keep_dims_0, x = square_17)[name = tensor<string, []>("reduce_mean_53")];
            tensor<fp32, []> add_34_y_0 = const()[name = tensor<string, []>("add_34_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [108, 1, 1, 1]> add_34 = add(x = reduce_mean_53, y = add_34_y_0)[name = tensor<string, []>("add_34")];
            tensor<fp32, [108, 1, 1, 1]> sqrt_17 = sqrt(x = add_34)[name = tensor<string, []>("sqrt_17")];
            tensor<fp32, [108, 1, 16, 476]> real_div_17 = real_div(x = sub_34, y = sqrt_17)[name = tensor<string, []>("real_div_17")];
            tensor<int32, [3]> reshape_69_shape_0 = const()[name = tensor<string, []>("reshape_69_shape_0"), val = tensor<int32, [3]>([108, 16, 476])];
            tensor<fp32, [108, 16, 476]> reshape_69 = reshape(shape = reshape_69_shape_0, x = real_div_17)[name = tensor<string, []>("reshape_69")];
            tensor<fp32, [1, 16, 1]> reshape_70 = const()[name = tensor<string, []>("reshape_70"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10694400)))];
            tensor<fp32, [108, 16, 476]> mul_17 = mul(x = reshape_69, y = reshape_70)[name = tensor<string, []>("mul_17")];
            tensor<fp32, [1, 16, 1]> reshape_71 = const()[name = tensor<string, []>("reshape_71"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10694528)))];
            tensor<fp32, [108, 16, 476]> add_35 = add(x = mul_17, y = reshape_71)[name = tensor<string, []>("add_35")];
            tensor<fp32, [108, 16, 476]> input_121 = silu(x = add_35)[name = tensor<string, []>("input_121")];
            tensor<string, []> var_636_pad_type_0 = const()[name = tensor<string, []>("op_636_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_636_strides_0 = const()[name = tensor<string, []>("op_636_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_636_pad_0 = const()[name = tensor<string, []>("op_636_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_636_dilations_0 = const()[name = tensor<string, []>("op_636_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_636_groups_0 = const()[name = tensor<string, []>("op_636_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [108, 64, 476]> var_636 = conv(bias = encoder_1_conv_modules_0_layers_2_6_bias, dilations = var_636_dilations_0, groups = var_636_groups_0, pad = var_636_pad_0, pad_type = var_636_pad_type_0, strides = var_636_strides_0, weight = encoder_1_conv_modules_0_layers_2_6_weight, x = input_121)[name = tensor<string, []>("op_636")];
            tensor<fp32, [108, 64, 476]> var_637 = add(x = input_111, y = var_636)[name = tensor<string, []>("op_637")];
            tensor<int32, [4]> var_642 = const()[name = tensor<string, []>("op_642"), val = tensor<int32, [4]>([1, 108, 64, 476])];
            tensor<fp32, [1, 108, 64, 476]> var_643 = reshape(shape = var_642, x = var_637)[name = tensor<string, []>("op_643")];
            tensor<string, []> band_19_mode_0 = const()[name = tensor<string, []>("band_19_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 108, 64, 476]> band_19 = gelu(mode = band_19_mode_0, x = var_643)[name = tensor<string, []>("band_19")];
            tensor<int32, [4]> var_647 = const()[name = tensor<string, []>("op_647"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_651 = const()[name = tensor<string, []>("op_651"), val = tensor<int32, [3]>([-1, 64, 476])];
            tensor<fp32, [1, 61, 64, 476]> var_648 = transpose(perm = var_647, x = band_15)[name = tensor<string, []>("transpose_117")];
            tensor<fp32, [61, 64, 476]> input_123 = reshape(shape = var_651, x = var_648)[name = tensor<string, []>("input_123")];
            tensor<int32, [4]> reshape_72_shape_0 = const()[name = tensor<string, []>("reshape_72_shape_0"), val = tensor<int32, [4]>([61, 1, 64, 476])];
            tensor<fp32, [61, 1, 64, 476]> reshape_72 = reshape(shape = reshape_72_shape_0, x = input_123)[name = tensor<string, []>("reshape_72")];
            tensor<int32, [2]> reduce_mean_54_axes_0 = const()[name = tensor<string, []>("reduce_mean_54_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_54_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_54_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [61, 1, 1, 1]> reduce_mean_54 = reduce_mean(axes = reduce_mean_54_axes_0, keep_dims = reduce_mean_54_keep_dims_0, x = reshape_72)[name = tensor<string, []>("reduce_mean_54")];
            tensor<fp32, [61, 1, 64, 476]> sub_36 = sub(x = reshape_72, y = reduce_mean_54)[name = tensor<string, []>("sub_36")];
            tensor<fp32, [61, 1, 64, 476]> square_18 = square(x = sub_36)[name = tensor<string, []>("square_18")];
            tensor<int32, [2]> reduce_mean_56_axes_0 = const()[name = tensor<string, []>("reduce_mean_56_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_56_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_56_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [61, 1, 1, 1]> reduce_mean_56 = reduce_mean(axes = reduce_mean_56_axes_0, keep_dims = reduce_mean_56_keep_dims_0, x = square_18)[name = tensor<string, []>("reduce_mean_56")];
            tensor<fp32, []> add_36_y_0 = const()[name = tensor<string, []>("add_36_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [61, 1, 1, 1]> add_36 = add(x = reduce_mean_56, y = add_36_y_0)[name = tensor<string, []>("add_36")];
            tensor<fp32, [61, 1, 1, 1]> sqrt_18 = sqrt(x = add_36)[name = tensor<string, []>("sqrt_18")];
            tensor<fp32, [61, 1, 64, 476]> real_div_18 = real_div(x = sub_36, y = sqrt_18)[name = tensor<string, []>("real_div_18")];
            tensor<int32, [3]> reshape_73_shape_0 = const()[name = tensor<string, []>("reshape_73_shape_0"), val = tensor<int32, [3]>([61, 64, 476])];
            tensor<fp32, [61, 64, 476]> reshape_73 = reshape(shape = reshape_73_shape_0, x = real_div_18)[name = tensor<string, []>("reshape_73")];
            tensor<fp32, [1, 64, 1]> reshape_74 = const()[name = tensor<string, []>("reshape_74"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10694656)))];
            tensor<fp32, [61, 64, 476]> mul_18 = mul(x = reshape_73, y = reshape_74)[name = tensor<string, []>("mul_18")];
            tensor<fp32, [1, 64, 1]> reshape_75 = const()[name = tensor<string, []>("reshape_75"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10694976)))];
            tensor<fp32, [61, 64, 476]> add_37 = add(x = mul_18, y = reshape_75)[name = tensor<string, []>("add_37")];
            tensor<string, []> input_127_pad_type_0 = const()[name = tensor<string, []>("input_127_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_127_pad_0 = const()[name = tensor<string, []>("input_127_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_127_strides_0 = const()[name = tensor<string, []>("input_127_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_127_dilations_0 = const()[name = tensor<string, []>("input_127_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_127_groups_0 = const()[name = tensor<string, []>("input_127_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [61, 32, 476]> input_127 = conv(bias = encoder_1_conv_modules_1_layers_0_1_bias, dilations = input_127_dilations_0, groups = input_127_groups_0, pad = input_127_pad_0, pad_type = input_127_pad_type_0, strides = input_127_strides_0, weight = encoder_1_conv_modules_1_layers_0_1_weight, x = add_37)[name = tensor<string, []>("input_127")];
            tensor<int32, []> input_129_split_num_splits_0 = const()[name = tensor<string, []>("input_129_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_129_split_axis_0 = const()[name = tensor<string, []>("input_129_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [61, 16, 476]> input_129_split_0, tensor<fp32, [61, 16, 476]> input_129_split_1 = split(axis = input_129_split_axis_0, num_splits = input_129_split_num_splits_0, x = input_127)[name = tensor<string, []>("input_129_split")];
            tensor<fp32, [61, 16, 476]> input_129_split_1_sigmoid = sigmoid(x = input_129_split_1)[name = tensor<string, []>("input_129_split_1_sigmoid")];
            tensor<fp32, [61, 16, 476]> input_129 = mul(x = input_129_split_0, y = input_129_split_1_sigmoid)[name = tensor<string, []>("input_129")];
            tensor<string, []> input_131_pad_type_0 = const()[name = tensor<string, []>("input_131_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_131_pad_0 = const()[name = tensor<string, []>("input_131_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_131_groups_0 = const()[name = tensor<string, []>("input_131_groups_0"), val = tensor<int32, []>(16)];
            tensor<int32, [1]> input_131_strides_0 = const()[name = tensor<string, []>("input_131_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_131_dilations_0 = const()[name = tensor<string, []>("input_131_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [61, 16, 476]> input_131 = conv(bias = encoder_1_conv_modules_1_layers_0_3_bias, dilations = input_131_dilations_0, groups = input_131_groups_0, pad = input_131_pad_0, pad_type = input_131_pad_type_0, strides = input_131_strides_0, weight = encoder_1_conv_modules_1_layers_0_3_weight, x = input_129)[name = tensor<string, []>("input_131")];
            tensor<int32, [4]> reshape_76_shape_0 = const()[name = tensor<string, []>("reshape_76_shape_0"), val = tensor<int32, [4]>([61, 1, 16, 476])];
            tensor<fp32, [61, 1, 16, 476]> reshape_76 = reshape(shape = reshape_76_shape_0, x = input_131)[name = tensor<string, []>("reshape_76")];
            tensor<int32, [2]> reduce_mean_57_axes_0 = const()[name = tensor<string, []>("reduce_mean_57_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_57_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_57_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [61, 1, 1, 1]> reduce_mean_57 = reduce_mean(axes = reduce_mean_57_axes_0, keep_dims = reduce_mean_57_keep_dims_0, x = reshape_76)[name = tensor<string, []>("reduce_mean_57")];
            tensor<fp32, [61, 1, 16, 476]> sub_38 = sub(x = reshape_76, y = reduce_mean_57)[name = tensor<string, []>("sub_38")];
            tensor<fp32, [61, 1, 16, 476]> square_19 = square(x = sub_38)[name = tensor<string, []>("square_19")];
            tensor<int32, [2]> reduce_mean_59_axes_0 = const()[name = tensor<string, []>("reduce_mean_59_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_59_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_59_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [61, 1, 1, 1]> reduce_mean_59 = reduce_mean(axes = reduce_mean_59_axes_0, keep_dims = reduce_mean_59_keep_dims_0, x = square_19)[name = tensor<string, []>("reduce_mean_59")];
            tensor<fp32, []> add_38_y_0 = const()[name = tensor<string, []>("add_38_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [61, 1, 1, 1]> add_38 = add(x = reduce_mean_59, y = add_38_y_0)[name = tensor<string, []>("add_38")];
            tensor<fp32, [61, 1, 1, 1]> sqrt_19 = sqrt(x = add_38)[name = tensor<string, []>("sqrt_19")];
            tensor<fp32, [61, 1, 16, 476]> real_div_19 = real_div(x = sub_38, y = sqrt_19)[name = tensor<string, []>("real_div_19")];
            tensor<int32, [3]> reshape_77_shape_0 = const()[name = tensor<string, []>("reshape_77_shape_0"), val = tensor<int32, [3]>([61, 16, 476])];
            tensor<fp32, [61, 16, 476]> reshape_77 = reshape(shape = reshape_77_shape_0, x = real_div_19)[name = tensor<string, []>("reshape_77")];
            tensor<fp32, [1, 16, 1]> reshape_78 = const()[name = tensor<string, []>("reshape_78"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10695296)))];
            tensor<fp32, [61, 16, 476]> mul_19 = mul(x = reshape_77, y = reshape_78)[name = tensor<string, []>("mul_19")];
            tensor<fp32, [1, 16, 1]> reshape_79 = const()[name = tensor<string, []>("reshape_79"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10695424)))];
            tensor<fp32, [61, 16, 476]> add_39 = add(x = mul_19, y = reshape_79)[name = tensor<string, []>("add_39")];
            tensor<fp32, [61, 16, 476]> input_133 = silu(x = add_39)[name = tensor<string, []>("input_133")];
            tensor<string, []> var_691_pad_type_0 = const()[name = tensor<string, []>("op_691_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_691_strides_0 = const()[name = tensor<string, []>("op_691_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_691_pad_0 = const()[name = tensor<string, []>("op_691_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_691_dilations_0 = const()[name = tensor<string, []>("op_691_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_691_groups_0 = const()[name = tensor<string, []>("op_691_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [61, 64, 476]> var_691 = conv(bias = encoder_1_conv_modules_1_layers_0_6_bias, dilations = var_691_dilations_0, groups = var_691_groups_0, pad = var_691_pad_0, pad_type = var_691_pad_type_0, strides = var_691_strides_0, weight = encoder_1_conv_modules_1_layers_0_6_weight, x = input_133)[name = tensor<string, []>("op_691")];
            tensor<fp32, [61, 64, 476]> input_135 = add(x = input_123, y = var_691)[name = tensor<string, []>("input_135")];
            tensor<int32, [4]> reshape_80_shape_0 = const()[name = tensor<string, []>("reshape_80_shape_0"), val = tensor<int32, [4]>([61, 1, 64, 476])];
            tensor<fp32, [61, 1, 64, 476]> reshape_80 = reshape(shape = reshape_80_shape_0, x = input_135)[name = tensor<string, []>("reshape_80")];
            tensor<int32, [2]> reduce_mean_60_axes_0 = const()[name = tensor<string, []>("reduce_mean_60_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_60_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_60_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [61, 1, 1, 1]> reduce_mean_60 = reduce_mean(axes = reduce_mean_60_axes_0, keep_dims = reduce_mean_60_keep_dims_0, x = reshape_80)[name = tensor<string, []>("reduce_mean_60")];
            tensor<fp32, [61, 1, 64, 476]> sub_40 = sub(x = reshape_80, y = reduce_mean_60)[name = tensor<string, []>("sub_40")];
            tensor<fp32, [61, 1, 64, 476]> square_20 = square(x = sub_40)[name = tensor<string, []>("square_20")];
            tensor<int32, [2]> reduce_mean_62_axes_0 = const()[name = tensor<string, []>("reduce_mean_62_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_62_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_62_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [61, 1, 1, 1]> reduce_mean_62 = reduce_mean(axes = reduce_mean_62_axes_0, keep_dims = reduce_mean_62_keep_dims_0, x = square_20)[name = tensor<string, []>("reduce_mean_62")];
            tensor<fp32, []> add_40_y_0 = const()[name = tensor<string, []>("add_40_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [61, 1, 1, 1]> add_40 = add(x = reduce_mean_62, y = add_40_y_0)[name = tensor<string, []>("add_40")];
            tensor<fp32, [61, 1, 1, 1]> sqrt_20 = sqrt(x = add_40)[name = tensor<string, []>("sqrt_20")];
            tensor<fp32, [61, 1, 64, 476]> real_div_20 = real_div(x = sub_40, y = sqrt_20)[name = tensor<string, []>("real_div_20")];
            tensor<int32, [3]> reshape_81_shape_0 = const()[name = tensor<string, []>("reshape_81_shape_0"), val = tensor<int32, [3]>([61, 64, 476])];
            tensor<fp32, [61, 64, 476]> reshape_81 = reshape(shape = reshape_81_shape_0, x = real_div_20)[name = tensor<string, []>("reshape_81")];
            tensor<fp32, [1, 64, 1]> reshape_82 = const()[name = tensor<string, []>("reshape_82"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10695552)))];
            tensor<fp32, [61, 64, 476]> mul_20 = mul(x = reshape_81, y = reshape_82)[name = tensor<string, []>("mul_20")];
            tensor<fp32, [1, 64, 1]> reshape_83 = const()[name = tensor<string, []>("reshape_83"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10695872)))];
            tensor<fp32, [61, 64, 476]> add_41 = add(x = mul_20, y = reshape_83)[name = tensor<string, []>("add_41")];
            tensor<string, []> input_139_pad_type_0 = const()[name = tensor<string, []>("input_139_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_139_pad_0 = const()[name = tensor<string, []>("input_139_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_139_strides_0 = const()[name = tensor<string, []>("input_139_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_139_dilations_0 = const()[name = tensor<string, []>("input_139_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_139_groups_0 = const()[name = tensor<string, []>("input_139_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [61, 32, 476]> input_139 = conv(bias = encoder_1_conv_modules_1_layers_1_1_bias, dilations = input_139_dilations_0, groups = input_139_groups_0, pad = input_139_pad_0, pad_type = input_139_pad_type_0, strides = input_139_strides_0, weight = encoder_1_conv_modules_1_layers_1_1_weight, x = add_41)[name = tensor<string, []>("input_139")];
            tensor<int32, []> input_141_split_num_splits_0 = const()[name = tensor<string, []>("input_141_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_141_split_axis_0 = const()[name = tensor<string, []>("input_141_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [61, 16, 476]> input_141_split_0, tensor<fp32, [61, 16, 476]> input_141_split_1 = split(axis = input_141_split_axis_0, num_splits = input_141_split_num_splits_0, x = input_139)[name = tensor<string, []>("input_141_split")];
            tensor<fp32, [61, 16, 476]> input_141_split_1_sigmoid = sigmoid(x = input_141_split_1)[name = tensor<string, []>("input_141_split_1_sigmoid")];
            tensor<fp32, [61, 16, 476]> input_141 = mul(x = input_141_split_0, y = input_141_split_1_sigmoid)[name = tensor<string, []>("input_141")];
            tensor<string, []> input_143_pad_type_0 = const()[name = tensor<string, []>("input_143_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_143_pad_0 = const()[name = tensor<string, []>("input_143_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_143_groups_0 = const()[name = tensor<string, []>("input_143_groups_0"), val = tensor<int32, []>(16)];
            tensor<int32, [1]> input_143_strides_0 = const()[name = tensor<string, []>("input_143_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_143_dilations_0 = const()[name = tensor<string, []>("input_143_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [61, 16, 476]> input_143 = conv(bias = encoder_1_conv_modules_1_layers_1_3_bias, dilations = input_143_dilations_0, groups = input_143_groups_0, pad = input_143_pad_0, pad_type = input_143_pad_type_0, strides = input_143_strides_0, weight = encoder_1_conv_modules_1_layers_1_3_weight, x = input_141)[name = tensor<string, []>("input_143")];
            tensor<int32, [4]> reshape_84_shape_0 = const()[name = tensor<string, []>("reshape_84_shape_0"), val = tensor<int32, [4]>([61, 1, 16, 476])];
            tensor<fp32, [61, 1, 16, 476]> reshape_84 = reshape(shape = reshape_84_shape_0, x = input_143)[name = tensor<string, []>("reshape_84")];
            tensor<int32, [2]> reduce_mean_63_axes_0 = const()[name = tensor<string, []>("reduce_mean_63_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_63_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_63_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [61, 1, 1, 1]> reduce_mean_63 = reduce_mean(axes = reduce_mean_63_axes_0, keep_dims = reduce_mean_63_keep_dims_0, x = reshape_84)[name = tensor<string, []>("reduce_mean_63")];
            tensor<fp32, [61, 1, 16, 476]> sub_42 = sub(x = reshape_84, y = reduce_mean_63)[name = tensor<string, []>("sub_42")];
            tensor<fp32, [61, 1, 16, 476]> square_21 = square(x = sub_42)[name = tensor<string, []>("square_21")];
            tensor<int32, [2]> reduce_mean_65_axes_0 = const()[name = tensor<string, []>("reduce_mean_65_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_65_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_65_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [61, 1, 1, 1]> reduce_mean_65 = reduce_mean(axes = reduce_mean_65_axes_0, keep_dims = reduce_mean_65_keep_dims_0, x = square_21)[name = tensor<string, []>("reduce_mean_65")];
            tensor<fp32, []> add_42_y_0 = const()[name = tensor<string, []>("add_42_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [61, 1, 1, 1]> add_42 = add(x = reduce_mean_65, y = add_42_y_0)[name = tensor<string, []>("add_42")];
            tensor<fp32, [61, 1, 1, 1]> sqrt_21 = sqrt(x = add_42)[name = tensor<string, []>("sqrt_21")];
            tensor<fp32, [61, 1, 16, 476]> real_div_21 = real_div(x = sub_42, y = sqrt_21)[name = tensor<string, []>("real_div_21")];
            tensor<int32, [3]> reshape_85_shape_0 = const()[name = tensor<string, []>("reshape_85_shape_0"), val = tensor<int32, [3]>([61, 16, 476])];
            tensor<fp32, [61, 16, 476]> reshape_85 = reshape(shape = reshape_85_shape_0, x = real_div_21)[name = tensor<string, []>("reshape_85")];
            tensor<fp32, [1, 16, 1]> reshape_86 = const()[name = tensor<string, []>("reshape_86"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10696192)))];
            tensor<fp32, [61, 16, 476]> mul_21 = mul(x = reshape_85, y = reshape_86)[name = tensor<string, []>("mul_21")];
            tensor<fp32, [1, 16, 1]> reshape_87 = const()[name = tensor<string, []>("reshape_87"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10696320)))];
            tensor<fp32, [61, 16, 476]> add_43 = add(x = mul_21, y = reshape_87)[name = tensor<string, []>("add_43")];
            tensor<fp32, [61, 16, 476]> input_145 = silu(x = add_43)[name = tensor<string, []>("input_145")];
            tensor<string, []> var_727_pad_type_0 = const()[name = tensor<string, []>("op_727_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_727_strides_0 = const()[name = tensor<string, []>("op_727_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_727_pad_0 = const()[name = tensor<string, []>("op_727_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_727_dilations_0 = const()[name = tensor<string, []>("op_727_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_727_groups_0 = const()[name = tensor<string, []>("op_727_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [61, 64, 476]> var_727 = conv(bias = encoder_1_conv_modules_1_layers_1_6_bias, dilations = var_727_dilations_0, groups = var_727_groups_0, pad = var_727_pad_0, pad_type = var_727_pad_type_0, strides = var_727_strides_0, weight = encoder_1_conv_modules_1_layers_1_6_weight, x = input_145)[name = tensor<string, []>("op_727")];
            tensor<fp32, [61, 64, 476]> var_728 = add(x = input_135, y = var_727)[name = tensor<string, []>("op_728")];
            tensor<int32, [4]> var_733 = const()[name = tensor<string, []>("op_733"), val = tensor<int32, [4]>([1, 61, 64, 476])];
            tensor<fp32, [1, 61, 64, 476]> var_734 = reshape(shape = var_733, x = var_728)[name = tensor<string, []>("op_734")];
            tensor<string, []> band_21_mode_0 = const()[name = tensor<string, []>("band_21_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 61, 64, 476]> band_21 = gelu(mode = band_21_mode_0, x = var_734)[name = tensor<string, []>("band_21")];
            tensor<int32, [4]> var_738 = const()[name = tensor<string, []>("op_738"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_742 = const()[name = tensor<string, []>("op_742"), val = tensor<int32, [3]>([-1, 64, 476])];
            tensor<fp32, [1, 17, 64, 476]> var_739 = transpose(perm = var_738, x = band_17)[name = tensor<string, []>("transpose_116")];
            tensor<fp32, [17, 64, 476]> input_147 = reshape(shape = var_742, x = var_739)[name = tensor<string, []>("input_147")];
            tensor<int32, [4]> reshape_88_shape_0 = const()[name = tensor<string, []>("reshape_88_shape_0"), val = tensor<int32, [4]>([17, 1, 64, 476])];
            tensor<fp32, [17, 1, 64, 476]> reshape_88 = reshape(shape = reshape_88_shape_0, x = input_147)[name = tensor<string, []>("reshape_88")];
            tensor<int32, [2]> reduce_mean_66_axes_0 = const()[name = tensor<string, []>("reduce_mean_66_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_66_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_66_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [17, 1, 1, 1]> reduce_mean_66 = reduce_mean(axes = reduce_mean_66_axes_0, keep_dims = reduce_mean_66_keep_dims_0, x = reshape_88)[name = tensor<string, []>("reduce_mean_66")];
            tensor<fp32, [17, 1, 64, 476]> sub_44 = sub(x = reshape_88, y = reduce_mean_66)[name = tensor<string, []>("sub_44")];
            tensor<fp32, [17, 1, 64, 476]> square_22 = square(x = sub_44)[name = tensor<string, []>("square_22")];
            tensor<int32, [2]> reduce_mean_68_axes_0 = const()[name = tensor<string, []>("reduce_mean_68_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_68_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_68_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [17, 1, 1, 1]> reduce_mean_68 = reduce_mean(axes = reduce_mean_68_axes_0, keep_dims = reduce_mean_68_keep_dims_0, x = square_22)[name = tensor<string, []>("reduce_mean_68")];
            tensor<fp32, []> add_44_y_0 = const()[name = tensor<string, []>("add_44_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [17, 1, 1, 1]> add_44 = add(x = reduce_mean_68, y = add_44_y_0)[name = tensor<string, []>("add_44")];
            tensor<fp32, [17, 1, 1, 1]> sqrt_22 = sqrt(x = add_44)[name = tensor<string, []>("sqrt_22")];
            tensor<fp32, [17, 1, 64, 476]> real_div_22 = real_div(x = sub_44, y = sqrt_22)[name = tensor<string, []>("real_div_22")];
            tensor<int32, [3]> reshape_89_shape_0 = const()[name = tensor<string, []>("reshape_89_shape_0"), val = tensor<int32, [3]>([17, 64, 476])];
            tensor<fp32, [17, 64, 476]> reshape_89 = reshape(shape = reshape_89_shape_0, x = real_div_22)[name = tensor<string, []>("reshape_89")];
            tensor<fp32, [1, 64, 1]> reshape_90 = const()[name = tensor<string, []>("reshape_90"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10696448)))];
            tensor<fp32, [17, 64, 476]> mul_22 = mul(x = reshape_89, y = reshape_90)[name = tensor<string, []>("mul_22")];
            tensor<fp32, [1, 64, 1]> reshape_91 = const()[name = tensor<string, []>("reshape_91"), val = tensor<fp32, [1, 64, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10696768)))];
            tensor<fp32, [17, 64, 476]> add_45 = add(x = mul_22, y = reshape_91)[name = tensor<string, []>("add_45")];
            tensor<string, []> input_151_pad_type_0 = const()[name = tensor<string, []>("input_151_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_151_pad_0 = const()[name = tensor<string, []>("input_151_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_151_strides_0 = const()[name = tensor<string, []>("input_151_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_151_dilations_0 = const()[name = tensor<string, []>("input_151_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_151_groups_0 = const()[name = tensor<string, []>("input_151_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [17, 32, 476]> input_151 = conv(bias = encoder_1_conv_modules_2_layers_0_1_bias, dilations = input_151_dilations_0, groups = input_151_groups_0, pad = input_151_pad_0, pad_type = input_151_pad_type_0, strides = input_151_strides_0, weight = encoder_1_conv_modules_2_layers_0_1_weight, x = add_45)[name = tensor<string, []>("input_151")];
            tensor<int32, []> input_153_split_num_splits_0 = const()[name = tensor<string, []>("input_153_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_153_split_axis_0 = const()[name = tensor<string, []>("input_153_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [17, 16, 476]> input_153_split_0, tensor<fp32, [17, 16, 476]> input_153_split_1 = split(axis = input_153_split_axis_0, num_splits = input_153_split_num_splits_0, x = input_151)[name = tensor<string, []>("input_153_split")];
            tensor<fp32, [17, 16, 476]> input_153_split_1_sigmoid = sigmoid(x = input_153_split_1)[name = tensor<string, []>("input_153_split_1_sigmoid")];
            tensor<fp32, [17, 16, 476]> input_153 = mul(x = input_153_split_0, y = input_153_split_1_sigmoid)[name = tensor<string, []>("input_153")];
            tensor<string, []> input_155_pad_type_0 = const()[name = tensor<string, []>("input_155_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_155_pad_0 = const()[name = tensor<string, []>("input_155_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_155_groups_0 = const()[name = tensor<string, []>("input_155_groups_0"), val = tensor<int32, []>(16)];
            tensor<int32, [1]> input_155_strides_0 = const()[name = tensor<string, []>("input_155_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_155_dilations_0 = const()[name = tensor<string, []>("input_155_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [17, 16, 476]> input_155 = conv(bias = encoder_1_conv_modules_2_layers_0_3_bias, dilations = input_155_dilations_0, groups = input_155_groups_0, pad = input_155_pad_0, pad_type = input_155_pad_type_0, strides = input_155_strides_0, weight = encoder_1_conv_modules_2_layers_0_3_weight, x = input_153)[name = tensor<string, []>("input_155")];
            tensor<int32, [4]> reshape_92_shape_0 = const()[name = tensor<string, []>("reshape_92_shape_0"), val = tensor<int32, [4]>([17, 1, 16, 476])];
            tensor<fp32, [17, 1, 16, 476]> reshape_92 = reshape(shape = reshape_92_shape_0, x = input_155)[name = tensor<string, []>("reshape_92")];
            tensor<int32, [2]> reduce_mean_69_axes_0 = const()[name = tensor<string, []>("reduce_mean_69_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_69_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_69_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [17, 1, 1, 1]> reduce_mean_69 = reduce_mean(axes = reduce_mean_69_axes_0, keep_dims = reduce_mean_69_keep_dims_0, x = reshape_92)[name = tensor<string, []>("reduce_mean_69")];
            tensor<fp32, [17, 1, 16, 476]> sub_46 = sub(x = reshape_92, y = reduce_mean_69)[name = tensor<string, []>("sub_46")];
            tensor<fp32, [17, 1, 16, 476]> square_23 = square(x = sub_46)[name = tensor<string, []>("square_23")];
            tensor<int32, [2]> reduce_mean_71_axes_0 = const()[name = tensor<string, []>("reduce_mean_71_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_71_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_71_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [17, 1, 1, 1]> reduce_mean_71 = reduce_mean(axes = reduce_mean_71_axes_0, keep_dims = reduce_mean_71_keep_dims_0, x = square_23)[name = tensor<string, []>("reduce_mean_71")];
            tensor<fp32, []> add_46_y_0 = const()[name = tensor<string, []>("add_46_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [17, 1, 1, 1]> add_46 = add(x = reduce_mean_71, y = add_46_y_0)[name = tensor<string, []>("add_46")];
            tensor<fp32, [17, 1, 1, 1]> sqrt_23 = sqrt(x = add_46)[name = tensor<string, []>("sqrt_23")];
            tensor<fp32, [17, 1, 16, 476]> real_div_23 = real_div(x = sub_46, y = sqrt_23)[name = tensor<string, []>("real_div_23")];
            tensor<int32, [3]> reshape_93_shape_0 = const()[name = tensor<string, []>("reshape_93_shape_0"), val = tensor<int32, [3]>([17, 16, 476])];
            tensor<fp32, [17, 16, 476]> reshape_93 = reshape(shape = reshape_93_shape_0, x = real_div_23)[name = tensor<string, []>("reshape_93")];
            tensor<fp32, [1, 16, 1]> reshape_94 = const()[name = tensor<string, []>("reshape_94"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10697088)))];
            tensor<fp32, [17, 16, 476]> mul_23 = mul(x = reshape_93, y = reshape_94)[name = tensor<string, []>("mul_23")];
            tensor<fp32, [1, 16, 1]> reshape_95 = const()[name = tensor<string, []>("reshape_95"), val = tensor<fp32, [1, 16, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10697216)))];
            tensor<fp32, [17, 16, 476]> add_47 = add(x = mul_23, y = reshape_95)[name = tensor<string, []>("add_47")];
            tensor<fp32, [17, 16, 476]> input_157 = silu(x = add_47)[name = tensor<string, []>("input_157")];
            tensor<string, []> var_780_pad_type_0 = const()[name = tensor<string, []>("op_780_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_780_strides_0 = const()[name = tensor<string, []>("op_780_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_780_pad_0 = const()[name = tensor<string, []>("op_780_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_780_dilations_0 = const()[name = tensor<string, []>("op_780_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_780_groups_0 = const()[name = tensor<string, []>("op_780_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [17, 64, 476]> var_780 = conv(bias = encoder_1_conv_modules_2_layers_0_6_bias, dilations = var_780_dilations_0, groups = var_780_groups_0, pad = var_780_pad_0, pad_type = var_780_pad_type_0, strides = var_780_strides_0, weight = encoder_1_conv_modules_2_layers_0_6_weight, x = input_157)[name = tensor<string, []>("op_780")];
            tensor<fp32, [17, 64, 476]> var_781 = add(x = input_147, y = var_780)[name = tensor<string, []>("op_781")];
            tensor<int32, [4]> var_786 = const()[name = tensor<string, []>("op_786"), val = tensor<int32, [4]>([1, 17, 64, 476])];
            tensor<fp32, [1, 17, 64, 476]> var_787 = reshape(shape = var_786, x = var_781)[name = tensor<string, []>("op_787")];
            tensor<string, []> band_23_mode_0 = const()[name = tensor<string, []>("band_23_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 17, 64, 476]> band_23 = gelu(mode = band_23_mode_0, x = var_787)[name = tensor<string, []>("band_23")];
            tensor<bool, []> input_159_interleave_0 = const()[name = tensor<string, []>("input_159_interleave_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> const_130 = const()[name = tensor<string, []>("const_130"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 186, 64, 476]> input_159 = concat(axis = const_130, interleave = input_159_interleave_0, values = (band_19, band_21, band_23))[name = tensor<string, []>("input_159")];
            tensor<string, []> x_29_pad_type_0 = const()[name = tensor<string, []>("x_29_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [4]> x_29_pad_0 = const()[name = tensor<string, []>("x_29_pad_0"), val = tensor<int32, [4]>([1, 1, 1, 1])];
            tensor<int32, [2]> x_29_strides_0 = const()[name = tensor<string, []>("x_29_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> x_29_dilations_0 = const()[name = tensor<string, []>("x_29_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> x_29_groups_0 = const()[name = tensor<string, []>("x_29_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> transpose_26_perm_0 = const()[name = tensor<string, []>("transpose_26_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 64, 186, 476]> transpose_26 = transpose(perm = transpose_26_perm_0, x = input_159)[name = tensor<string, []>("transpose_115")];
            tensor<fp32, [1, 64, 186, 476]> x_29 = conv(bias = encoder_1_globalconv_bias, dilations = x_29_dilations_0, groups = x_29_groups_0, pad = x_29_pad_0, pad_type = x_29_pad_type_0, strides = x_29_strides_0, weight = encoder_1_globalconv_weight, x = transpose_26)[name = tensor<string, []>("x_29")];
            tensor<int32, [4]> var_849_begin_0 = const()[name = tensor<string, []>("op_849_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_849_end_0 = const()[name = tensor<string, []>("op_849_end_0"), val = tensor<int32, [4]>([1, 64, 33, 476])];
            tensor<bool, [4]> var_849_end_mask_0 = const()[name = tensor<string, []>("op_849_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 64, 33, 476]> var_849 = slice_by_index(begin = var_849_begin_0, end = var_849_end_0, end_mask = var_849_end_mask_0, x = x_29)[name = tensor<string, []>("op_849")];
            tensor<fp32, []> const_53 = const()[name = tensor<string, []>("const_53"), val = tensor<fp32, []>(0x0p+0)];
            tensor<int32, [8]> input_161_pad_0 = const()[name = tensor<string, []>("input_161_pad_0"), val = tensor<int32, [8]>([0, 0, 0, 0, 1, 1, 0, 0])];
            tensor<string, []> input_161_mode_0 = const()[name = tensor<string, []>("input_161_mode_0"), val = tensor<string, []>("constant")];
            tensor<fp32, [1, 64, 35, 476]> input_161 = pad(constant_val = const_53, mode = input_161_mode_0, pad = input_161_pad_0, x = var_849)[name = tensor<string, []>("input_161")];
            tensor<string, []> band_25_pad_type_0 = const()[name = tensor<string, []>("band_25_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_25_strides_0 = const()[name = tensor<string, []>("band_25_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [4]> band_25_pad_0 = const()[name = tensor<string, []>("band_25_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_25_dilations_0 = const()[name = tensor<string, []>("band_25_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_25_groups_0 = const()[name = tensor<string, []>("band_25_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 128, 33, 476]> band_25 = conv(bias = encoder_2_SDlayer_convs_0_bias, dilations = band_25_dilations_0, groups = band_25_groups_0, pad = band_25_pad_0, pad_type = band_25_pad_type_0, strides = band_25_strides_0, weight = encoder_2_SDlayer_convs_0_weight, x = input_161)[name = tensor<string, []>("band_25")];
            tensor<int32, [4]> var_862_begin_0 = const()[name = tensor<string, []>("op_862_begin_0"), val = tensor<int32, [4]>([0, 0, 33, 0])];
            tensor<int32, [4]> var_862_end_0 = const()[name = tensor<string, []>("op_862_end_0"), val = tensor<int32, [4]>([1, 64, 106, 476])];
            tensor<bool, [4]> var_862_end_mask_0 = const()[name = tensor<string, []>("op_862_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 64, 73, 476]> var_862 = slice_by_index(begin = var_862_begin_0, end = var_862_end_0, end_mask = var_862_end_mask_0, x = x_29)[name = tensor<string, []>("op_862")];
            tensor<fp32, []> const_55 = const()[name = tensor<string, []>("const_55"), val = tensor<fp32, []>(0x0p+0)];
            tensor<int32, [8]> input_163_pad_0 = const()[name = tensor<string, []>("input_163_pad_0"), val = tensor<int32, [8]>([0, 0, 0, 0, 1, 2, 0, 0])];
            tensor<string, []> input_163_mode_0 = const()[name = tensor<string, []>("input_163_mode_0"), val = tensor<string, []>("constant")];
            tensor<fp32, [1, 64, 76, 476]> input_163 = pad(constant_val = const_55, mode = input_163_mode_0, pad = input_163_pad_0, x = var_862)[name = tensor<string, []>("input_163")];
            tensor<string, []> band_27_pad_type_0 = const()[name = tensor<string, []>("band_27_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_27_strides_0 = const()[name = tensor<string, []>("band_27_strides_0"), val = tensor<int32, [2]>([4, 1])];
            tensor<int32, [4]> band_27_pad_0 = const()[name = tensor<string, []>("band_27_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_27_dilations_0 = const()[name = tensor<string, []>("band_27_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_27_groups_0 = const()[name = tensor<string, []>("band_27_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 128, 19, 476]> band_27 = conv(bias = encoder_2_SDlayer_convs_1_bias, dilations = band_27_dilations_0, groups = band_27_groups_0, pad = band_27_pad_0, pad_type = band_27_pad_type_0, strides = band_27_strides_0, weight = encoder_2_SDlayer_convs_1_weight, x = input_163)[name = tensor<string, []>("band_27")];
            tensor<int32, [4]> var_884_begin_0 = const()[name = tensor<string, []>("op_884_begin_0"), val = tensor<int32, [4]>([0, 0, 106, 0])];
            tensor<int32, [4]> var_884_end_0 = const()[name = tensor<string, []>("op_884_end_0"), val = tensor<int32, [4]>([1, 64, 1, 476])];
            tensor<bool, [4]> var_884_end_mask_0 = const()[name = tensor<string, []>("op_884_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 64, 80, 476]> var_884 = slice_by_index(begin = var_884_begin_0, end = var_884_end_0, end_mask = var_884_end_mask_0, x = x_29)[name = tensor<string, []>("op_884")];
            tensor<string, []> band_29_pad_type_0 = const()[name = tensor<string, []>("band_29_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> band_29_strides_0 = const()[name = tensor<string, []>("band_29_strides_0"), val = tensor<int32, [2]>([16, 1])];
            tensor<int32, [4]> band_29_pad_0 = const()[name = tensor<string, []>("band_29_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> band_29_dilations_0 = const()[name = tensor<string, []>("band_29_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> band_29_groups_0 = const()[name = tensor<string, []>("band_29_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 128, 5, 476]> band_29 = conv(bias = encoder_2_SDlayer_convs_2_bias, dilations = band_29_dilations_0, groups = band_29_groups_0, pad = band_29_pad_0, pad_type = band_29_pad_type_0, strides = band_29_strides_0, weight = encoder_2_SDlayer_convs_2_weight, x = var_884)[name = tensor<string, []>("band_29")];
            tensor<int32, [4]> var_911 = const()[name = tensor<string, []>("op_911"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_915 = const()[name = tensor<string, []>("op_915"), val = tensor<int32, [3]>([-1, 128, 476])];
            tensor<fp32, [1, 33, 128, 476]> var_912 = transpose(perm = var_911, x = band_25)[name = tensor<string, []>("transpose_114")];
            tensor<fp32, [33, 128, 476]> input_167 = reshape(shape = var_915, x = var_912)[name = tensor<string, []>("input_167")];
            tensor<int32, [4]> reshape_96_shape_0 = const()[name = tensor<string, []>("reshape_96_shape_0"), val = tensor<int32, [4]>([33, 1, 128, 476])];
            tensor<fp32, [33, 1, 128, 476]> reshape_96 = reshape(shape = reshape_96_shape_0, x = input_167)[name = tensor<string, []>("reshape_96")];
            tensor<int32, [2]> reduce_mean_72_axes_0 = const()[name = tensor<string, []>("reduce_mean_72_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_72_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_72_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_72 = reduce_mean(axes = reduce_mean_72_axes_0, keep_dims = reduce_mean_72_keep_dims_0, x = reshape_96)[name = tensor<string, []>("reduce_mean_72")];
            tensor<fp32, [33, 1, 128, 476]> sub_48 = sub(x = reshape_96, y = reduce_mean_72)[name = tensor<string, []>("sub_48")];
            tensor<fp32, [33, 1, 128, 476]> square_24 = square(x = sub_48)[name = tensor<string, []>("square_24")];
            tensor<int32, [2]> reduce_mean_74_axes_0 = const()[name = tensor<string, []>("reduce_mean_74_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_74_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_74_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_74 = reduce_mean(axes = reduce_mean_74_axes_0, keep_dims = reduce_mean_74_keep_dims_0, x = square_24)[name = tensor<string, []>("reduce_mean_74")];
            tensor<fp32, []> add_48_y_0 = const()[name = tensor<string, []>("add_48_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [33, 1, 1, 1]> add_48 = add(x = reduce_mean_74, y = add_48_y_0)[name = tensor<string, []>("add_48")];
            tensor<fp32, [33, 1, 1, 1]> sqrt_24 = sqrt(x = add_48)[name = tensor<string, []>("sqrt_24")];
            tensor<fp32, [33, 1, 128, 476]> real_div_24 = real_div(x = sub_48, y = sqrt_24)[name = tensor<string, []>("real_div_24")];
            tensor<int32, [3]> reshape_97_shape_0 = const()[name = tensor<string, []>("reshape_97_shape_0"), val = tensor<int32, [3]>([33, 128, 476])];
            tensor<fp32, [33, 128, 476]> reshape_97 = reshape(shape = reshape_97_shape_0, x = real_div_24)[name = tensor<string, []>("reshape_97")];
            tensor<fp32, [1, 128, 1]> reshape_98 = const()[name = tensor<string, []>("reshape_98"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10697344)))];
            tensor<fp32, [33, 128, 476]> mul_24 = mul(x = reshape_97, y = reshape_98)[name = tensor<string, []>("mul_24")];
            tensor<fp32, [1, 128, 1]> reshape_99 = const()[name = tensor<string, []>("reshape_99"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10697920)))];
            tensor<fp32, [33, 128, 476]> add_49 = add(x = mul_24, y = reshape_99)[name = tensor<string, []>("add_49")];
            tensor<string, []> input_171_pad_type_0 = const()[name = tensor<string, []>("input_171_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_171_pad_0 = const()[name = tensor<string, []>("input_171_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_171_strides_0 = const()[name = tensor<string, []>("input_171_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_171_dilations_0 = const()[name = tensor<string, []>("input_171_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_171_groups_0 = const()[name = tensor<string, []>("input_171_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 64, 476]> input_171 = conv(bias = encoder_2_conv_modules_0_layers_0_1_bias, dilations = input_171_dilations_0, groups = input_171_groups_0, pad = input_171_pad_0, pad_type = input_171_pad_type_0, strides = input_171_strides_0, weight = encoder_2_conv_modules_0_layers_0_1_weight, x = add_49)[name = tensor<string, []>("input_171")];
            tensor<int32, []> input_173_split_num_splits_0 = const()[name = tensor<string, []>("input_173_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_173_split_axis_0 = const()[name = tensor<string, []>("input_173_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 32, 476]> input_173_split_0, tensor<fp32, [33, 32, 476]> input_173_split_1 = split(axis = input_173_split_axis_0, num_splits = input_173_split_num_splits_0, x = input_171)[name = tensor<string, []>("input_173_split")];
            tensor<fp32, [33, 32, 476]> input_173_split_1_sigmoid = sigmoid(x = input_173_split_1)[name = tensor<string, []>("input_173_split_1_sigmoid")];
            tensor<fp32, [33, 32, 476]> input_173 = mul(x = input_173_split_0, y = input_173_split_1_sigmoid)[name = tensor<string, []>("input_173")];
            tensor<string, []> input_175_pad_type_0 = const()[name = tensor<string, []>("input_175_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_175_pad_0 = const()[name = tensor<string, []>("input_175_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_175_groups_0 = const()[name = tensor<string, []>("input_175_groups_0"), val = tensor<int32, []>(32)];
            tensor<int32, [1]> input_175_strides_0 = const()[name = tensor<string, []>("input_175_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_175_dilations_0 = const()[name = tensor<string, []>("input_175_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [33, 32, 476]> input_175 = conv(bias = encoder_2_conv_modules_0_layers_0_3_bias, dilations = input_175_dilations_0, groups = input_175_groups_0, pad = input_175_pad_0, pad_type = input_175_pad_type_0, strides = input_175_strides_0, weight = encoder_2_conv_modules_0_layers_0_3_weight, x = input_173)[name = tensor<string, []>("input_175")];
            tensor<int32, [4]> reshape_100_shape_0 = const()[name = tensor<string, []>("reshape_100_shape_0"), val = tensor<int32, [4]>([33, 1, 32, 476])];
            tensor<fp32, [33, 1, 32, 476]> reshape_100 = reshape(shape = reshape_100_shape_0, x = input_175)[name = tensor<string, []>("reshape_100")];
            tensor<int32, [2]> reduce_mean_75_axes_0 = const()[name = tensor<string, []>("reduce_mean_75_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_75_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_75_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_75 = reduce_mean(axes = reduce_mean_75_axes_0, keep_dims = reduce_mean_75_keep_dims_0, x = reshape_100)[name = tensor<string, []>("reduce_mean_75")];
            tensor<fp32, [33, 1, 32, 476]> sub_50 = sub(x = reshape_100, y = reduce_mean_75)[name = tensor<string, []>("sub_50")];
            tensor<fp32, [33, 1, 32, 476]> square_25 = square(x = sub_50)[name = tensor<string, []>("square_25")];
            tensor<int32, [2]> reduce_mean_77_axes_0 = const()[name = tensor<string, []>("reduce_mean_77_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_77_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_77_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_77 = reduce_mean(axes = reduce_mean_77_axes_0, keep_dims = reduce_mean_77_keep_dims_0, x = square_25)[name = tensor<string, []>("reduce_mean_77")];
            tensor<fp32, []> add_50_y_0 = const()[name = tensor<string, []>("add_50_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [33, 1, 1, 1]> add_50 = add(x = reduce_mean_77, y = add_50_y_0)[name = tensor<string, []>("add_50")];
            tensor<fp32, [33, 1, 1, 1]> sqrt_25 = sqrt(x = add_50)[name = tensor<string, []>("sqrt_25")];
            tensor<fp32, [33, 1, 32, 476]> real_div_25 = real_div(x = sub_50, y = sqrt_25)[name = tensor<string, []>("real_div_25")];
            tensor<int32, [3]> reshape_101_shape_0 = const()[name = tensor<string, []>("reshape_101_shape_0"), val = tensor<int32, [3]>([33, 32, 476])];
            tensor<fp32, [33, 32, 476]> reshape_101 = reshape(shape = reshape_101_shape_0, x = real_div_25)[name = tensor<string, []>("reshape_101")];
            tensor<fp32, [1, 32, 1]> reshape_102 = const()[name = tensor<string, []>("reshape_102"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10698496)))];
            tensor<fp32, [33, 32, 476]> mul_25 = mul(x = reshape_101, y = reshape_102)[name = tensor<string, []>("mul_25")];
            tensor<fp32, [1, 32, 1]> reshape_103 = const()[name = tensor<string, []>("reshape_103"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10698688)))];
            tensor<fp32, [33, 32, 476]> add_51 = add(x = mul_25, y = reshape_103)[name = tensor<string, []>("add_51")];
            tensor<fp32, [33, 32, 476]> input_177 = silu(x = add_51)[name = tensor<string, []>("input_177")];
            tensor<string, []> var_957_pad_type_0 = const()[name = tensor<string, []>("op_957_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_957_strides_0 = const()[name = tensor<string, []>("op_957_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_957_pad_0 = const()[name = tensor<string, []>("op_957_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_957_dilations_0 = const()[name = tensor<string, []>("op_957_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_957_groups_0 = const()[name = tensor<string, []>("op_957_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 128, 476]> var_957 = conv(bias = encoder_2_conv_modules_0_layers_0_6_bias, dilations = var_957_dilations_0, groups = var_957_groups_0, pad = var_957_pad_0, pad_type = var_957_pad_type_0, strides = var_957_strides_0, weight = encoder_2_conv_modules_0_layers_0_6_weight, x = input_177)[name = tensor<string, []>("op_957")];
            tensor<fp32, [33, 128, 476]> input_179 = add(x = input_167, y = var_957)[name = tensor<string, []>("input_179")];
            tensor<int32, [4]> reshape_104_shape_0 = const()[name = tensor<string, []>("reshape_104_shape_0"), val = tensor<int32, [4]>([33, 1, 128, 476])];
            tensor<fp32, [33, 1, 128, 476]> reshape_104 = reshape(shape = reshape_104_shape_0, x = input_179)[name = tensor<string, []>("reshape_104")];
            tensor<int32, [2]> reduce_mean_78_axes_0 = const()[name = tensor<string, []>("reduce_mean_78_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_78_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_78_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_78 = reduce_mean(axes = reduce_mean_78_axes_0, keep_dims = reduce_mean_78_keep_dims_0, x = reshape_104)[name = tensor<string, []>("reduce_mean_78")];
            tensor<fp32, [33, 1, 128, 476]> sub_52 = sub(x = reshape_104, y = reduce_mean_78)[name = tensor<string, []>("sub_52")];
            tensor<fp32, [33, 1, 128, 476]> square_26 = square(x = sub_52)[name = tensor<string, []>("square_26")];
            tensor<int32, [2]> reduce_mean_80_axes_0 = const()[name = tensor<string, []>("reduce_mean_80_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_80_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_80_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_80 = reduce_mean(axes = reduce_mean_80_axes_0, keep_dims = reduce_mean_80_keep_dims_0, x = square_26)[name = tensor<string, []>("reduce_mean_80")];
            tensor<fp32, []> add_52_y_0 = const()[name = tensor<string, []>("add_52_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [33, 1, 1, 1]> add_52 = add(x = reduce_mean_80, y = add_52_y_0)[name = tensor<string, []>("add_52")];
            tensor<fp32, [33, 1, 1, 1]> sqrt_26 = sqrt(x = add_52)[name = tensor<string, []>("sqrt_26")];
            tensor<fp32, [33, 1, 128, 476]> real_div_26 = real_div(x = sub_52, y = sqrt_26)[name = tensor<string, []>("real_div_26")];
            tensor<int32, [3]> reshape_105_shape_0 = const()[name = tensor<string, []>("reshape_105_shape_0"), val = tensor<int32, [3]>([33, 128, 476])];
            tensor<fp32, [33, 128, 476]> reshape_105 = reshape(shape = reshape_105_shape_0, x = real_div_26)[name = tensor<string, []>("reshape_105")];
            tensor<fp32, [1, 128, 1]> reshape_106 = const()[name = tensor<string, []>("reshape_106"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10698880)))];
            tensor<fp32, [33, 128, 476]> mul_26 = mul(x = reshape_105, y = reshape_106)[name = tensor<string, []>("mul_26")];
            tensor<fp32, [1, 128, 1]> reshape_107 = const()[name = tensor<string, []>("reshape_107"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10699456)))];
            tensor<fp32, [33, 128, 476]> add_53 = add(x = mul_26, y = reshape_107)[name = tensor<string, []>("add_53")];
            tensor<string, []> input_183_pad_type_0 = const()[name = tensor<string, []>("input_183_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_183_pad_0 = const()[name = tensor<string, []>("input_183_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_183_strides_0 = const()[name = tensor<string, []>("input_183_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_183_dilations_0 = const()[name = tensor<string, []>("input_183_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_183_groups_0 = const()[name = tensor<string, []>("input_183_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 64, 476]> input_183 = conv(bias = encoder_2_conv_modules_0_layers_1_1_bias, dilations = input_183_dilations_0, groups = input_183_groups_0, pad = input_183_pad_0, pad_type = input_183_pad_type_0, strides = input_183_strides_0, weight = encoder_2_conv_modules_0_layers_1_1_weight, x = add_53)[name = tensor<string, []>("input_183")];
            tensor<int32, []> input_185_split_num_splits_0 = const()[name = tensor<string, []>("input_185_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_185_split_axis_0 = const()[name = tensor<string, []>("input_185_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 32, 476]> input_185_split_0, tensor<fp32, [33, 32, 476]> input_185_split_1 = split(axis = input_185_split_axis_0, num_splits = input_185_split_num_splits_0, x = input_183)[name = tensor<string, []>("input_185_split")];
            tensor<fp32, [33, 32, 476]> input_185_split_1_sigmoid = sigmoid(x = input_185_split_1)[name = tensor<string, []>("input_185_split_1_sigmoid")];
            tensor<fp32, [33, 32, 476]> input_185 = mul(x = input_185_split_0, y = input_185_split_1_sigmoid)[name = tensor<string, []>("input_185")];
            tensor<string, []> input_187_pad_type_0 = const()[name = tensor<string, []>("input_187_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_187_pad_0 = const()[name = tensor<string, []>("input_187_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_187_groups_0 = const()[name = tensor<string, []>("input_187_groups_0"), val = tensor<int32, []>(32)];
            tensor<int32, [1]> input_187_strides_0 = const()[name = tensor<string, []>("input_187_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_187_dilations_0 = const()[name = tensor<string, []>("input_187_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [33, 32, 476]> input_187 = conv(bias = encoder_2_conv_modules_0_layers_1_3_bias, dilations = input_187_dilations_0, groups = input_187_groups_0, pad = input_187_pad_0, pad_type = input_187_pad_type_0, strides = input_187_strides_0, weight = encoder_2_conv_modules_0_layers_1_3_weight, x = input_185)[name = tensor<string, []>("input_187")];
            tensor<int32, [4]> reshape_108_shape_0 = const()[name = tensor<string, []>("reshape_108_shape_0"), val = tensor<int32, [4]>([33, 1, 32, 476])];
            tensor<fp32, [33, 1, 32, 476]> reshape_108 = reshape(shape = reshape_108_shape_0, x = input_187)[name = tensor<string, []>("reshape_108")];
            tensor<int32, [2]> reduce_mean_81_axes_0 = const()[name = tensor<string, []>("reduce_mean_81_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_81_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_81_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_81 = reduce_mean(axes = reduce_mean_81_axes_0, keep_dims = reduce_mean_81_keep_dims_0, x = reshape_108)[name = tensor<string, []>("reduce_mean_81")];
            tensor<fp32, [33, 1, 32, 476]> sub_54 = sub(x = reshape_108, y = reduce_mean_81)[name = tensor<string, []>("sub_54")];
            tensor<fp32, [33, 1, 32, 476]> square_27 = square(x = sub_54)[name = tensor<string, []>("square_27")];
            tensor<int32, [2]> reduce_mean_83_axes_0 = const()[name = tensor<string, []>("reduce_mean_83_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_83_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_83_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_83 = reduce_mean(axes = reduce_mean_83_axes_0, keep_dims = reduce_mean_83_keep_dims_0, x = square_27)[name = tensor<string, []>("reduce_mean_83")];
            tensor<fp32, []> add_54_y_0 = const()[name = tensor<string, []>("add_54_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [33, 1, 1, 1]> add_54 = add(x = reduce_mean_83, y = add_54_y_0)[name = tensor<string, []>("add_54")];
            tensor<fp32, [33, 1, 1, 1]> sqrt_27 = sqrt(x = add_54)[name = tensor<string, []>("sqrt_27")];
            tensor<fp32, [33, 1, 32, 476]> real_div_27 = real_div(x = sub_54, y = sqrt_27)[name = tensor<string, []>("real_div_27")];
            tensor<int32, [3]> reshape_109_shape_0 = const()[name = tensor<string, []>("reshape_109_shape_0"), val = tensor<int32, [3]>([33, 32, 476])];
            tensor<fp32, [33, 32, 476]> reshape_109 = reshape(shape = reshape_109_shape_0, x = real_div_27)[name = tensor<string, []>("reshape_109")];
            tensor<fp32, [1, 32, 1]> reshape_110 = const()[name = tensor<string, []>("reshape_110"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10700032)))];
            tensor<fp32, [33, 32, 476]> mul_27 = mul(x = reshape_109, y = reshape_110)[name = tensor<string, []>("mul_27")];
            tensor<fp32, [1, 32, 1]> reshape_111 = const()[name = tensor<string, []>("reshape_111"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10700224)))];
            tensor<fp32, [33, 32, 476]> add_55 = add(x = mul_27, y = reshape_111)[name = tensor<string, []>("add_55")];
            tensor<fp32, [33, 32, 476]> input_189 = silu(x = add_55)[name = tensor<string, []>("input_189")];
            tensor<string, []> var_993_pad_type_0 = const()[name = tensor<string, []>("op_993_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_993_strides_0 = const()[name = tensor<string, []>("op_993_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_993_pad_0 = const()[name = tensor<string, []>("op_993_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_993_dilations_0 = const()[name = tensor<string, []>("op_993_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_993_groups_0 = const()[name = tensor<string, []>("op_993_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 128, 476]> var_993 = conv(bias = encoder_2_conv_modules_0_layers_1_6_bias, dilations = var_993_dilations_0, groups = var_993_groups_0, pad = var_993_pad_0, pad_type = var_993_pad_type_0, strides = var_993_strides_0, weight = encoder_2_conv_modules_0_layers_1_6_weight, x = input_189)[name = tensor<string, []>("op_993")];
            tensor<fp32, [33, 128, 476]> input_191 = add(x = input_179, y = var_993)[name = tensor<string, []>("input_191")];
            tensor<int32, [4]> reshape_112_shape_0 = const()[name = tensor<string, []>("reshape_112_shape_0"), val = tensor<int32, [4]>([33, 1, 128, 476])];
            tensor<fp32, [33, 1, 128, 476]> reshape_112 = reshape(shape = reshape_112_shape_0, x = input_191)[name = tensor<string, []>("reshape_112")];
            tensor<int32, [2]> reduce_mean_84_axes_0 = const()[name = tensor<string, []>("reduce_mean_84_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_84_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_84_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_84 = reduce_mean(axes = reduce_mean_84_axes_0, keep_dims = reduce_mean_84_keep_dims_0, x = reshape_112)[name = tensor<string, []>("reduce_mean_84")];
            tensor<fp32, [33, 1, 128, 476]> sub_56 = sub(x = reshape_112, y = reduce_mean_84)[name = tensor<string, []>("sub_56")];
            tensor<fp32, [33, 1, 128, 476]> square_28 = square(x = sub_56)[name = tensor<string, []>("square_28")];
            tensor<int32, [2]> reduce_mean_86_axes_0 = const()[name = tensor<string, []>("reduce_mean_86_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_86_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_86_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_86 = reduce_mean(axes = reduce_mean_86_axes_0, keep_dims = reduce_mean_86_keep_dims_0, x = square_28)[name = tensor<string, []>("reduce_mean_86")];
            tensor<fp32, []> add_56_y_0 = const()[name = tensor<string, []>("add_56_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [33, 1, 1, 1]> add_56 = add(x = reduce_mean_86, y = add_56_y_0)[name = tensor<string, []>("add_56")];
            tensor<fp32, [33, 1, 1, 1]> sqrt_28 = sqrt(x = add_56)[name = tensor<string, []>("sqrt_28")];
            tensor<fp32, [33, 1, 128, 476]> real_div_28 = real_div(x = sub_56, y = sqrt_28)[name = tensor<string, []>("real_div_28")];
            tensor<int32, [3]> reshape_113_shape_0 = const()[name = tensor<string, []>("reshape_113_shape_0"), val = tensor<int32, [3]>([33, 128, 476])];
            tensor<fp32, [33, 128, 476]> reshape_113 = reshape(shape = reshape_113_shape_0, x = real_div_28)[name = tensor<string, []>("reshape_113")];
            tensor<fp32, [1, 128, 1]> reshape_114 = const()[name = tensor<string, []>("reshape_114"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10700416)))];
            tensor<fp32, [33, 128, 476]> mul_28 = mul(x = reshape_113, y = reshape_114)[name = tensor<string, []>("mul_28")];
            tensor<fp32, [1, 128, 1]> reshape_115 = const()[name = tensor<string, []>("reshape_115"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10700992)))];
            tensor<fp32, [33, 128, 476]> add_57 = add(x = mul_28, y = reshape_115)[name = tensor<string, []>("add_57")];
            tensor<string, []> input_195_pad_type_0 = const()[name = tensor<string, []>("input_195_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_195_pad_0 = const()[name = tensor<string, []>("input_195_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_195_strides_0 = const()[name = tensor<string, []>("input_195_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_195_dilations_0 = const()[name = tensor<string, []>("input_195_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_195_groups_0 = const()[name = tensor<string, []>("input_195_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 64, 476]> input_195 = conv(bias = encoder_2_conv_modules_0_layers_2_1_bias, dilations = input_195_dilations_0, groups = input_195_groups_0, pad = input_195_pad_0, pad_type = input_195_pad_type_0, strides = input_195_strides_0, weight = encoder_2_conv_modules_0_layers_2_1_weight, x = add_57)[name = tensor<string, []>("input_195")];
            tensor<int32, []> input_197_split_num_splits_0 = const()[name = tensor<string, []>("input_197_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_197_split_axis_0 = const()[name = tensor<string, []>("input_197_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 32, 476]> input_197_split_0, tensor<fp32, [33, 32, 476]> input_197_split_1 = split(axis = input_197_split_axis_0, num_splits = input_197_split_num_splits_0, x = input_195)[name = tensor<string, []>("input_197_split")];
            tensor<fp32, [33, 32, 476]> input_197_split_1_sigmoid = sigmoid(x = input_197_split_1)[name = tensor<string, []>("input_197_split_1_sigmoid")];
            tensor<fp32, [33, 32, 476]> input_197 = mul(x = input_197_split_0, y = input_197_split_1_sigmoid)[name = tensor<string, []>("input_197")];
            tensor<string, []> input_199_pad_type_0 = const()[name = tensor<string, []>("input_199_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_199_pad_0 = const()[name = tensor<string, []>("input_199_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_199_groups_0 = const()[name = tensor<string, []>("input_199_groups_0"), val = tensor<int32, []>(32)];
            tensor<int32, [1]> input_199_strides_0 = const()[name = tensor<string, []>("input_199_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_199_dilations_0 = const()[name = tensor<string, []>("input_199_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [33, 32, 476]> input_199 = conv(bias = encoder_2_conv_modules_0_layers_2_3_bias, dilations = input_199_dilations_0, groups = input_199_groups_0, pad = input_199_pad_0, pad_type = input_199_pad_type_0, strides = input_199_strides_0, weight = encoder_2_conv_modules_0_layers_2_3_weight, x = input_197)[name = tensor<string, []>("input_199")];
            tensor<int32, [4]> reshape_116_shape_0 = const()[name = tensor<string, []>("reshape_116_shape_0"), val = tensor<int32, [4]>([33, 1, 32, 476])];
            tensor<fp32, [33, 1, 32, 476]> reshape_116 = reshape(shape = reshape_116_shape_0, x = input_199)[name = tensor<string, []>("reshape_116")];
            tensor<int32, [2]> reduce_mean_87_axes_0 = const()[name = tensor<string, []>("reduce_mean_87_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_87_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_87_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_87 = reduce_mean(axes = reduce_mean_87_axes_0, keep_dims = reduce_mean_87_keep_dims_0, x = reshape_116)[name = tensor<string, []>("reduce_mean_87")];
            tensor<fp32, [33, 1, 32, 476]> sub_58 = sub(x = reshape_116, y = reduce_mean_87)[name = tensor<string, []>("sub_58")];
            tensor<fp32, [33, 1, 32, 476]> square_29 = square(x = sub_58)[name = tensor<string, []>("square_29")];
            tensor<int32, [2]> reduce_mean_89_axes_0 = const()[name = tensor<string, []>("reduce_mean_89_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_89_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_89_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [33, 1, 1, 1]> reduce_mean_89 = reduce_mean(axes = reduce_mean_89_axes_0, keep_dims = reduce_mean_89_keep_dims_0, x = square_29)[name = tensor<string, []>("reduce_mean_89")];
            tensor<fp32, []> add_58_y_0 = const()[name = tensor<string, []>("add_58_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [33, 1, 1, 1]> add_58 = add(x = reduce_mean_89, y = add_58_y_0)[name = tensor<string, []>("add_58")];
            tensor<fp32, [33, 1, 1, 1]> sqrt_29 = sqrt(x = add_58)[name = tensor<string, []>("sqrt_29")];
            tensor<fp32, [33, 1, 32, 476]> real_div_29 = real_div(x = sub_58, y = sqrt_29)[name = tensor<string, []>("real_div_29")];
            tensor<int32, [3]> reshape_117_shape_0 = const()[name = tensor<string, []>("reshape_117_shape_0"), val = tensor<int32, [3]>([33, 32, 476])];
            tensor<fp32, [33, 32, 476]> reshape_117 = reshape(shape = reshape_117_shape_0, x = real_div_29)[name = tensor<string, []>("reshape_117")];
            tensor<fp32, [1, 32, 1]> reshape_118 = const()[name = tensor<string, []>("reshape_118"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10701568)))];
            tensor<fp32, [33, 32, 476]> mul_29 = mul(x = reshape_117, y = reshape_118)[name = tensor<string, []>("mul_29")];
            tensor<fp32, [1, 32, 1]> reshape_119 = const()[name = tensor<string, []>("reshape_119"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10701760)))];
            tensor<fp32, [33, 32, 476]> add_59 = add(x = mul_29, y = reshape_119)[name = tensor<string, []>("add_59")];
            tensor<fp32, [33, 32, 476]> input_201 = silu(x = add_59)[name = tensor<string, []>("input_201")];
            tensor<string, []> var_1029_pad_type_0 = const()[name = tensor<string, []>("op_1029_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_1029_strides_0 = const()[name = tensor<string, []>("op_1029_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_1029_pad_0 = const()[name = tensor<string, []>("op_1029_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_1029_dilations_0 = const()[name = tensor<string, []>("op_1029_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_1029_groups_0 = const()[name = tensor<string, []>("op_1029_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [33, 128, 476]> var_1029 = conv(bias = encoder_2_conv_modules_0_layers_2_6_bias, dilations = var_1029_dilations_0, groups = var_1029_groups_0, pad = var_1029_pad_0, pad_type = var_1029_pad_type_0, strides = var_1029_strides_0, weight = encoder_2_conv_modules_0_layers_2_6_weight, x = input_201)[name = tensor<string, []>("op_1029")];
            tensor<fp32, [33, 128, 476]> var_1030 = add(x = input_191, y = var_1029)[name = tensor<string, []>("op_1030")];
            tensor<int32, [4]> var_1035 = const()[name = tensor<string, []>("op_1035"), val = tensor<int32, [4]>([1, 33, 128, 476])];
            tensor<fp32, [1, 33, 128, 476]> var_1036 = reshape(shape = var_1035, x = var_1030)[name = tensor<string, []>("op_1036")];
            tensor<string, []> band_31_mode_0 = const()[name = tensor<string, []>("band_31_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 33, 128, 476]> band_31 = gelu(mode = band_31_mode_0, x = var_1036)[name = tensor<string, []>("band_31")];
            tensor<int32, [4]> var_1040 = const()[name = tensor<string, []>("op_1040"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_1044 = const()[name = tensor<string, []>("op_1044"), val = tensor<int32, [3]>([-1, 128, 476])];
            tensor<fp32, [1, 19, 128, 476]> var_1041 = transpose(perm = var_1040, x = band_27)[name = tensor<string, []>("transpose_113")];
            tensor<fp32, [19, 128, 476]> input_203 = reshape(shape = var_1044, x = var_1041)[name = tensor<string, []>("input_203")];
            tensor<int32, [4]> reshape_120_shape_0 = const()[name = tensor<string, []>("reshape_120_shape_0"), val = tensor<int32, [4]>([19, 1, 128, 476])];
            tensor<fp32, [19, 1, 128, 476]> reshape_120 = reshape(shape = reshape_120_shape_0, x = input_203)[name = tensor<string, []>("reshape_120")];
            tensor<int32, [2]> reduce_mean_90_axes_0 = const()[name = tensor<string, []>("reduce_mean_90_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_90_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_90_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [19, 1, 1, 1]> reduce_mean_90 = reduce_mean(axes = reduce_mean_90_axes_0, keep_dims = reduce_mean_90_keep_dims_0, x = reshape_120)[name = tensor<string, []>("reduce_mean_90")];
            tensor<fp32, [19, 1, 128, 476]> sub_60 = sub(x = reshape_120, y = reduce_mean_90)[name = tensor<string, []>("sub_60")];
            tensor<fp32, [19, 1, 128, 476]> square_30 = square(x = sub_60)[name = tensor<string, []>("square_30")];
            tensor<int32, [2]> reduce_mean_92_axes_0 = const()[name = tensor<string, []>("reduce_mean_92_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_92_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_92_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [19, 1, 1, 1]> reduce_mean_92 = reduce_mean(axes = reduce_mean_92_axes_0, keep_dims = reduce_mean_92_keep_dims_0, x = square_30)[name = tensor<string, []>("reduce_mean_92")];
            tensor<fp32, []> add_60_y_0 = const()[name = tensor<string, []>("add_60_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [19, 1, 1, 1]> add_60 = add(x = reduce_mean_92, y = add_60_y_0)[name = tensor<string, []>("add_60")];
            tensor<fp32, [19, 1, 1, 1]> sqrt_30 = sqrt(x = add_60)[name = tensor<string, []>("sqrt_30")];
            tensor<fp32, [19, 1, 128, 476]> real_div_30 = real_div(x = sub_60, y = sqrt_30)[name = tensor<string, []>("real_div_30")];
            tensor<int32, [3]> reshape_121_shape_0 = const()[name = tensor<string, []>("reshape_121_shape_0"), val = tensor<int32, [3]>([19, 128, 476])];
            tensor<fp32, [19, 128, 476]> reshape_121 = reshape(shape = reshape_121_shape_0, x = real_div_30)[name = tensor<string, []>("reshape_121")];
            tensor<fp32, [1, 128, 1]> reshape_122 = const()[name = tensor<string, []>("reshape_122"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10701952)))];
            tensor<fp32, [19, 128, 476]> mul_30 = mul(x = reshape_121, y = reshape_122)[name = tensor<string, []>("mul_30")];
            tensor<fp32, [1, 128, 1]> reshape_123 = const()[name = tensor<string, []>("reshape_123"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10702528)))];
            tensor<fp32, [19, 128, 476]> add_61 = add(x = mul_30, y = reshape_123)[name = tensor<string, []>("add_61")];
            tensor<string, []> input_207_pad_type_0 = const()[name = tensor<string, []>("input_207_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_207_pad_0 = const()[name = tensor<string, []>("input_207_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_207_strides_0 = const()[name = tensor<string, []>("input_207_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_207_dilations_0 = const()[name = tensor<string, []>("input_207_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_207_groups_0 = const()[name = tensor<string, []>("input_207_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [19, 64, 476]> input_207 = conv(bias = encoder_2_conv_modules_1_layers_0_1_bias, dilations = input_207_dilations_0, groups = input_207_groups_0, pad = input_207_pad_0, pad_type = input_207_pad_type_0, strides = input_207_strides_0, weight = encoder_2_conv_modules_1_layers_0_1_weight, x = add_61)[name = tensor<string, []>("input_207")];
            tensor<int32, []> input_209_split_num_splits_0 = const()[name = tensor<string, []>("input_209_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_209_split_axis_0 = const()[name = tensor<string, []>("input_209_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [19, 32, 476]> input_209_split_0, tensor<fp32, [19, 32, 476]> input_209_split_1 = split(axis = input_209_split_axis_0, num_splits = input_209_split_num_splits_0, x = input_207)[name = tensor<string, []>("input_209_split")];
            tensor<fp32, [19, 32, 476]> input_209_split_1_sigmoid = sigmoid(x = input_209_split_1)[name = tensor<string, []>("input_209_split_1_sigmoid")];
            tensor<fp32, [19, 32, 476]> input_209 = mul(x = input_209_split_0, y = input_209_split_1_sigmoid)[name = tensor<string, []>("input_209")];
            tensor<string, []> input_211_pad_type_0 = const()[name = tensor<string, []>("input_211_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_211_pad_0 = const()[name = tensor<string, []>("input_211_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_211_groups_0 = const()[name = tensor<string, []>("input_211_groups_0"), val = tensor<int32, []>(32)];
            tensor<int32, [1]> input_211_strides_0 = const()[name = tensor<string, []>("input_211_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_211_dilations_0 = const()[name = tensor<string, []>("input_211_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [19, 32, 476]> input_211 = conv(bias = encoder_2_conv_modules_1_layers_0_3_bias, dilations = input_211_dilations_0, groups = input_211_groups_0, pad = input_211_pad_0, pad_type = input_211_pad_type_0, strides = input_211_strides_0, weight = encoder_2_conv_modules_1_layers_0_3_weight, x = input_209)[name = tensor<string, []>("input_211")];
            tensor<int32, [4]> reshape_124_shape_0 = const()[name = tensor<string, []>("reshape_124_shape_0"), val = tensor<int32, [4]>([19, 1, 32, 476])];
            tensor<fp32, [19, 1, 32, 476]> reshape_124 = reshape(shape = reshape_124_shape_0, x = input_211)[name = tensor<string, []>("reshape_124")];
            tensor<int32, [2]> reduce_mean_93_axes_0 = const()[name = tensor<string, []>("reduce_mean_93_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_93_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_93_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [19, 1, 1, 1]> reduce_mean_93 = reduce_mean(axes = reduce_mean_93_axes_0, keep_dims = reduce_mean_93_keep_dims_0, x = reshape_124)[name = tensor<string, []>("reduce_mean_93")];
            tensor<fp32, [19, 1, 32, 476]> sub_62 = sub(x = reshape_124, y = reduce_mean_93)[name = tensor<string, []>("sub_62")];
            tensor<fp32, [19, 1, 32, 476]> square_31 = square(x = sub_62)[name = tensor<string, []>("square_31")];
            tensor<int32, [2]> reduce_mean_95_axes_0 = const()[name = tensor<string, []>("reduce_mean_95_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_95_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_95_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [19, 1, 1, 1]> reduce_mean_95 = reduce_mean(axes = reduce_mean_95_axes_0, keep_dims = reduce_mean_95_keep_dims_0, x = square_31)[name = tensor<string, []>("reduce_mean_95")];
            tensor<fp32, []> add_62_y_0 = const()[name = tensor<string, []>("add_62_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [19, 1, 1, 1]> add_62 = add(x = reduce_mean_95, y = add_62_y_0)[name = tensor<string, []>("add_62")];
            tensor<fp32, [19, 1, 1, 1]> sqrt_31 = sqrt(x = add_62)[name = tensor<string, []>("sqrt_31")];
            tensor<fp32, [19, 1, 32, 476]> real_div_31 = real_div(x = sub_62, y = sqrt_31)[name = tensor<string, []>("real_div_31")];
            tensor<int32, [3]> reshape_125_shape_0 = const()[name = tensor<string, []>("reshape_125_shape_0"), val = tensor<int32, [3]>([19, 32, 476])];
            tensor<fp32, [19, 32, 476]> reshape_125 = reshape(shape = reshape_125_shape_0, x = real_div_31)[name = tensor<string, []>("reshape_125")];
            tensor<fp32, [1, 32, 1]> reshape_126 = const()[name = tensor<string, []>("reshape_126"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10703104)))];
            tensor<fp32, [19, 32, 476]> mul_31 = mul(x = reshape_125, y = reshape_126)[name = tensor<string, []>("mul_31")];
            tensor<fp32, [1, 32, 1]> reshape_127 = const()[name = tensor<string, []>("reshape_127"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10703296)))];
            tensor<fp32, [19, 32, 476]> add_63 = add(x = mul_31, y = reshape_127)[name = tensor<string, []>("add_63")];
            tensor<fp32, [19, 32, 476]> input_213 = silu(x = add_63)[name = tensor<string, []>("input_213")];
            tensor<string, []> var_1084_pad_type_0 = const()[name = tensor<string, []>("op_1084_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_1084_strides_0 = const()[name = tensor<string, []>("op_1084_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_1084_pad_0 = const()[name = tensor<string, []>("op_1084_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_1084_dilations_0 = const()[name = tensor<string, []>("op_1084_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_1084_groups_0 = const()[name = tensor<string, []>("op_1084_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [19, 128, 476]> var_1084 = conv(bias = encoder_2_conv_modules_1_layers_0_6_bias, dilations = var_1084_dilations_0, groups = var_1084_groups_0, pad = var_1084_pad_0, pad_type = var_1084_pad_type_0, strides = var_1084_strides_0, weight = encoder_2_conv_modules_1_layers_0_6_weight, x = input_213)[name = tensor<string, []>("op_1084")];
            tensor<fp32, [19, 128, 476]> input_215 = add(x = input_203, y = var_1084)[name = tensor<string, []>("input_215")];
            tensor<int32, [4]> reshape_128_shape_0 = const()[name = tensor<string, []>("reshape_128_shape_0"), val = tensor<int32, [4]>([19, 1, 128, 476])];
            tensor<fp32, [19, 1, 128, 476]> reshape_128 = reshape(shape = reshape_128_shape_0, x = input_215)[name = tensor<string, []>("reshape_128")];
            tensor<int32, [2]> reduce_mean_96_axes_0 = const()[name = tensor<string, []>("reduce_mean_96_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_96_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_96_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [19, 1, 1, 1]> reduce_mean_96 = reduce_mean(axes = reduce_mean_96_axes_0, keep_dims = reduce_mean_96_keep_dims_0, x = reshape_128)[name = tensor<string, []>("reduce_mean_96")];
            tensor<fp32, [19, 1, 128, 476]> sub_64 = sub(x = reshape_128, y = reduce_mean_96)[name = tensor<string, []>("sub_64")];
            tensor<fp32, [19, 1, 128, 476]> square_32 = square(x = sub_64)[name = tensor<string, []>("square_32")];
            tensor<int32, [2]> reduce_mean_98_axes_0 = const()[name = tensor<string, []>("reduce_mean_98_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_98_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_98_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [19, 1, 1, 1]> reduce_mean_98 = reduce_mean(axes = reduce_mean_98_axes_0, keep_dims = reduce_mean_98_keep_dims_0, x = square_32)[name = tensor<string, []>("reduce_mean_98")];
            tensor<fp32, []> add_64_y_0 = const()[name = tensor<string, []>("add_64_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [19, 1, 1, 1]> add_64 = add(x = reduce_mean_98, y = add_64_y_0)[name = tensor<string, []>("add_64")];
            tensor<fp32, [19, 1, 1, 1]> sqrt_32 = sqrt(x = add_64)[name = tensor<string, []>("sqrt_32")];
            tensor<fp32, [19, 1, 128, 476]> real_div_32 = real_div(x = sub_64, y = sqrt_32)[name = tensor<string, []>("real_div_32")];
            tensor<int32, [3]> reshape_129_shape_0 = const()[name = tensor<string, []>("reshape_129_shape_0"), val = tensor<int32, [3]>([19, 128, 476])];
            tensor<fp32, [19, 128, 476]> reshape_129 = reshape(shape = reshape_129_shape_0, x = real_div_32)[name = tensor<string, []>("reshape_129")];
            tensor<fp32, [1, 128, 1]> reshape_130 = const()[name = tensor<string, []>("reshape_130"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10703488)))];
            tensor<fp32, [19, 128, 476]> mul_32 = mul(x = reshape_129, y = reshape_130)[name = tensor<string, []>("mul_32")];
            tensor<fp32, [1, 128, 1]> reshape_131 = const()[name = tensor<string, []>("reshape_131"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10704064)))];
            tensor<fp32, [19, 128, 476]> add_65 = add(x = mul_32, y = reshape_131)[name = tensor<string, []>("add_65")];
            tensor<string, []> input_219_pad_type_0 = const()[name = tensor<string, []>("input_219_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_219_pad_0 = const()[name = tensor<string, []>("input_219_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_219_strides_0 = const()[name = tensor<string, []>("input_219_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_219_dilations_0 = const()[name = tensor<string, []>("input_219_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_219_groups_0 = const()[name = tensor<string, []>("input_219_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [19, 64, 476]> input_219 = conv(bias = encoder_2_conv_modules_1_layers_1_1_bias, dilations = input_219_dilations_0, groups = input_219_groups_0, pad = input_219_pad_0, pad_type = input_219_pad_type_0, strides = input_219_strides_0, weight = encoder_2_conv_modules_1_layers_1_1_weight, x = add_65)[name = tensor<string, []>("input_219")];
            tensor<int32, []> input_221_split_num_splits_0 = const()[name = tensor<string, []>("input_221_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_221_split_axis_0 = const()[name = tensor<string, []>("input_221_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [19, 32, 476]> input_221_split_0, tensor<fp32, [19, 32, 476]> input_221_split_1 = split(axis = input_221_split_axis_0, num_splits = input_221_split_num_splits_0, x = input_219)[name = tensor<string, []>("input_221_split")];
            tensor<fp32, [19, 32, 476]> input_221_split_1_sigmoid = sigmoid(x = input_221_split_1)[name = tensor<string, []>("input_221_split_1_sigmoid")];
            tensor<fp32, [19, 32, 476]> input_221 = mul(x = input_221_split_0, y = input_221_split_1_sigmoid)[name = tensor<string, []>("input_221")];
            tensor<string, []> input_223_pad_type_0 = const()[name = tensor<string, []>("input_223_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_223_pad_0 = const()[name = tensor<string, []>("input_223_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_223_groups_0 = const()[name = tensor<string, []>("input_223_groups_0"), val = tensor<int32, []>(32)];
            tensor<int32, [1]> input_223_strides_0 = const()[name = tensor<string, []>("input_223_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_223_dilations_0 = const()[name = tensor<string, []>("input_223_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [19, 32, 476]> input_223 = conv(bias = encoder_2_conv_modules_1_layers_1_3_bias, dilations = input_223_dilations_0, groups = input_223_groups_0, pad = input_223_pad_0, pad_type = input_223_pad_type_0, strides = input_223_strides_0, weight = encoder_2_conv_modules_1_layers_1_3_weight, x = input_221)[name = tensor<string, []>("input_223")];
            tensor<int32, [4]> reshape_132_shape_0 = const()[name = tensor<string, []>("reshape_132_shape_0"), val = tensor<int32, [4]>([19, 1, 32, 476])];
            tensor<fp32, [19, 1, 32, 476]> reshape_132 = reshape(shape = reshape_132_shape_0, x = input_223)[name = tensor<string, []>("reshape_132")];
            tensor<int32, [2]> reduce_mean_99_axes_0 = const()[name = tensor<string, []>("reduce_mean_99_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_99_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_99_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [19, 1, 1, 1]> reduce_mean_99 = reduce_mean(axes = reduce_mean_99_axes_0, keep_dims = reduce_mean_99_keep_dims_0, x = reshape_132)[name = tensor<string, []>("reduce_mean_99")];
            tensor<fp32, [19, 1, 32, 476]> sub_66 = sub(x = reshape_132, y = reduce_mean_99)[name = tensor<string, []>("sub_66")];
            tensor<fp32, [19, 1, 32, 476]> square_33 = square(x = sub_66)[name = tensor<string, []>("square_33")];
            tensor<int32, [2]> reduce_mean_101_axes_0 = const()[name = tensor<string, []>("reduce_mean_101_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_101_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_101_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [19, 1, 1, 1]> reduce_mean_101 = reduce_mean(axes = reduce_mean_101_axes_0, keep_dims = reduce_mean_101_keep_dims_0, x = square_33)[name = tensor<string, []>("reduce_mean_101")];
            tensor<fp32, []> add_66_y_0 = const()[name = tensor<string, []>("add_66_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [19, 1, 1, 1]> add_66 = add(x = reduce_mean_101, y = add_66_y_0)[name = tensor<string, []>("add_66")];
            tensor<fp32, [19, 1, 1, 1]> sqrt_33 = sqrt(x = add_66)[name = tensor<string, []>("sqrt_33")];
            tensor<fp32, [19, 1, 32, 476]> real_div_33 = real_div(x = sub_66, y = sqrt_33)[name = tensor<string, []>("real_div_33")];
            tensor<int32, [3]> reshape_133_shape_0 = const()[name = tensor<string, []>("reshape_133_shape_0"), val = tensor<int32, [3]>([19, 32, 476])];
            tensor<fp32, [19, 32, 476]> reshape_133 = reshape(shape = reshape_133_shape_0, x = real_div_33)[name = tensor<string, []>("reshape_133")];
            tensor<fp32, [1, 32, 1]> reshape_134 = const()[name = tensor<string, []>("reshape_134"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10704640)))];
            tensor<fp32, [19, 32, 476]> mul_33 = mul(x = reshape_133, y = reshape_134)[name = tensor<string, []>("mul_33")];
            tensor<fp32, [1, 32, 1]> reshape_135 = const()[name = tensor<string, []>("reshape_135"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10704832)))];
            tensor<fp32, [19, 32, 476]> add_67 = add(x = mul_33, y = reshape_135)[name = tensor<string, []>("add_67")];
            tensor<fp32, [19, 32, 476]> input_225 = silu(x = add_67)[name = tensor<string, []>("input_225")];
            tensor<string, []> var_1120_pad_type_0 = const()[name = tensor<string, []>("op_1120_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_1120_strides_0 = const()[name = tensor<string, []>("op_1120_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_1120_pad_0 = const()[name = tensor<string, []>("op_1120_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_1120_dilations_0 = const()[name = tensor<string, []>("op_1120_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_1120_groups_0 = const()[name = tensor<string, []>("op_1120_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [19, 128, 476]> var_1120 = conv(bias = encoder_2_conv_modules_1_layers_1_6_bias, dilations = var_1120_dilations_0, groups = var_1120_groups_0, pad = var_1120_pad_0, pad_type = var_1120_pad_type_0, strides = var_1120_strides_0, weight = encoder_2_conv_modules_1_layers_1_6_weight, x = input_225)[name = tensor<string, []>("op_1120")];
            tensor<fp32, [19, 128, 476]> var_1121 = add(x = input_215, y = var_1120)[name = tensor<string, []>("op_1121")];
            tensor<int32, [4]> var_1126 = const()[name = tensor<string, []>("op_1126"), val = tensor<int32, [4]>([1, 19, 128, 476])];
            tensor<fp32, [1, 19, 128, 476]> var_1127 = reshape(shape = var_1126, x = var_1121)[name = tensor<string, []>("op_1127")];
            tensor<string, []> band_33_mode_0 = const()[name = tensor<string, []>("band_33_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 19, 128, 476]> band_33 = gelu(mode = band_33_mode_0, x = var_1127)[name = tensor<string, []>("band_33")];
            tensor<int32, [4]> var_1131 = const()[name = tensor<string, []>("op_1131"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_1135 = const()[name = tensor<string, []>("op_1135"), val = tensor<int32, [3]>([-1, 128, 476])];
            tensor<fp32, [1, 5, 128, 476]> var_1132 = transpose(perm = var_1131, x = band_29)[name = tensor<string, []>("transpose_112")];
            tensor<fp32, [5, 128, 476]> input_227 = reshape(shape = var_1135, x = var_1132)[name = tensor<string, []>("input_227")];
            tensor<int32, [4]> reshape_136_shape_0 = const()[name = tensor<string, []>("reshape_136_shape_0"), val = tensor<int32, [4]>([5, 1, 128, 476])];
            tensor<fp32, [5, 1, 128, 476]> reshape_136 = reshape(shape = reshape_136_shape_0, x = input_227)[name = tensor<string, []>("reshape_136")];
            tensor<int32, [2]> reduce_mean_102_axes_0 = const()[name = tensor<string, []>("reduce_mean_102_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_102_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_102_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [5, 1, 1, 1]> reduce_mean_102 = reduce_mean(axes = reduce_mean_102_axes_0, keep_dims = reduce_mean_102_keep_dims_0, x = reshape_136)[name = tensor<string, []>("reduce_mean_102")];
            tensor<fp32, [5, 1, 128, 476]> sub_68 = sub(x = reshape_136, y = reduce_mean_102)[name = tensor<string, []>("sub_68")];
            tensor<fp32, [5, 1, 128, 476]> square_34 = square(x = sub_68)[name = tensor<string, []>("square_34")];
            tensor<int32, [2]> reduce_mean_104_axes_0 = const()[name = tensor<string, []>("reduce_mean_104_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_104_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_104_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [5, 1, 1, 1]> reduce_mean_104 = reduce_mean(axes = reduce_mean_104_axes_0, keep_dims = reduce_mean_104_keep_dims_0, x = square_34)[name = tensor<string, []>("reduce_mean_104")];
            tensor<fp32, []> add_68_y_0 = const()[name = tensor<string, []>("add_68_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [5, 1, 1, 1]> add_68 = add(x = reduce_mean_104, y = add_68_y_0)[name = tensor<string, []>("add_68")];
            tensor<fp32, [5, 1, 1, 1]> sqrt_34 = sqrt(x = add_68)[name = tensor<string, []>("sqrt_34")];
            tensor<fp32, [5, 1, 128, 476]> real_div_34 = real_div(x = sub_68, y = sqrt_34)[name = tensor<string, []>("real_div_34")];
            tensor<int32, [3]> reshape_137_shape_0 = const()[name = tensor<string, []>("reshape_137_shape_0"), val = tensor<int32, [3]>([5, 128, 476])];
            tensor<fp32, [5, 128, 476]> reshape_137 = reshape(shape = reshape_137_shape_0, x = real_div_34)[name = tensor<string, []>("reshape_137")];
            tensor<fp32, [1, 128, 1]> reshape_138 = const()[name = tensor<string, []>("reshape_138"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10705024)))];
            tensor<fp32, [5, 128, 476]> mul_34 = mul(x = reshape_137, y = reshape_138)[name = tensor<string, []>("mul_34")];
            tensor<fp32, [1, 128, 1]> reshape_139 = const()[name = tensor<string, []>("reshape_139"), val = tensor<fp32, [1, 128, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10705600)))];
            tensor<fp32, [5, 128, 476]> add_69 = add(x = mul_34, y = reshape_139)[name = tensor<string, []>("add_69")];
            tensor<string, []> input_231_pad_type_0 = const()[name = tensor<string, []>("input_231_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_231_pad_0 = const()[name = tensor<string, []>("input_231_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [1]> input_231_strides_0 = const()[name = tensor<string, []>("input_231_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_231_dilations_0 = const()[name = tensor<string, []>("input_231_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> input_231_groups_0 = const()[name = tensor<string, []>("input_231_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [5, 64, 476]> input_231 = conv(bias = encoder_2_conv_modules_2_layers_0_1_bias, dilations = input_231_dilations_0, groups = input_231_groups_0, pad = input_231_pad_0, pad_type = input_231_pad_type_0, strides = input_231_strides_0, weight = encoder_2_conv_modules_2_layers_0_1_weight, x = add_69)[name = tensor<string, []>("input_231")];
            tensor<int32, []> input_233_split_num_splits_0 = const()[name = tensor<string, []>("input_233_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> input_233_split_axis_0 = const()[name = tensor<string, []>("input_233_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [5, 32, 476]> input_233_split_0, tensor<fp32, [5, 32, 476]> input_233_split_1 = split(axis = input_233_split_axis_0, num_splits = input_233_split_num_splits_0, x = input_231)[name = tensor<string, []>("input_233_split")];
            tensor<fp32, [5, 32, 476]> input_233_split_1_sigmoid = sigmoid(x = input_233_split_1)[name = tensor<string, []>("input_233_split_1_sigmoid")];
            tensor<fp32, [5, 32, 476]> input_233 = mul(x = input_233_split_0, y = input_233_split_1_sigmoid)[name = tensor<string, []>("input_233")];
            tensor<string, []> input_235_pad_type_0 = const()[name = tensor<string, []>("input_235_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [2]> input_235_pad_0 = const()[name = tensor<string, []>("input_235_pad_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_235_groups_0 = const()[name = tensor<string, []>("input_235_groups_0"), val = tensor<int32, []>(32)];
            tensor<int32, [1]> input_235_strides_0 = const()[name = tensor<string, []>("input_235_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [1]> input_235_dilations_0 = const()[name = tensor<string, []>("input_235_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<fp32, [5, 32, 476]> input_235 = conv(bias = encoder_2_conv_modules_2_layers_0_3_bias, dilations = input_235_dilations_0, groups = input_235_groups_0, pad = input_235_pad_0, pad_type = input_235_pad_type_0, strides = input_235_strides_0, weight = encoder_2_conv_modules_2_layers_0_3_weight, x = input_233)[name = tensor<string, []>("input_235")];
            tensor<int32, [4]> reshape_140_shape_0 = const()[name = tensor<string, []>("reshape_140_shape_0"), val = tensor<int32, [4]>([5, 1, 32, 476])];
            tensor<fp32, [5, 1, 32, 476]> reshape_140 = reshape(shape = reshape_140_shape_0, x = input_235)[name = tensor<string, []>("reshape_140")];
            tensor<int32, [2]> reduce_mean_105_axes_0 = const()[name = tensor<string, []>("reduce_mean_105_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_105_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_105_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [5, 1, 1, 1]> reduce_mean_105 = reduce_mean(axes = reduce_mean_105_axes_0, keep_dims = reduce_mean_105_keep_dims_0, x = reshape_140)[name = tensor<string, []>("reduce_mean_105")];
            tensor<fp32, [5, 1, 32, 476]> sub_70 = sub(x = reshape_140, y = reduce_mean_105)[name = tensor<string, []>("sub_70")];
            tensor<fp32, [5, 1, 32, 476]> square_35 = square(x = sub_70)[name = tensor<string, []>("square_35")];
            tensor<int32, [2]> reduce_mean_107_axes_0 = const()[name = tensor<string, []>("reduce_mean_107_axes_0"), val = tensor<int32, [2]>([2, 3])];
            tensor<bool, []> reduce_mean_107_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_107_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [5, 1, 1, 1]> reduce_mean_107 = reduce_mean(axes = reduce_mean_107_axes_0, keep_dims = reduce_mean_107_keep_dims_0, x = square_35)[name = tensor<string, []>("reduce_mean_107")];
            tensor<fp32, []> add_70_y_0 = const()[name = tensor<string, []>("add_70_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [5, 1, 1, 1]> add_70 = add(x = reduce_mean_107, y = add_70_y_0)[name = tensor<string, []>("add_70")];
            tensor<fp32, [5, 1, 1, 1]> sqrt_35 = sqrt(x = add_70)[name = tensor<string, []>("sqrt_35")];
            tensor<fp32, [5, 1, 32, 476]> real_div_35 = real_div(x = sub_70, y = sqrt_35)[name = tensor<string, []>("real_div_35")];
            tensor<int32, [3]> reshape_141_shape_0 = const()[name = tensor<string, []>("reshape_141_shape_0"), val = tensor<int32, [3]>([5, 32, 476])];
            tensor<fp32, [5, 32, 476]> reshape_141 = reshape(shape = reshape_141_shape_0, x = real_div_35)[name = tensor<string, []>("reshape_141")];
            tensor<fp32, [1, 32, 1]> reshape_142 = const()[name = tensor<string, []>("reshape_142"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10706176)))];
            tensor<fp32, [5, 32, 476]> mul_35 = mul(x = reshape_141, y = reshape_142)[name = tensor<string, []>("mul_35")];
            tensor<fp32, [1, 32, 1]> reshape_143 = const()[name = tensor<string, []>("reshape_143"), val = tensor<fp32, [1, 32, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10706368)))];
            tensor<fp32, [5, 32, 476]> add_71 = add(x = mul_35, y = reshape_143)[name = tensor<string, []>("add_71")];
            tensor<fp32, [5, 32, 476]> input_237 = silu(x = add_71)[name = tensor<string, []>("input_237")];
            tensor<string, []> var_1173_pad_type_0 = const()[name = tensor<string, []>("op_1173_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [1]> var_1173_strides_0 = const()[name = tensor<string, []>("op_1173_strides_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, [2]> var_1173_pad_0 = const()[name = tensor<string, []>("op_1173_pad_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [1]> var_1173_dilations_0 = const()[name = tensor<string, []>("op_1173_dilations_0"), val = tensor<int32, [1]>([1])];
            tensor<int32, []> var_1173_groups_0 = const()[name = tensor<string, []>("op_1173_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [5, 128, 476]> var_1173 = conv(bias = encoder_2_conv_modules_2_layers_0_6_bias, dilations = var_1173_dilations_0, groups = var_1173_groups_0, pad = var_1173_pad_0, pad_type = var_1173_pad_type_0, strides = var_1173_strides_0, weight = encoder_2_conv_modules_2_layers_0_6_weight, x = input_237)[name = tensor<string, []>("op_1173")];
            tensor<fp32, [5, 128, 476]> var_1174 = add(x = input_227, y = var_1173)[name = tensor<string, []>("op_1174")];
            tensor<int32, [4]> var_1179 = const()[name = tensor<string, []>("op_1179"), val = tensor<int32, [4]>([1, 5, 128, 476])];
            tensor<fp32, [1, 5, 128, 476]> var_1180 = reshape(shape = var_1179, x = var_1174)[name = tensor<string, []>("op_1180")];
            tensor<string, []> band_mode_0 = const()[name = tensor<string, []>("band_mode_0"), val = tensor<string, []>("EXACT")];
            tensor<fp32, [1, 5, 128, 476]> band = gelu(mode = band_mode_0, x = var_1180)[name = tensor<string, []>("band")];
            tensor<bool, []> input_239_interleave_0 = const()[name = tensor<string, []>("input_239_interleave_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> const_131 = const()[name = tensor<string, []>("const_131"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 57, 128, 476]> input_239 = concat(axis = const_131, interleave = input_239_interleave_0, values = (band_31, band_33, band))[name = tensor<string, []>("input_239")];
            tensor<string, []> x_43_pad_type_0 = const()[name = tensor<string, []>("x_43_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [4]> x_43_pad_0 = const()[name = tensor<string, []>("x_43_pad_0"), val = tensor<int32, [4]>([1, 1, 1, 1])];
            tensor<int32, [2]> x_43_strides_0 = const()[name = tensor<string, []>("x_43_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> x_43_dilations_0 = const()[name = tensor<string, []>("x_43_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> x_43_groups_0 = const()[name = tensor<string, []>("x_43_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> transpose_28_perm_0 = const()[name = tensor<string, []>("transpose_28_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 128, 57, 476]> transpose_28 = transpose(perm = transpose_28_perm_0, x = input_239)[name = tensor<string, []>("transpose_111")];
            tensor<fp32, [1, 128, 57, 476]> x_43 = conv(bias = encoder_2_globalconv_bias, dilations = x_43_dilations_0, groups = x_43_groups_0, pad = x_43_pad_0, pad_type = x_43_pad_type_0, strides = x_43_strides_0, weight = encoder_2_globalconv_weight, x = transpose_28)[name = tensor<string, []>("x_43")];
            tensor<int32, []> var_1209 = const()[name = tensor<string, []>("op_1209"), val = tensor<int32, []>(1)];
            tensor<int32, [5]> reshape_144_shape_0 = const()[name = tensor<string, []>("reshape_144_shape_0"), val = tensor<int32, [5]>([1, 1, 128, 57, 476])];
            tensor<fp32, [1, 1, 128, 57, 476]> reshape_144 = reshape(shape = reshape_144_shape_0, x = x_43)[name = tensor<string, []>("reshape_144")];
            tensor<int32, [3]> reduce_mean_108_axes_0 = const()[name = tensor<string, []>("reduce_mean_108_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_108_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_108_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_108 = reduce_mean(axes = reduce_mean_108_axes_0, keep_dims = reduce_mean_108_keep_dims_0, x = reshape_144)[name = tensor<string, []>("reduce_mean_108")];
            tensor<fp32, [1, 1, 128, 57, 476]> sub_72 = sub(x = reshape_144, y = reduce_mean_108)[name = tensor<string, []>("sub_72")];
            tensor<fp32, [1, 1, 128, 57, 476]> square_36 = square(x = sub_72)[name = tensor<string, []>("square_36")];
            tensor<int32, [3]> reduce_mean_110_axes_0 = const()[name = tensor<string, []>("reduce_mean_110_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_110_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_110_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_110 = reduce_mean(axes = reduce_mean_110_axes_0, keep_dims = reduce_mean_110_keep_dims_0, x = square_36)[name = tensor<string, []>("reduce_mean_110")];
            tensor<fp32, []> add_72_y_0 = const()[name = tensor<string, []>("add_72_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_72 = add(x = reduce_mean_110, y = add_72_y_0)[name = tensor<string, []>("add_72")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_36 = sqrt(x = add_72)[name = tensor<string, []>("sqrt_36")];
            tensor<fp32, [1, 1, 128, 57, 476]> real_div_36 = real_div(x = sub_72, y = sqrt_36)[name = tensor<string, []>("real_div_36")];
            tensor<int32, [4]> reshape_145_shape_0 = const()[name = tensor<string, []>("reshape_145_shape_0"), val = tensor<int32, [4]>([1, 128, 57, 476])];
            tensor<fp32, [1, 128, 57, 476]> reshape_145 = reshape(shape = reshape_145_shape_0, x = real_div_36)[name = tensor<string, []>("reshape_145")];
            tensor<fp32, [128]> add_73_mean_0 = const()[name = tensor<string, []>("add_73_mean_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10706560)))];
            tensor<fp32, [128]> add_73_variance_0 = const()[name = tensor<string, []>("add_73_variance_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10707136)))];
            tensor<fp32, [128]> add_73_gamma_0 = const()[name = tensor<string, []>("add_73_gamma_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10707712)))];
            tensor<fp32, [128]> add_73_beta_0 = const()[name = tensor<string, []>("add_73_beta_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10708288)))];
            tensor<fp32, []> add_73_epsilon_0 = const()[name = tensor<string, []>("add_73_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 128, 57, 476]> add_73 = batch_norm(beta = add_73_beta_0, epsilon = add_73_epsilon_0, gamma = add_73_gamma_0, mean = add_73_mean_0, variance = add_73_variance_0, x = reshape_145)[name = tensor<string, []>("add_73")];
            tensor<int32, [4]> var_1254_perm_0 = const()[name = tensor<string, []>("op_1254_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<int32, [3]> var_1258 = const()[name = tensor<string, []>("op_1258"), val = tensor<int32, [3]>([476, 57, 128])];
            tensor<fp32, [1, 476, 57, 128]> var_1254 = transpose(perm = var_1254_perm_0, x = add_73)[name = tensor<string, []>("transpose_110")];
            tensor<fp32, [476, 57, 128]> input_241 = reshape(shape = var_1258, x = var_1254)[name = tensor<string, []>("input_241")];
            tensor<int32, [3]> input_241_batch_first_transpose_perm_0 = const()[name = tensor<string, []>("input_241_batch_first_transpose_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [512]> add_74 = const()[name = tensor<string, []>("add_74"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10708864)))];
            tensor<fp32, [512]> add_75 = const()[name = tensor<string, []>("add_75"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10710976)))];
            tensor<fp32, [512, 128]> concat_4 = const()[name = tensor<string, []>("concat_4"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10713088)))];
            tensor<fp32, [512, 128]> concat_5 = const()[name = tensor<string, []>("concat_5"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(10975296)))];
            tensor<fp32, [512, 128]> concat_6 = const()[name = tensor<string, []>("concat_6"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11237504)))];
            tensor<fp32, [512, 128]> concat_7 = const()[name = tensor<string, []>("concat_7"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11499712)))];
            tensor<fp32, [476, 256]> input_243_batch_first_lstm_h0_reshaped = const()[name = tensor<string, []>("input_243_batch_first_lstm_h0_reshaped"), val = tensor<fp32, [476, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(11761920)))];
            tensor<string, []> input_243_batch_first_direction_0 = const()[name = tensor<string, []>("input_243_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_243_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_243_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_243_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_243_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_243_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_243_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_243_batch_first_activation_0 = const()[name = tensor<string, []>("input_243_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [57, 476, 128]> input_241_batch_first_transpose = transpose(perm = input_241_batch_first_transpose_perm_0, x = input_241)[name = tensor<string, []>("transpose_109")];
            tensor<fp32, [57, 476, 256]> input_243_batch_first_0, tensor<fp32, [476, 256]> input_243_batch_first_1, tensor<fp32, [476, 256]> input_243_batch_first_2 = lstm(activation = input_243_batch_first_activation_0, bias = add_74, bias_back = add_75, cell_activation = input_243_batch_first_cell_activation_0, direction = input_243_batch_first_direction_0, initial_c = input_243_batch_first_lstm_h0_reshaped, initial_h = input_243_batch_first_lstm_h0_reshaped, output_sequence = input_243_batch_first_output_sequence_0, recurrent_activation = input_243_batch_first_recurrent_activation_0, weight_hh = concat_5, weight_hh_back = concat_7, weight_ih = concat_4, weight_ih_back = concat_6, x = input_241_batch_first_transpose)[name = tensor<string, []>("input_243_batch_first")];
            tensor<int32, [3]> input_243_perm_0 = const()[name = tensor<string, []>("input_243_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [476, 57, 256]> input_243 = transpose(perm = input_243_perm_0, x = input_243_batch_first_0)[name = tensor<string, []>("transpose_108")];
            tensor<fp32, [476, 57, 128]> x_47 = linear(bias = separation_net_dp_modules_0_linear_layers_0_bias, weight = separation_net_dp_modules_0_linear_layers_0_weight, x = input_243)[name = tensor<string, []>("linear_0")];
            tensor<int32, [4]> var_1281 = const()[name = tensor<string, []>("op_1281"), val = tensor<int32, [4]>([1, 476, 57, 128])];
            tensor<fp32, [1, 476, 57, 128]> var_1282 = reshape(shape = var_1281, x = x_47)[name = tensor<string, []>("op_1282")];
            tensor<int32, [4]> x_49_perm_0 = const()[name = tensor<string, []>("x_49_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<fp32, [1, 128, 57, 476]> x_49 = transpose(perm = x_49_perm_0, x = var_1282)[name = tensor<string, []>("transpose_107")];
            tensor<fp32, [1, 128, 57, 476]> input_245 = add(x = x_49, y = x_43)[name = tensor<string, []>("input_245")];
            tensor<int32, [5]> reshape_148_shape_0 = const()[name = tensor<string, []>("reshape_148_shape_0"), val = tensor<int32, [5]>([1, 1, 128, 57, 476])];
            tensor<fp32, [1, 1, 128, 57, 476]> reshape_148 = reshape(shape = reshape_148_shape_0, x = input_245)[name = tensor<string, []>("reshape_148")];
            tensor<int32, [3]> reduce_mean_111_axes_0 = const()[name = tensor<string, []>("reduce_mean_111_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_111_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_111_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_111 = reduce_mean(axes = reduce_mean_111_axes_0, keep_dims = reduce_mean_111_keep_dims_0, x = reshape_148)[name = tensor<string, []>("reduce_mean_111")];
            tensor<fp32, [1, 1, 128, 57, 476]> sub_74 = sub(x = reshape_148, y = reduce_mean_111)[name = tensor<string, []>("sub_74")];
            tensor<fp32, [1, 1, 128, 57, 476]> square_37 = square(x = sub_74)[name = tensor<string, []>("square_37")];
            tensor<int32, [3]> reduce_mean_113_axes_0 = const()[name = tensor<string, []>("reduce_mean_113_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_113_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_113_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_113 = reduce_mean(axes = reduce_mean_113_axes_0, keep_dims = reduce_mean_113_keep_dims_0, x = square_37)[name = tensor<string, []>("reduce_mean_113")];
            tensor<fp32, []> add_76_y_0 = const()[name = tensor<string, []>("add_76_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_76 = add(x = reduce_mean_113, y = add_76_y_0)[name = tensor<string, []>("add_76")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_37 = sqrt(x = add_76)[name = tensor<string, []>("sqrt_37")];
            tensor<fp32, [1, 1, 128, 57, 476]> real_div_37 = real_div(x = sub_74, y = sqrt_37)[name = tensor<string, []>("real_div_37")];
            tensor<int32, [4]> reshape_149_shape_0 = const()[name = tensor<string, []>("reshape_149_shape_0"), val = tensor<int32, [4]>([1, 128, 57, 476])];
            tensor<fp32, [1, 128, 57, 476]> reshape_149 = reshape(shape = reshape_149_shape_0, x = real_div_37)[name = tensor<string, []>("reshape_149")];
            tensor<fp32, [128]> add_77_gamma_0 = const()[name = tensor<string, []>("add_77_gamma_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12249408)))];
            tensor<fp32, [128]> add_77_beta_0 = const()[name = tensor<string, []>("add_77_beta_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12249984)))];
            tensor<fp32, []> add_77_epsilon_0 = const()[name = tensor<string, []>("add_77_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 128, 57, 476]> add_77 = batch_norm(beta = add_77_beta_0, epsilon = add_77_epsilon_0, gamma = add_77_gamma_0, mean = add_73_mean_0, variance = add_73_variance_0, x = reshape_149)[name = tensor<string, []>("add_77")];
            tensor<int32, [4]> var_1288_perm_0 = const()[name = tensor<string, []>("op_1288_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_1292 = const()[name = tensor<string, []>("op_1292"), val = tensor<int32, [3]>([57, 128, 476])];
            tensor<fp32, [1, 57, 128, 476]> var_1288 = transpose(perm = var_1288_perm_0, x = add_77)[name = tensor<string, []>("transpose_106")];
            tensor<fp32, [57, 128, 476]> var_1293 = reshape(shape = var_1292, x = var_1288)[name = tensor<string, []>("op_1293")];
            tensor<int32, [3]> transpose_31_perm_0 = const()[name = tensor<string, []>("transpose_31_perm_0"), val = tensor<int32, [3]>([2, 0, 1])];
            tensor<fp32, [512]> add_78 = const()[name = tensor<string, []>("add_78"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12250560)))];
            tensor<fp32, [512]> add_79 = const()[name = tensor<string, []>("add_79"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12252672)))];
            tensor<fp32, [512, 128]> concat_14 = const()[name = tensor<string, []>("concat_14"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12254784)))];
            tensor<fp32, [512, 128]> concat_15 = const()[name = tensor<string, []>("concat_15"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12516992)))];
            tensor<fp32, [512, 128]> concat_16 = const()[name = tensor<string, []>("concat_16"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(12779200)))];
            tensor<fp32, [512, 128]> concat_17 = const()[name = tensor<string, []>("concat_17"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13041408)))];
            tensor<fp32, [57, 256]> input_249_batch_first_lstm_h0_reshaped = const()[name = tensor<string, []>("input_249_batch_first_lstm_h0_reshaped"), val = tensor<fp32, [57, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13303616)))];
            tensor<string, []> input_249_batch_first_direction_0 = const()[name = tensor<string, []>("input_249_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_249_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_249_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_249_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_249_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_249_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_249_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_249_batch_first_activation_0 = const()[name = tensor<string, []>("input_249_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [476, 57, 128]> transpose_31 = transpose(perm = transpose_31_perm_0, x = var_1293)[name = tensor<string, []>("transpose_105")];
            tensor<fp32, [476, 57, 256]> input_249_batch_first_0, tensor<fp32, [57, 256]> input_249_batch_first_1, tensor<fp32, [57, 256]> input_249_batch_first_2 = lstm(activation = input_249_batch_first_activation_0, bias = add_78, bias_back = add_79, cell_activation = input_249_batch_first_cell_activation_0, direction = input_249_batch_first_direction_0, initial_c = input_249_batch_first_lstm_h0_reshaped, initial_h = input_249_batch_first_lstm_h0_reshaped, output_sequence = input_249_batch_first_output_sequence_0, recurrent_activation = input_249_batch_first_recurrent_activation_0, weight_hh = concat_15, weight_hh_back = concat_17, weight_ih = concat_14, weight_ih_back = concat_16, x = transpose_31)[name = tensor<string, []>("input_249_batch_first")];
            tensor<int32, [3]> input_249_perm_0 = const()[name = tensor<string, []>("input_249_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [57, 476, 256]> input_249 = transpose(perm = input_249_perm_0, x = input_249_batch_first_0)[name = tensor<string, []>("transpose_104")];
            tensor<fp32, [57, 476, 128]> x_53 = linear(bias = separation_net_dp_modules_0_linear_layers_1_bias, weight = separation_net_dp_modules_0_linear_layers_1_weight, x = input_249)[name = tensor<string, []>("linear_1")];
            tensor<int32, [3]> var_1316_perm_0 = const()[name = tensor<string, []>("op_1316_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
            tensor<int32, [4]> var_1318 = const()[name = tensor<string, []>("op_1318"), val = tensor<int32, [4]>([1, 57, 128, 476])];
            tensor<fp32, [57, 128, 476]> var_1316 = transpose(perm = var_1316_perm_0, x = x_53)[name = tensor<string, []>("transpose_103")];
            tensor<fp32, [1, 57, 128, 476]> var_1319 = reshape(shape = var_1318, x = var_1316)[name = tensor<string, []>("op_1319")];
            tensor<int32, [4]> x_55_perm_0 = const()[name = tensor<string, []>("x_55_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 128, 57, 476]> x_55 = transpose(perm = x_55_perm_0, x = var_1319)[name = tensor<string, []>("transpose_102")];
            tensor<fp32, [1, 128, 57, 476]> x_57 = add(x = x_55, y = input_245)[name = tensor<string, []>("x_57")];
            tensor<int32, [4]> transpose_0_perm_0 = const()[name = tensor<string, []>("transpose_0_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [2]> reshape_192_shape_0 = const()[name = tensor<string, []>("reshape_192_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_0 = transpose(perm = transpose_0_perm_0, x = x_57)[name = tensor<string, []>("transpose_101")];
            tensor<fp32, [476, 7296]> reshape_192 = reshape(shape = reshape_192_shape_0, x = transpose_0)[name = tensor<string, []>("reshape_192")];
            tensor<fp32, [476, 476]> cos_0 = const()[name = tensor<string, []>("cos_0"), val = tensor<fp32, [476, 476]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(13362048)))];
            tensor<fp32, [476, 476]> sin_0 = const()[name = tensor<string, []>("sin_0"), val = tensor<fp32, [476, 476]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(14268416)))];
            tensor<bool, []> matmul_1_transpose_x_0 = const()[name = tensor<string, []>("matmul_1_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_1_transpose_y_0 = const()[name = tensor<string, []>("matmul_1_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_1 = matmul(transpose_x = matmul_1_transpose_x_0, transpose_y = matmul_1_transpose_y_0, x = cos_0, y = reshape_192)[name = tensor<string, []>("matmul_1")];
            tensor<bool, []> matmul_3_transpose_x_0 = const()[name = tensor<string, []>("matmul_3_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_3_transpose_y_0 = const()[name = tensor<string, []>("matmul_3_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_3 = matmul(transpose_x = matmul_3_transpose_x_0, transpose_y = matmul_3_transpose_y_0, x = sin_0, y = reshape_192)[name = tensor<string, []>("matmul_3")];
            tensor<fp32, []> mul_49_y_0 = const()[name = tensor<string, []>("mul_49_y_0"), val = tensor<fp32, []>(-0x1p+0)];
            tensor<fp32, [476, 7296]> mul_49 = mul(x = matmul_3, y = mul_49_y_0)[name = tensor<string, []>("mul_49")];
            tensor<int32, [4]> reshape_196_shape_0 = const()[name = tensor<string, []>("reshape_196_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_196 = reshape(shape = reshape_196_shape_0, x = matmul_1)[name = tensor<string, []>("reshape_196")];
            tensor<int32, [4]> reshape_197_shape_0 = const()[name = tensor<string, []>("reshape_197_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_197 = reshape(shape = reshape_197_shape_0, x = mul_49)[name = tensor<string, []>("reshape_197")];
            tensor<int32, [4]> transpose_2_perm_0 = const()[name = tensor<string, []>("transpose_2_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [4]> transpose_3_perm_0 = const()[name = tensor<string, []>("transpose_3_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<fp32, []> _inversed_real_div_49_y_0 = const()[name = tensor<string, []>("_inversed_real_div_49_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [1, 128, 57, 476]> transpose_2 = transpose(perm = transpose_2_perm_0, x = reshape_196)[name = tensor<string, []>("transpose_100")];
            tensor<fp32, [1, 128, 57, 476]> _inversed_real_div_49 = mul(x = transpose_2, y = _inversed_real_div_49_y_0)[name = tensor<string, []>("_inversed_real_div_49")];
            tensor<fp32, []> _inversed_real_div_50_y_0 = const()[name = tensor<string, []>("_inversed_real_div_50_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [1, 128, 57, 476]> transpose_3 = transpose(perm = transpose_3_perm_0, x = reshape_197)[name = tensor<string, []>("transpose_99")];
            tensor<fp32, [1, 128, 57, 476]> _inversed_real_div_50 = mul(x = transpose_3, y = _inversed_real_div_50_y_0)[name = tensor<string, []>("_inversed_real_div_50")];
            tensor<int32, []> gather_0_batch_dims_0 = const()[name = tensor<string, []>("gather_0_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_0_validate_indices_0 = const()[name = tensor<string, []>("gather_0_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, [239]> select_0 = const()[name = tensor<string, []>("select_0"), val = tensor<int32, [239]>([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238])];
            tensor<int32, []> gather_0_axis_1 = const()[name = tensor<string, []>("gather_0_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 239]> gather_0 = gather(axis = gather_0_axis_1, batch_dims = gather_0_batch_dims_0, indices = select_0, validate_indices = gather_0_validate_indices_0, x = _inversed_real_div_49)[name = tensor<string, []>("gather_0")];
            tensor<int32, []> gather_1_batch_dims_0 = const()[name = tensor<string, []>("gather_1_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_1_validate_indices_0 = const()[name = tensor<string, []>("gather_1_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_1_axis_1 = const()[name = tensor<string, []>("gather_1_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 239]> gather_1 = gather(axis = gather_1_axis_1, batch_dims = gather_1_batch_dims_0, indices = select_0, validate_indices = gather_1_validate_indices_0, x = _inversed_real_div_50)[name = tensor<string, []>("gather_1")];
            tensor<bool, []> x_63_interleave_0 = const()[name = tensor<string, []>("x_63_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 256, 57, 239]> x_63 = concat(axis = var_1209, interleave = x_63_interleave_0, values = (gather_0, gather_1))[name = tensor<string, []>("x_63")];
            tensor<int32, [5]> reshape_152_shape_0 = const()[name = tensor<string, []>("reshape_152_shape_0"), val = tensor<int32, [5]>([1, 1, 256, 57, 239])];
            tensor<fp32, [1, 1, 256, 57, 239]> reshape_152 = reshape(shape = reshape_152_shape_0, x = x_63)[name = tensor<string, []>("reshape_152")];
            tensor<int32, [3]> reduce_mean_114_axes_0 = const()[name = tensor<string, []>("reduce_mean_114_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_114_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_114_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_114 = reduce_mean(axes = reduce_mean_114_axes_0, keep_dims = reduce_mean_114_keep_dims_0, x = reshape_152)[name = tensor<string, []>("reduce_mean_114")];
            tensor<fp32, [1, 1, 256, 57, 239]> sub_76 = sub(x = reshape_152, y = reduce_mean_114)[name = tensor<string, []>("sub_76")];
            tensor<fp32, [1, 1, 256, 57, 239]> square_38 = square(x = sub_76)[name = tensor<string, []>("square_38")];
            tensor<int32, [3]> reduce_mean_116_axes_0 = const()[name = tensor<string, []>("reduce_mean_116_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_116_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_116_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_116 = reduce_mean(axes = reduce_mean_116_axes_0, keep_dims = reduce_mean_116_keep_dims_0, x = square_38)[name = tensor<string, []>("reduce_mean_116")];
            tensor<fp32, []> add_80_y_0 = const()[name = tensor<string, []>("add_80_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_80 = add(x = reduce_mean_116, y = add_80_y_0)[name = tensor<string, []>("add_80")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_38 = sqrt(x = add_80)[name = tensor<string, []>("sqrt_38")];
            tensor<fp32, [1, 1, 256, 57, 239]> real_div_38 = real_div(x = sub_76, y = sqrt_38)[name = tensor<string, []>("real_div_38")];
            tensor<int32, [4]> reshape_153_shape_0 = const()[name = tensor<string, []>("reshape_153_shape_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<fp32, [1, 256, 57, 239]> reshape_153 = reshape(shape = reshape_153_shape_0, x = real_div_38)[name = tensor<string, []>("reshape_153")];
            tensor<fp32, [256]> add_81_mean_0 = const()[name = tensor<string, []>("add_81_mean_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15174784)))];
            tensor<fp32, [256]> add_81_variance_0 = const()[name = tensor<string, []>("add_81_variance_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15175872)))];
            tensor<fp32, [256]> add_81_gamma_0 = const()[name = tensor<string, []>("add_81_gamma_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15176960)))];
            tensor<fp32, [256]> add_81_beta_0 = const()[name = tensor<string, []>("add_81_beta_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15178048)))];
            tensor<fp32, []> add_81_epsilon_0 = const()[name = tensor<string, []>("add_81_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 256, 57, 239]> add_81 = batch_norm(beta = add_81_beta_0, epsilon = add_81_epsilon_0, gamma = add_81_gamma_0, mean = add_81_mean_0, variance = add_81_variance_0, x = reshape_153)[name = tensor<string, []>("add_81")];
            tensor<int32, [4]> var_1350_perm_0 = const()[name = tensor<string, []>("op_1350_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<int32, [3]> var_1354 = const()[name = tensor<string, []>("op_1354"), val = tensor<int32, [3]>([239, 57, 256])];
            tensor<fp32, [1, 239, 57, 256]> var_1350 = transpose(perm = var_1350_perm_0, x = add_81)[name = tensor<string, []>("transpose_98")];
            tensor<fp32, [239, 57, 256]> input_251 = reshape(shape = var_1354, x = var_1350)[name = tensor<string, []>("input_251")];
            tensor<int32, [3]> input_251_batch_first_transpose_perm_0 = const()[name = tensor<string, []>("input_251_batch_first_transpose_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [1024]> add_82 = const()[name = tensor<string, []>("add_82"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15179136)))];
            tensor<fp32, [1024]> add_83 = const()[name = tensor<string, []>("add_83"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15183296)))];
            tensor<fp32, [1024, 256]> concat_24 = const()[name = tensor<string, []>("concat_24"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(15187456)))];
            tensor<fp32, [1024, 256]> concat_25 = const()[name = tensor<string, []>("concat_25"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(16236096)))];
            tensor<fp32, [1024, 256]> concat_26 = const()[name = tensor<string, []>("concat_26"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(17284736)))];
            tensor<fp32, [1024, 256]> concat_27 = const()[name = tensor<string, []>("concat_27"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(18333376)))];
            tensor<fp32, [239, 512]> input_253_batch_first_lstm_h0_reshaped = const()[name = tensor<string, []>("input_253_batch_first_lstm_h0_reshaped"), val = tensor<fp32, [239, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(19382016)))];
            tensor<string, []> input_253_batch_first_direction_0 = const()[name = tensor<string, []>("input_253_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_253_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_253_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_253_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_253_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_253_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_253_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_253_batch_first_activation_0 = const()[name = tensor<string, []>("input_253_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [57, 239, 256]> input_251_batch_first_transpose = transpose(perm = input_251_batch_first_transpose_perm_0, x = input_251)[name = tensor<string, []>("transpose_97")];
            tensor<fp32, [57, 239, 512]> input_253_batch_first_0, tensor<fp32, [239, 512]> input_253_batch_first_1, tensor<fp32, [239, 512]> input_253_batch_first_2 = lstm(activation = input_253_batch_first_activation_0, bias = add_82, bias_back = add_83, cell_activation = input_253_batch_first_cell_activation_0, direction = input_253_batch_first_direction_0, initial_c = input_253_batch_first_lstm_h0_reshaped, initial_h = input_253_batch_first_lstm_h0_reshaped, output_sequence = input_253_batch_first_output_sequence_0, recurrent_activation = input_253_batch_first_recurrent_activation_0, weight_hh = concat_25, weight_hh_back = concat_27, weight_ih = concat_24, weight_ih_back = concat_26, x = input_251_batch_first_transpose)[name = tensor<string, []>("input_253_batch_first")];
            tensor<int32, [3]> input_253_perm_0 = const()[name = tensor<string, []>("input_253_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [239, 57, 512]> input_253 = transpose(perm = input_253_perm_0, x = input_253_batch_first_0)[name = tensor<string, []>("transpose_96")];
            tensor<fp32, [239, 57, 256]> x_67 = linear(bias = separation_net_dp_modules_1_linear_layers_0_bias, weight = separation_net_dp_modules_1_linear_layers_0_weight, x = input_253)[name = tensor<string, []>("linear_2")];
            tensor<int32, [4]> var_1377 = const()[name = tensor<string, []>("op_1377"), val = tensor<int32, [4]>([1, 239, 57, 256])];
            tensor<fp32, [1, 239, 57, 256]> var_1378 = reshape(shape = var_1377, x = x_67)[name = tensor<string, []>("op_1378")];
            tensor<int32, [4]> x_69_perm_0 = const()[name = tensor<string, []>("x_69_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<fp32, [1, 256, 57, 239]> x_69 = transpose(perm = x_69_perm_0, x = var_1378)[name = tensor<string, []>("transpose_95")];
            tensor<fp32, [1, 256, 57, 239]> input_255 = add(x = x_69, y = x_63)[name = tensor<string, []>("input_255")];
            tensor<int32, [5]> reshape_156_shape_0 = const()[name = tensor<string, []>("reshape_156_shape_0"), val = tensor<int32, [5]>([1, 1, 256, 57, 239])];
            tensor<fp32, [1, 1, 256, 57, 239]> reshape_156 = reshape(shape = reshape_156_shape_0, x = input_255)[name = tensor<string, []>("reshape_156")];
            tensor<int32, [3]> reduce_mean_117_axes_0 = const()[name = tensor<string, []>("reduce_mean_117_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_117_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_117_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_117 = reduce_mean(axes = reduce_mean_117_axes_0, keep_dims = reduce_mean_117_keep_dims_0, x = reshape_156)[name = tensor<string, []>("reduce_mean_117")];
            tensor<fp32, [1, 1, 256, 57, 239]> sub_78 = sub(x = reshape_156, y = reduce_mean_117)[name = tensor<string, []>("sub_78")];
            tensor<fp32, [1, 1, 256, 57, 239]> square_39 = square(x = sub_78)[name = tensor<string, []>("square_39")];
            tensor<int32, [3]> reduce_mean_119_axes_0 = const()[name = tensor<string, []>("reduce_mean_119_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_119_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_119_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_119 = reduce_mean(axes = reduce_mean_119_axes_0, keep_dims = reduce_mean_119_keep_dims_0, x = square_39)[name = tensor<string, []>("reduce_mean_119")];
            tensor<fp32, []> add_84_y_0 = const()[name = tensor<string, []>("add_84_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_84 = add(x = reduce_mean_119, y = add_84_y_0)[name = tensor<string, []>("add_84")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_39 = sqrt(x = add_84)[name = tensor<string, []>("sqrt_39")];
            tensor<fp32, [1, 1, 256, 57, 239]> real_div_39 = real_div(x = sub_78, y = sqrt_39)[name = tensor<string, []>("real_div_39")];
            tensor<int32, [4]> reshape_157_shape_0 = const()[name = tensor<string, []>("reshape_157_shape_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<fp32, [1, 256, 57, 239]> reshape_157 = reshape(shape = reshape_157_shape_0, x = real_div_39)[name = tensor<string, []>("reshape_157")];
            tensor<fp32, [256]> add_85_gamma_0 = const()[name = tensor<string, []>("add_85_gamma_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(19871552)))];
            tensor<fp32, [256]> add_85_beta_0 = const()[name = tensor<string, []>("add_85_beta_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(19872640)))];
            tensor<fp32, []> add_85_epsilon_0 = const()[name = tensor<string, []>("add_85_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 256, 57, 239]> add_85 = batch_norm(beta = add_85_beta_0, epsilon = add_85_epsilon_0, gamma = add_85_gamma_0, mean = add_81_mean_0, variance = add_81_variance_0, x = reshape_157)[name = tensor<string, []>("add_85")];
            tensor<int32, [4]> var_1384_perm_0 = const()[name = tensor<string, []>("op_1384_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_1388 = const()[name = tensor<string, []>("op_1388"), val = tensor<int32, [3]>([57, 256, 239])];
            tensor<fp32, [1, 57, 256, 239]> var_1384 = transpose(perm = var_1384_perm_0, x = add_85)[name = tensor<string, []>("transpose_94")];
            tensor<fp32, [57, 256, 239]> var_1389 = reshape(shape = var_1388, x = var_1384)[name = tensor<string, []>("op_1389")];
            tensor<int32, [3]> transpose_32_perm_0 = const()[name = tensor<string, []>("transpose_32_perm_0"), val = tensor<int32, [3]>([2, 0, 1])];
            tensor<fp32, [1024]> add_86 = const()[name = tensor<string, []>("add_86"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(19873728)))];
            tensor<fp32, [1024]> add_87 = const()[name = tensor<string, []>("add_87"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(19877888)))];
            tensor<fp32, [1024, 256]> concat_34 = const()[name = tensor<string, []>("concat_34"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(19882048)))];
            tensor<fp32, [1024, 256]> concat_35 = const()[name = tensor<string, []>("concat_35"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(20930688)))];
            tensor<fp32, [1024, 256]> concat_36 = const()[name = tensor<string, []>("concat_36"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(21979328)))];
            tensor<fp32, [1024, 256]> concat_37 = const()[name = tensor<string, []>("concat_37"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(23027968)))];
            tensor<fp32, [57, 512]> input_259_batch_first_lstm_h0_reshaped = const()[name = tensor<string, []>("input_259_batch_first_lstm_h0_reshaped"), val = tensor<fp32, [57, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24076608)))];
            tensor<string, []> input_259_batch_first_direction_0 = const()[name = tensor<string, []>("input_259_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_259_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_259_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_259_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_259_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_259_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_259_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_259_batch_first_activation_0 = const()[name = tensor<string, []>("input_259_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [239, 57, 256]> transpose_32 = transpose(perm = transpose_32_perm_0, x = var_1389)[name = tensor<string, []>("transpose_93")];
            tensor<fp32, [239, 57, 512]> input_259_batch_first_0, tensor<fp32, [57, 512]> input_259_batch_first_1, tensor<fp32, [57, 512]> input_259_batch_first_2 = lstm(activation = input_259_batch_first_activation_0, bias = add_86, bias_back = add_87, cell_activation = input_259_batch_first_cell_activation_0, direction = input_259_batch_first_direction_0, initial_c = input_259_batch_first_lstm_h0_reshaped, initial_h = input_259_batch_first_lstm_h0_reshaped, output_sequence = input_259_batch_first_output_sequence_0, recurrent_activation = input_259_batch_first_recurrent_activation_0, weight_hh = concat_35, weight_hh_back = concat_37, weight_ih = concat_34, weight_ih_back = concat_36, x = transpose_32)[name = tensor<string, []>("input_259_batch_first")];
            tensor<int32, [3]> input_259_perm_0 = const()[name = tensor<string, []>("input_259_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [57, 239, 512]> input_259 = transpose(perm = input_259_perm_0, x = input_259_batch_first_0)[name = tensor<string, []>("transpose_92")];
            tensor<fp32, [57, 239, 256]> x_73 = linear(bias = separation_net_dp_modules_1_linear_layers_1_bias, weight = separation_net_dp_modules_1_linear_layers_1_weight, x = input_259)[name = tensor<string, []>("linear_3")];
            tensor<int32, [3]> var_1412_perm_0 = const()[name = tensor<string, []>("op_1412_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
            tensor<int32, [4]> var_1414 = const()[name = tensor<string, []>("op_1414"), val = tensor<int32, [4]>([1, 57, 256, 239])];
            tensor<fp32, [57, 256, 239]> var_1412 = transpose(perm = var_1412_perm_0, x = x_73)[name = tensor<string, []>("transpose_91")];
            tensor<fp32, [1, 57, 256, 239]> var_1415 = reshape(shape = var_1414, x = var_1412)[name = tensor<string, []>("op_1415")];
            tensor<int32, [4]> x_75_perm_0 = const()[name = tensor<string, []>("x_75_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 256, 57, 239]> x_75 = transpose(perm = x_75_perm_0, x = var_1415)[name = tensor<string, []>("transpose_90")];
            tensor<fp32, [1, 256, 57, 239]> x_77 = add(x = x_75, y = input_255)[name = tensor<string, []>("x_77")];
            tensor<int32, [4]> var_1420_begin_0 = const()[name = tensor<string, []>("op_1420_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_1420_end_0 = const()[name = tensor<string, []>("op_1420_end_0"), val = tensor<int32, [4]>([1, 128, 57, 239])];
            tensor<bool, [4]> var_1420_end_mask_0 = const()[name = tensor<string, []>("op_1420_end_mask_0"), val = tensor<bool, [4]>([true, false, true, true])];
            tensor<fp32, [1, 128, 57, 239]> var_1420 = slice_by_index(begin = var_1420_begin_0, end = var_1420_end_0, end_mask = var_1420_end_mask_0, x = x_77)[name = tensor<string, []>("op_1420")];
            tensor<int32, [4]> var_1424_begin_0 = const()[name = tensor<string, []>("op_1424_begin_0"), val = tensor<int32, [4]>([0, 128, 0, 0])];
            tensor<int32, [4]> var_1424_end_0 = const()[name = tensor<string, []>("op_1424_end_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<bool, [4]> var_1424_end_mask_0 = const()[name = tensor<string, []>("op_1424_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 128, 57, 239]> var_1424 = slice_by_index(begin = var_1424_begin_0, end = var_1424_end_0, end_mask = var_1424_end_mask_0, x = x_77)[name = tensor<string, []>("op_1424")];
            tensor<int32, []> gather_2_batch_dims_0 = const()[name = tensor<string, []>("gather_2_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_2_validate_indices_0 = const()[name = tensor<string, []>("gather_2_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, [237]> select_2 = const()[name = tensor<string, []>("select_2"), val = tensor<int32, [237]>([237, 236, 235, 234, 233, 232, 231, 230, 229, 228, 227, 226, 225, 224, 223, 222, 221, 220, 219, 218, 217, 216, 215, 214, 213, 212, 211, 210, 209, 208, 207, 206, 205, 204, 203, 202, 201, 200, 199, 198, 197, 196, 195, 194, 193, 192, 191, 190, 189, 188, 187, 186, 185, 184, 183, 182, 181, 180, 179, 178, 177, 176, 175, 174, 173, 172, 171, 170, 169, 168, 167, 166, 165, 164, 163, 162, 161, 160, 159, 158, 157, 156, 155, 154, 153, 152, 151, 150, 149, 148, 147, 146, 145, 144, 143, 142, 141, 140, 139, 138, 137, 136, 135, 134, 133, 132, 131, 130, 129, 128, 127, 126, 125, 124, 123, 122, 121, 120, 119, 118, 117, 116, 115, 114, 113, 112, 111, 110, 109, 108, 107, 106, 105, 104, 103, 102, 101, 100, 99, 98, 97, 96, 95, 94, 93, 92, 91, 90, 89, 88, 87, 86, 85, 84, 83, 82, 81, 80, 79, 78, 77, 76, 75, 74, 73, 72, 71, 70, 69, 68, 67, 66, 65, 64, 63, 62, 61, 60, 59, 58, 57, 56, 55, 54, 53, 52, 51, 50, 49, 48, 47, 46, 45, 44, 43, 42, 41, 40, 39, 38, 37, 36, 35, 34, 33, 32, 31, 30, 29, 28, 27, 26, 25, 24, 23, 22, 21, 20, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1])];
            tensor<int32, []> gather_2_axis_1 = const()[name = tensor<string, []>("gather_2_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 237]> gather_2 = gather(axis = gather_2_axis_1, batch_dims = gather_2_batch_dims_0, indices = select_2, validate_indices = gather_2_validate_indices_0, x = var_1420)[name = tensor<string, []>("gather_2")];
            tensor<int32, []> gather_3_batch_dims_0 = const()[name = tensor<string, []>("gather_3_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_3_validate_indices_0 = const()[name = tensor<string, []>("gather_3_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_3_axis_1 = const()[name = tensor<string, []>("gather_3_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 237]> gather_3 = gather(axis = gather_3_axis_1, batch_dims = gather_3_batch_dims_0, indices = select_2, validate_indices = gather_3_validate_indices_0, x = var_1424)[name = tensor<string, []>("gather_3")];
            tensor<fp32, []> mul_50_y_0 = const()[name = tensor<string, []>("mul_50_y_0"), val = tensor<fp32, []>(-0x1p+0)];
            tensor<fp32, [1, 128, 57, 237]> mul_50 = mul(x = gather_3, y = mul_50_y_0)[name = tensor<string, []>("mul_50")];
            tensor<int32, []> concat_120_axis_0 = const()[name = tensor<string, []>("concat_120_axis_0"), val = tensor<int32, []>(3)];
            tensor<bool, []> concat_120_interleave_0 = const()[name = tensor<string, []>("concat_120_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 128, 57, 476]> concat_120 = concat(axis = concat_120_axis_0, interleave = concat_120_interleave_0, values = (var_1420, gather_2))[name = tensor<string, []>("concat_120")];
            tensor<int32, []> concat_121_axis_0 = const()[name = tensor<string, []>("concat_121_axis_0"), val = tensor<int32, []>(3)];
            tensor<bool, []> concat_121_interleave_0 = const()[name = tensor<string, []>("concat_121_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 128, 57, 476]> concat_121 = concat(axis = concat_121_axis_0, interleave = concat_121_interleave_0, values = (var_1424, mul_50))[name = tensor<string, []>("concat_121")];
            tensor<int32, [4]> transpose_4_perm_0 = const()[name = tensor<string, []>("transpose_4_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [4]> transpose_5_perm_0 = const()[name = tensor<string, []>("transpose_5_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [2]> reshape_198_shape_0 = const()[name = tensor<string, []>("reshape_198_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_4 = transpose(perm = transpose_4_perm_0, x = concat_120)[name = tensor<string, []>("transpose_89")];
            tensor<fp32, [476, 7296]> reshape_198 = reshape(shape = reshape_198_shape_0, x = transpose_4)[name = tensor<string, []>("reshape_198")];
            tensor<int32, [2]> reshape_199_shape_0 = const()[name = tensor<string, []>("reshape_199_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_5 = transpose(perm = transpose_5_perm_0, x = concat_121)[name = tensor<string, []>("transpose_88")];
            tensor<fp32, [476, 7296]> reshape_199 = reshape(shape = reshape_199_shape_0, x = transpose_5)[name = tensor<string, []>("reshape_199")];
            tensor<bool, []> matmul_6_transpose_x_0 = const()[name = tensor<string, []>("matmul_6_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_6_transpose_y_0 = const()[name = tensor<string, []>("matmul_6_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_6 = matmul(transpose_x = matmul_6_transpose_x_0, transpose_y = matmul_6_transpose_y_0, x = cos_0, y = reshape_198)[name = tensor<string, []>("matmul_6")];
            tensor<bool, []> matmul_7_transpose_x_0 = const()[name = tensor<string, []>("matmul_7_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_7_transpose_y_0 = const()[name = tensor<string, []>("matmul_7_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_7 = matmul(transpose_x = matmul_7_transpose_x_0, transpose_y = matmul_7_transpose_y_0, x = sin_0, y = reshape_199)[name = tensor<string, []>("matmul_7")];
            tensor<fp32, [476, 7296]> sub_97 = sub(x = matmul_6, y = matmul_7)[name = tensor<string, []>("sub_97")];
            tensor<int32, [4]> reshape_202_shape_0 = const()[name = tensor<string, []>("reshape_202_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_202 = reshape(shape = reshape_202_shape_0, x = sub_97)[name = tensor<string, []>("reshape_202")];
            tensor<int32, [4]> transpose_6_perm_0 = const()[name = tensor<string, []>("transpose_6_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<fp32, []> _inversed_real_div_52_y_0 = const()[name = tensor<string, []>("_inversed_real_div_52_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [1, 128, 57, 476]> transpose_6 = transpose(perm = transpose_6_perm_0, x = reshape_202)[name = tensor<string, []>("transpose_87")];
            tensor<fp32, [1, 128, 57, 476]> _inversed_real_div_52 = mul(x = transpose_6, y = _inversed_real_div_52_y_0)[name = tensor<string, []>("_inversed_real_div_52")];
            tensor<int32, [5]> reshape_160_shape_0 = const()[name = tensor<string, []>("reshape_160_shape_0"), val = tensor<int32, [5]>([1, 1, 128, 57, 476])];
            tensor<fp32, [1, 1, 128, 57, 476]> reshape_160 = reshape(shape = reshape_160_shape_0, x = _inversed_real_div_52)[name = tensor<string, []>("reshape_160")];
            tensor<int32, [3]> reduce_mean_120_axes_0 = const()[name = tensor<string, []>("reduce_mean_120_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_120_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_120_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_120 = reduce_mean(axes = reduce_mean_120_axes_0, keep_dims = reduce_mean_120_keep_dims_0, x = reshape_160)[name = tensor<string, []>("reduce_mean_120")];
            tensor<fp32, [1, 1, 128, 57, 476]> sub_80 = sub(x = reshape_160, y = reduce_mean_120)[name = tensor<string, []>("sub_80")];
            tensor<fp32, [1, 1, 128, 57, 476]> square_40 = square(x = sub_80)[name = tensor<string, []>("square_40")];
            tensor<int32, [3]> reduce_mean_122_axes_0 = const()[name = tensor<string, []>("reduce_mean_122_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_122_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_122_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_122 = reduce_mean(axes = reduce_mean_122_axes_0, keep_dims = reduce_mean_122_keep_dims_0, x = square_40)[name = tensor<string, []>("reduce_mean_122")];
            tensor<fp32, []> add_88_y_0 = const()[name = tensor<string, []>("add_88_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_88 = add(x = reduce_mean_122, y = add_88_y_0)[name = tensor<string, []>("add_88")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_40 = sqrt(x = add_88)[name = tensor<string, []>("sqrt_40")];
            tensor<fp32, [1, 1, 128, 57, 476]> real_div_40 = real_div(x = sub_80, y = sqrt_40)[name = tensor<string, []>("real_div_40")];
            tensor<int32, [4]> reshape_161_shape_0 = const()[name = tensor<string, []>("reshape_161_shape_0"), val = tensor<int32, [4]>([1, 128, 57, 476])];
            tensor<fp32, [1, 128, 57, 476]> reshape_161 = reshape(shape = reshape_161_shape_0, x = real_div_40)[name = tensor<string, []>("reshape_161")];
            tensor<fp32, [128]> add_89_gamma_0 = const()[name = tensor<string, []>("add_89_gamma_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24193408)))];
            tensor<fp32, [128]> add_89_beta_0 = const()[name = tensor<string, []>("add_89_beta_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24193984)))];
            tensor<fp32, []> add_89_epsilon_0 = const()[name = tensor<string, []>("add_89_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 128, 57, 476]> add_89 = batch_norm(beta = add_89_beta_0, epsilon = add_89_epsilon_0, gamma = add_89_gamma_0, mean = add_73_mean_0, variance = add_73_variance_0, x = reshape_161)[name = tensor<string, []>("add_89")];
            tensor<int32, [4]> var_1451_perm_0 = const()[name = tensor<string, []>("op_1451_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<int32, [3]> var_1455 = const()[name = tensor<string, []>("op_1455"), val = tensor<int32, [3]>([476, 57, 128])];
            tensor<fp32, [1, 476, 57, 128]> var_1451 = transpose(perm = var_1451_perm_0, x = add_89)[name = tensor<string, []>("transpose_86")];
            tensor<fp32, [476, 57, 128]> input_261 = reshape(shape = var_1455, x = var_1451)[name = tensor<string, []>("input_261")];
            tensor<int32, [3]> input_261_batch_first_transpose_perm_0 = const()[name = tensor<string, []>("input_261_batch_first_transpose_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [512]> add_90 = const()[name = tensor<string, []>("add_90"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24194560)))];
            tensor<fp32, [512]> add_91 = const()[name = tensor<string, []>("add_91"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24196672)))];
            tensor<fp32, [512, 128]> concat_44 = const()[name = tensor<string, []>("concat_44"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24198784)))];
            tensor<fp32, [512, 128]> concat_45 = const()[name = tensor<string, []>("concat_45"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24460992)))];
            tensor<fp32, [512, 128]> concat_46 = const()[name = tensor<string, []>("concat_46"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24723200)))];
            tensor<fp32, [512, 128]> concat_47 = const()[name = tensor<string, []>("concat_47"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(24985408)))];
            tensor<string, []> input_263_batch_first_direction_0 = const()[name = tensor<string, []>("input_263_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_263_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_263_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_263_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_263_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_263_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_263_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_263_batch_first_activation_0 = const()[name = tensor<string, []>("input_263_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [57, 476, 128]> input_261_batch_first_transpose = transpose(perm = input_261_batch_first_transpose_perm_0, x = input_261)[name = tensor<string, []>("transpose_85")];
            tensor<fp32, [57, 476, 256]> input_263_batch_first_0, tensor<fp32, [476, 256]> input_263_batch_first_1, tensor<fp32, [476, 256]> input_263_batch_first_2 = lstm(activation = input_263_batch_first_activation_0, bias = add_90, bias_back = add_91, cell_activation = input_263_batch_first_cell_activation_0, direction = input_263_batch_first_direction_0, initial_c = input_243_batch_first_lstm_h0_reshaped, initial_h = input_243_batch_first_lstm_h0_reshaped, output_sequence = input_263_batch_first_output_sequence_0, recurrent_activation = input_263_batch_first_recurrent_activation_0, weight_hh = concat_45, weight_hh_back = concat_47, weight_ih = concat_44, weight_ih_back = concat_46, x = input_261_batch_first_transpose)[name = tensor<string, []>("input_263_batch_first")];
            tensor<int32, [3]> input_263_perm_0 = const()[name = tensor<string, []>("input_263_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [476, 57, 256]> input_263 = transpose(perm = input_263_perm_0, x = input_263_batch_first_0)[name = tensor<string, []>("transpose_84")];
            tensor<fp32, [476, 57, 128]> x_87 = linear(bias = separation_net_dp_modules_2_linear_layers_0_bias, weight = separation_net_dp_modules_2_linear_layers_0_weight, x = input_263)[name = tensor<string, []>("linear_4")];
            tensor<int32, [4]> var_1478 = const()[name = tensor<string, []>("op_1478"), val = tensor<int32, [4]>([1, 476, 57, 128])];
            tensor<fp32, [1, 476, 57, 128]> var_1479 = reshape(shape = var_1478, x = x_87)[name = tensor<string, []>("op_1479")];
            tensor<int32, [4]> x_89_perm_0 = const()[name = tensor<string, []>("x_89_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<fp32, [1, 128, 57, 476]> x_89 = transpose(perm = x_89_perm_0, x = var_1479)[name = tensor<string, []>("transpose_83")];
            tensor<fp32, [1, 128, 57, 476]> input_265 = add(x = x_89, y = _inversed_real_div_52)[name = tensor<string, []>("input_265")];
            tensor<int32, [5]> reshape_164_shape_0 = const()[name = tensor<string, []>("reshape_164_shape_0"), val = tensor<int32, [5]>([1, 1, 128, 57, 476])];
            tensor<fp32, [1, 1, 128, 57, 476]> reshape_164 = reshape(shape = reshape_164_shape_0, x = input_265)[name = tensor<string, []>("reshape_164")];
            tensor<int32, [3]> reduce_mean_123_axes_0 = const()[name = tensor<string, []>("reduce_mean_123_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_123_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_123_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_123 = reduce_mean(axes = reduce_mean_123_axes_0, keep_dims = reduce_mean_123_keep_dims_0, x = reshape_164)[name = tensor<string, []>("reduce_mean_123")];
            tensor<fp32, [1, 1, 128, 57, 476]> sub_82 = sub(x = reshape_164, y = reduce_mean_123)[name = tensor<string, []>("sub_82")];
            tensor<fp32, [1, 1, 128, 57, 476]> square_41 = square(x = sub_82)[name = tensor<string, []>("square_41")];
            tensor<int32, [3]> reduce_mean_125_axes_0 = const()[name = tensor<string, []>("reduce_mean_125_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_125_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_125_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_125 = reduce_mean(axes = reduce_mean_125_axes_0, keep_dims = reduce_mean_125_keep_dims_0, x = square_41)[name = tensor<string, []>("reduce_mean_125")];
            tensor<fp32, []> add_92_y_0 = const()[name = tensor<string, []>("add_92_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_92 = add(x = reduce_mean_125, y = add_92_y_0)[name = tensor<string, []>("add_92")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_41 = sqrt(x = add_92)[name = tensor<string, []>("sqrt_41")];
            tensor<fp32, [1, 1, 128, 57, 476]> real_div_41 = real_div(x = sub_82, y = sqrt_41)[name = tensor<string, []>("real_div_41")];
            tensor<int32, [4]> reshape_165_shape_0 = const()[name = tensor<string, []>("reshape_165_shape_0"), val = tensor<int32, [4]>([1, 128, 57, 476])];
            tensor<fp32, [1, 128, 57, 476]> reshape_165 = reshape(shape = reshape_165_shape_0, x = real_div_41)[name = tensor<string, []>("reshape_165")];
            tensor<fp32, [128]> add_93_gamma_0 = const()[name = tensor<string, []>("add_93_gamma_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(25247616)))];
            tensor<fp32, [128]> add_93_beta_0 = const()[name = tensor<string, []>("add_93_beta_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(25248192)))];
            tensor<fp32, []> add_93_epsilon_0 = const()[name = tensor<string, []>("add_93_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 128, 57, 476]> add_93 = batch_norm(beta = add_93_beta_0, epsilon = add_93_epsilon_0, gamma = add_93_gamma_0, mean = add_73_mean_0, variance = add_73_variance_0, x = reshape_165)[name = tensor<string, []>("add_93")];
            tensor<int32, [4]> var_1485_perm_0 = const()[name = tensor<string, []>("op_1485_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_1489 = const()[name = tensor<string, []>("op_1489"), val = tensor<int32, [3]>([57, 128, 476])];
            tensor<fp32, [1, 57, 128, 476]> var_1485 = transpose(perm = var_1485_perm_0, x = add_93)[name = tensor<string, []>("transpose_82")];
            tensor<fp32, [57, 128, 476]> var_1490 = reshape(shape = var_1489, x = var_1485)[name = tensor<string, []>("op_1490")];
            tensor<int32, [3]> transpose_33_perm_0 = const()[name = tensor<string, []>("transpose_33_perm_0"), val = tensor<int32, [3]>([2, 0, 1])];
            tensor<fp32, [512]> add_94 = const()[name = tensor<string, []>("add_94"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(25248768)))];
            tensor<fp32, [512]> add_95 = const()[name = tensor<string, []>("add_95"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(25250880)))];
            tensor<fp32, [512, 128]> concat_54 = const()[name = tensor<string, []>("concat_54"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(25252992)))];
            tensor<fp32, [512, 128]> concat_55 = const()[name = tensor<string, []>("concat_55"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(25515200)))];
            tensor<fp32, [512, 128]> concat_56 = const()[name = tensor<string, []>("concat_56"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(25777408)))];
            tensor<fp32, [512, 128]> concat_57 = const()[name = tensor<string, []>("concat_57"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26039616)))];
            tensor<string, []> input_269_batch_first_direction_0 = const()[name = tensor<string, []>("input_269_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_269_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_269_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_269_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_269_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_269_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_269_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_269_batch_first_activation_0 = const()[name = tensor<string, []>("input_269_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [476, 57, 128]> transpose_33 = transpose(perm = transpose_33_perm_0, x = var_1490)[name = tensor<string, []>("transpose_81")];
            tensor<fp32, [476, 57, 256]> input_269_batch_first_0, tensor<fp32, [57, 256]> input_269_batch_first_1, tensor<fp32, [57, 256]> input_269_batch_first_2 = lstm(activation = input_269_batch_first_activation_0, bias = add_94, bias_back = add_95, cell_activation = input_269_batch_first_cell_activation_0, direction = input_269_batch_first_direction_0, initial_c = input_249_batch_first_lstm_h0_reshaped, initial_h = input_249_batch_first_lstm_h0_reshaped, output_sequence = input_269_batch_first_output_sequence_0, recurrent_activation = input_269_batch_first_recurrent_activation_0, weight_hh = concat_55, weight_hh_back = concat_57, weight_ih = concat_54, weight_ih_back = concat_56, x = transpose_33)[name = tensor<string, []>("input_269_batch_first")];
            tensor<int32, [3]> input_269_perm_0 = const()[name = tensor<string, []>("input_269_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [57, 476, 256]> input_269 = transpose(perm = input_269_perm_0, x = input_269_batch_first_0)[name = tensor<string, []>("transpose_80")];
            tensor<fp32, [57, 476, 128]> x_93 = linear(bias = separation_net_dp_modules_2_linear_layers_1_bias, weight = separation_net_dp_modules_2_linear_layers_1_weight, x = input_269)[name = tensor<string, []>("linear_5")];
            tensor<int32, [3]> var_1513_perm_0 = const()[name = tensor<string, []>("op_1513_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
            tensor<int32, [4]> var_1515 = const()[name = tensor<string, []>("op_1515"), val = tensor<int32, [4]>([1, 57, 128, 476])];
            tensor<fp32, [57, 128, 476]> var_1513 = transpose(perm = var_1513_perm_0, x = x_93)[name = tensor<string, []>("transpose_79")];
            tensor<fp32, [1, 57, 128, 476]> var_1516 = reshape(shape = var_1515, x = var_1513)[name = tensor<string, []>("op_1516")];
            tensor<int32, [4]> x_95_perm_0 = const()[name = tensor<string, []>("x_95_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 128, 57, 476]> x_95 = transpose(perm = x_95_perm_0, x = var_1516)[name = tensor<string, []>("transpose_78")];
            tensor<fp32, [1, 128, 57, 476]> x_97 = add(x = x_95, y = input_265)[name = tensor<string, []>("x_97")];
            tensor<int32, [4]> transpose_8_perm_0 = const()[name = tensor<string, []>("transpose_8_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [2]> reshape_204_shape_0 = const()[name = tensor<string, []>("reshape_204_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_8 = transpose(perm = transpose_8_perm_0, x = x_97)[name = tensor<string, []>("transpose_77")];
            tensor<fp32, [476, 7296]> reshape_204 = reshape(shape = reshape_204_shape_0, x = transpose_8)[name = tensor<string, []>("reshape_204")];
            tensor<bool, []> matmul_11_transpose_x_0 = const()[name = tensor<string, []>("matmul_11_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_11_transpose_y_0 = const()[name = tensor<string, []>("matmul_11_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_11 = matmul(transpose_x = matmul_11_transpose_x_0, transpose_y = matmul_11_transpose_y_0, x = cos_0, y = reshape_204)[name = tensor<string, []>("matmul_11")];
            tensor<bool, []> matmul_13_transpose_x_0 = const()[name = tensor<string, []>("matmul_13_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_13_transpose_y_0 = const()[name = tensor<string, []>("matmul_13_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_13 = matmul(transpose_x = matmul_13_transpose_x_0, transpose_y = matmul_13_transpose_y_0, x = sin_0, y = reshape_204)[name = tensor<string, []>("matmul_13")];
            tensor<fp32, []> mul_53_y_0 = const()[name = tensor<string, []>("mul_53_y_0"), val = tensor<fp32, []>(-0x1p+0)];
            tensor<fp32, [476, 7296]> mul_53 = mul(x = matmul_13, y = mul_53_y_0)[name = tensor<string, []>("mul_53")];
            tensor<int32, [4]> reshape_208_shape_0 = const()[name = tensor<string, []>("reshape_208_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_208 = reshape(shape = reshape_208_shape_0, x = matmul_11)[name = tensor<string, []>("reshape_208")];
            tensor<int32, [4]> reshape_209_shape_0 = const()[name = tensor<string, []>("reshape_209_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_209 = reshape(shape = reshape_209_shape_0, x = mul_53)[name = tensor<string, []>("reshape_209")];
            tensor<int32, [4]> transpose_10_perm_0 = const()[name = tensor<string, []>("transpose_10_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [4]> transpose_11_perm_0 = const()[name = tensor<string, []>("transpose_11_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<fp32, []> _inversed_real_div_55_y_0 = const()[name = tensor<string, []>("_inversed_real_div_55_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [1, 128, 57, 476]> transpose_10 = transpose(perm = transpose_10_perm_0, x = reshape_208)[name = tensor<string, []>("transpose_76")];
            tensor<fp32, [1, 128, 57, 476]> _inversed_real_div_55 = mul(x = transpose_10, y = _inversed_real_div_55_y_0)[name = tensor<string, []>("_inversed_real_div_55")];
            tensor<fp32, []> _inversed_real_div_56_y_0 = const()[name = tensor<string, []>("_inversed_real_div_56_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [1, 128, 57, 476]> transpose_11 = transpose(perm = transpose_11_perm_0, x = reshape_209)[name = tensor<string, []>("transpose_75")];
            tensor<fp32, [1, 128, 57, 476]> _inversed_real_div_56 = mul(x = transpose_11, y = _inversed_real_div_56_y_0)[name = tensor<string, []>("_inversed_real_div_56")];
            tensor<int32, []> gather_4_batch_dims_0 = const()[name = tensor<string, []>("gather_4_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_4_validate_indices_0 = const()[name = tensor<string, []>("gather_4_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_4_axis_1 = const()[name = tensor<string, []>("gather_4_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 239]> gather_4 = gather(axis = gather_4_axis_1, batch_dims = gather_4_batch_dims_0, indices = select_0, validate_indices = gather_4_validate_indices_0, x = _inversed_real_div_55)[name = tensor<string, []>("gather_4")];
            tensor<int32, []> gather_5_batch_dims_0 = const()[name = tensor<string, []>("gather_5_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_5_validate_indices_0 = const()[name = tensor<string, []>("gather_5_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_5_axis_1 = const()[name = tensor<string, []>("gather_5_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 239]> gather_5 = gather(axis = gather_5_axis_1, batch_dims = gather_5_batch_dims_0, indices = select_0, validate_indices = gather_5_validate_indices_0, x = _inversed_real_div_56)[name = tensor<string, []>("gather_5")];
            tensor<bool, []> x_103_interleave_0 = const()[name = tensor<string, []>("x_103_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 256, 57, 239]> x_103 = concat(axis = var_1209, interleave = x_103_interleave_0, values = (gather_4, gather_5))[name = tensor<string, []>("x_103")];
            tensor<int32, [5]> reshape_168_shape_0 = const()[name = tensor<string, []>("reshape_168_shape_0"), val = tensor<int32, [5]>([1, 1, 256, 57, 239])];
            tensor<fp32, [1, 1, 256, 57, 239]> reshape_168 = reshape(shape = reshape_168_shape_0, x = x_103)[name = tensor<string, []>("reshape_168")];
            tensor<int32, [3]> reduce_mean_126_axes_0 = const()[name = tensor<string, []>("reduce_mean_126_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_126_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_126_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_126 = reduce_mean(axes = reduce_mean_126_axes_0, keep_dims = reduce_mean_126_keep_dims_0, x = reshape_168)[name = tensor<string, []>("reduce_mean_126")];
            tensor<fp32, [1, 1, 256, 57, 239]> sub_84 = sub(x = reshape_168, y = reduce_mean_126)[name = tensor<string, []>("sub_84")];
            tensor<fp32, [1, 1, 256, 57, 239]> square_42 = square(x = sub_84)[name = tensor<string, []>("square_42")];
            tensor<int32, [3]> reduce_mean_128_axes_0 = const()[name = tensor<string, []>("reduce_mean_128_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_128_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_128_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_128 = reduce_mean(axes = reduce_mean_128_axes_0, keep_dims = reduce_mean_128_keep_dims_0, x = square_42)[name = tensor<string, []>("reduce_mean_128")];
            tensor<fp32, []> add_96_y_0 = const()[name = tensor<string, []>("add_96_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_96 = add(x = reduce_mean_128, y = add_96_y_0)[name = tensor<string, []>("add_96")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_42 = sqrt(x = add_96)[name = tensor<string, []>("sqrt_42")];
            tensor<fp32, [1, 1, 256, 57, 239]> real_div_42 = real_div(x = sub_84, y = sqrt_42)[name = tensor<string, []>("real_div_42")];
            tensor<int32, [4]> reshape_169_shape_0 = const()[name = tensor<string, []>("reshape_169_shape_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<fp32, [1, 256, 57, 239]> reshape_169 = reshape(shape = reshape_169_shape_0, x = real_div_42)[name = tensor<string, []>("reshape_169")];
            tensor<fp32, [256]> add_97_gamma_0 = const()[name = tensor<string, []>("add_97_gamma_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26301824)))];
            tensor<fp32, [256]> add_97_beta_0 = const()[name = tensor<string, []>("add_97_beta_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26302912)))];
            tensor<fp32, []> add_97_epsilon_0 = const()[name = tensor<string, []>("add_97_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 256, 57, 239]> add_97 = batch_norm(beta = add_97_beta_0, epsilon = add_97_epsilon_0, gamma = add_97_gamma_0, mean = add_81_mean_0, variance = add_81_variance_0, x = reshape_169)[name = tensor<string, []>("add_97")];
            tensor<int32, [4]> var_1547_perm_0 = const()[name = tensor<string, []>("op_1547_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<int32, [3]> var_1551 = const()[name = tensor<string, []>("op_1551"), val = tensor<int32, [3]>([239, 57, 256])];
            tensor<fp32, [1, 239, 57, 256]> var_1547 = transpose(perm = var_1547_perm_0, x = add_97)[name = tensor<string, []>("transpose_74")];
            tensor<fp32, [239, 57, 256]> input_271 = reshape(shape = var_1551, x = var_1547)[name = tensor<string, []>("input_271")];
            tensor<int32, [3]> input_271_batch_first_transpose_perm_0 = const()[name = tensor<string, []>("input_271_batch_first_transpose_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [1024]> add_98 = const()[name = tensor<string, []>("add_98"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26304000)))];
            tensor<fp32, [1024]> add_99 = const()[name = tensor<string, []>("add_99"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26308160)))];
            tensor<fp32, [1024, 256]> concat_64 = const()[name = tensor<string, []>("concat_64"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(26312320)))];
            tensor<fp32, [1024, 256]> concat_65 = const()[name = tensor<string, []>("concat_65"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(27360960)))];
            tensor<fp32, [1024, 256]> concat_66 = const()[name = tensor<string, []>("concat_66"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(28409600)))];
            tensor<fp32, [1024, 256]> concat_67 = const()[name = tensor<string, []>("concat_67"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(29458240)))];
            tensor<string, []> input_273_batch_first_direction_0 = const()[name = tensor<string, []>("input_273_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_273_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_273_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_273_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_273_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_273_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_273_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_273_batch_first_activation_0 = const()[name = tensor<string, []>("input_273_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [57, 239, 256]> input_271_batch_first_transpose = transpose(perm = input_271_batch_first_transpose_perm_0, x = input_271)[name = tensor<string, []>("transpose_73")];
            tensor<fp32, [57, 239, 512]> input_273_batch_first_0, tensor<fp32, [239, 512]> input_273_batch_first_1, tensor<fp32, [239, 512]> input_273_batch_first_2 = lstm(activation = input_273_batch_first_activation_0, bias = add_98, bias_back = add_99, cell_activation = input_273_batch_first_cell_activation_0, direction = input_273_batch_first_direction_0, initial_c = input_253_batch_first_lstm_h0_reshaped, initial_h = input_253_batch_first_lstm_h0_reshaped, output_sequence = input_273_batch_first_output_sequence_0, recurrent_activation = input_273_batch_first_recurrent_activation_0, weight_hh = concat_65, weight_hh_back = concat_67, weight_ih = concat_64, weight_ih_back = concat_66, x = input_271_batch_first_transpose)[name = tensor<string, []>("input_273_batch_first")];
            tensor<int32, [3]> input_273_perm_0 = const()[name = tensor<string, []>("input_273_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [239, 57, 512]> input_273 = transpose(perm = input_273_perm_0, x = input_273_batch_first_0)[name = tensor<string, []>("transpose_72")];
            tensor<fp32, [239, 57, 256]> x_107 = linear(bias = separation_net_dp_modules_3_linear_layers_0_bias, weight = separation_net_dp_modules_3_linear_layers_0_weight, x = input_273)[name = tensor<string, []>("linear_6")];
            tensor<int32, [4]> var_1574 = const()[name = tensor<string, []>("op_1574"), val = tensor<int32, [4]>([1, 239, 57, 256])];
            tensor<fp32, [1, 239, 57, 256]> var_1575 = reshape(shape = var_1574, x = x_107)[name = tensor<string, []>("op_1575")];
            tensor<int32, [4]> x_109_perm_0 = const()[name = tensor<string, []>("x_109_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<fp32, [1, 256, 57, 239]> x_109 = transpose(perm = x_109_perm_0, x = var_1575)[name = tensor<string, []>("transpose_71")];
            tensor<fp32, [1, 256, 57, 239]> input_275 = add(x = x_109, y = x_103)[name = tensor<string, []>("input_275")];
            tensor<int32, [5]> reshape_172_shape_0 = const()[name = tensor<string, []>("reshape_172_shape_0"), val = tensor<int32, [5]>([1, 1, 256, 57, 239])];
            tensor<fp32, [1, 1, 256, 57, 239]> reshape_172 = reshape(shape = reshape_172_shape_0, x = input_275)[name = tensor<string, []>("reshape_172")];
            tensor<int32, [3]> reduce_mean_129_axes_0 = const()[name = tensor<string, []>("reduce_mean_129_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_129_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_129_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_129 = reduce_mean(axes = reduce_mean_129_axes_0, keep_dims = reduce_mean_129_keep_dims_0, x = reshape_172)[name = tensor<string, []>("reduce_mean_129")];
            tensor<fp32, [1, 1, 256, 57, 239]> sub_86 = sub(x = reshape_172, y = reduce_mean_129)[name = tensor<string, []>("sub_86")];
            tensor<fp32, [1, 1, 256, 57, 239]> square_43 = square(x = sub_86)[name = tensor<string, []>("square_43")];
            tensor<int32, [3]> reduce_mean_131_axes_0 = const()[name = tensor<string, []>("reduce_mean_131_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_131_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_131_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_131 = reduce_mean(axes = reduce_mean_131_axes_0, keep_dims = reduce_mean_131_keep_dims_0, x = square_43)[name = tensor<string, []>("reduce_mean_131")];
            tensor<fp32, []> add_100_y_0 = const()[name = tensor<string, []>("add_100_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_100 = add(x = reduce_mean_131, y = add_100_y_0)[name = tensor<string, []>("add_100")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_43 = sqrt(x = add_100)[name = tensor<string, []>("sqrt_43")];
            tensor<fp32, [1, 1, 256, 57, 239]> real_div_43 = real_div(x = sub_86, y = sqrt_43)[name = tensor<string, []>("real_div_43")];
            tensor<int32, [4]> reshape_173_shape_0 = const()[name = tensor<string, []>("reshape_173_shape_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<fp32, [1, 256, 57, 239]> reshape_173 = reshape(shape = reshape_173_shape_0, x = real_div_43)[name = tensor<string, []>("reshape_173")];
            tensor<fp32, [256]> add_101_gamma_0 = const()[name = tensor<string, []>("add_101_gamma_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30506880)))];
            tensor<fp32, [256]> add_101_beta_0 = const()[name = tensor<string, []>("add_101_beta_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30507968)))];
            tensor<fp32, []> add_101_epsilon_0 = const()[name = tensor<string, []>("add_101_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 256, 57, 239]> add_101 = batch_norm(beta = add_101_beta_0, epsilon = add_101_epsilon_0, gamma = add_101_gamma_0, mean = add_81_mean_0, variance = add_81_variance_0, x = reshape_173)[name = tensor<string, []>("add_101")];
            tensor<int32, [4]> var_1581_perm_0 = const()[name = tensor<string, []>("op_1581_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_1585 = const()[name = tensor<string, []>("op_1585"), val = tensor<int32, [3]>([57, 256, 239])];
            tensor<fp32, [1, 57, 256, 239]> var_1581 = transpose(perm = var_1581_perm_0, x = add_101)[name = tensor<string, []>("transpose_70")];
            tensor<fp32, [57, 256, 239]> var_1586 = reshape(shape = var_1585, x = var_1581)[name = tensor<string, []>("op_1586")];
            tensor<int32, [3]> transpose_34_perm_0 = const()[name = tensor<string, []>("transpose_34_perm_0"), val = tensor<int32, [3]>([2, 0, 1])];
            tensor<fp32, [1024]> add_102 = const()[name = tensor<string, []>("add_102"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30509056)))];
            tensor<fp32, [1024]> add_103 = const()[name = tensor<string, []>("add_103"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30513216)))];
            tensor<fp32, [1024, 256]> concat_74 = const()[name = tensor<string, []>("concat_74"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(30517376)))];
            tensor<fp32, [1024, 256]> concat_75 = const()[name = tensor<string, []>("concat_75"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(31566016)))];
            tensor<fp32, [1024, 256]> concat_76 = const()[name = tensor<string, []>("concat_76"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(32614656)))];
            tensor<fp32, [1024, 256]> concat_77 = const()[name = tensor<string, []>("concat_77"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(33663296)))];
            tensor<string, []> input_279_batch_first_direction_0 = const()[name = tensor<string, []>("input_279_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_279_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_279_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_279_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_279_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_279_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_279_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_279_batch_first_activation_0 = const()[name = tensor<string, []>("input_279_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [239, 57, 256]> transpose_34 = transpose(perm = transpose_34_perm_0, x = var_1586)[name = tensor<string, []>("transpose_69")];
            tensor<fp32, [239, 57, 512]> input_279_batch_first_0, tensor<fp32, [57, 512]> input_279_batch_first_1, tensor<fp32, [57, 512]> input_279_batch_first_2 = lstm(activation = input_279_batch_first_activation_0, bias = add_102, bias_back = add_103, cell_activation = input_279_batch_first_cell_activation_0, direction = input_279_batch_first_direction_0, initial_c = input_259_batch_first_lstm_h0_reshaped, initial_h = input_259_batch_first_lstm_h0_reshaped, output_sequence = input_279_batch_first_output_sequence_0, recurrent_activation = input_279_batch_first_recurrent_activation_0, weight_hh = concat_75, weight_hh_back = concat_77, weight_ih = concat_74, weight_ih_back = concat_76, x = transpose_34)[name = tensor<string, []>("input_279_batch_first")];
            tensor<int32, [3]> input_279_perm_0 = const()[name = tensor<string, []>("input_279_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [57, 239, 512]> input_279 = transpose(perm = input_279_perm_0, x = input_279_batch_first_0)[name = tensor<string, []>("transpose_68")];
            tensor<fp32, [57, 239, 256]> x_113 = linear(bias = separation_net_dp_modules_3_linear_layers_1_bias, weight = separation_net_dp_modules_3_linear_layers_1_weight, x = input_279)[name = tensor<string, []>("linear_7")];
            tensor<int32, [3]> var_1609_perm_0 = const()[name = tensor<string, []>("op_1609_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
            tensor<int32, [4]> var_1611 = const()[name = tensor<string, []>("op_1611"), val = tensor<int32, [4]>([1, 57, 256, 239])];
            tensor<fp32, [57, 256, 239]> var_1609 = transpose(perm = var_1609_perm_0, x = x_113)[name = tensor<string, []>("transpose_67")];
            tensor<fp32, [1, 57, 256, 239]> var_1612 = reshape(shape = var_1611, x = var_1609)[name = tensor<string, []>("op_1612")];
            tensor<int32, [4]> x_115_perm_0 = const()[name = tensor<string, []>("x_115_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 256, 57, 239]> x_115 = transpose(perm = x_115_perm_0, x = var_1612)[name = tensor<string, []>("transpose_66")];
            tensor<fp32, [1, 256, 57, 239]> x_117 = add(x = x_115, y = input_275)[name = tensor<string, []>("x_117")];
            tensor<int32, [4]> var_1617_begin_0 = const()[name = tensor<string, []>("op_1617_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_1617_end_0 = const()[name = tensor<string, []>("op_1617_end_0"), val = tensor<int32, [4]>([1, 128, 57, 239])];
            tensor<bool, [4]> var_1617_end_mask_0 = const()[name = tensor<string, []>("op_1617_end_mask_0"), val = tensor<bool, [4]>([true, false, true, true])];
            tensor<fp32, [1, 128, 57, 239]> var_1617 = slice_by_index(begin = var_1617_begin_0, end = var_1617_end_0, end_mask = var_1617_end_mask_0, x = x_117)[name = tensor<string, []>("op_1617")];
            tensor<int32, [4]> var_1621_begin_0 = const()[name = tensor<string, []>("op_1621_begin_0"), val = tensor<int32, [4]>([0, 128, 0, 0])];
            tensor<int32, [4]> var_1621_end_0 = const()[name = tensor<string, []>("op_1621_end_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<bool, [4]> var_1621_end_mask_0 = const()[name = tensor<string, []>("op_1621_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 128, 57, 239]> var_1621 = slice_by_index(begin = var_1621_begin_0, end = var_1621_end_0, end_mask = var_1621_end_mask_0, x = x_117)[name = tensor<string, []>("op_1621")];
            tensor<int32, []> gather_6_batch_dims_0 = const()[name = tensor<string, []>("gather_6_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_6_validate_indices_0 = const()[name = tensor<string, []>("gather_6_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_6_axis_1 = const()[name = tensor<string, []>("gather_6_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 237]> gather_6 = gather(axis = gather_6_axis_1, batch_dims = gather_6_batch_dims_0, indices = select_2, validate_indices = gather_6_validate_indices_0, x = var_1617)[name = tensor<string, []>("gather_6")];
            tensor<int32, []> gather_7_batch_dims_0 = const()[name = tensor<string, []>("gather_7_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_7_validate_indices_0 = const()[name = tensor<string, []>("gather_7_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_7_axis_1 = const()[name = tensor<string, []>("gather_7_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 237]> gather_7 = gather(axis = gather_7_axis_1, batch_dims = gather_7_batch_dims_0, indices = select_2, validate_indices = gather_7_validate_indices_0, x = var_1621)[name = tensor<string, []>("gather_7")];
            tensor<fp32, []> mul_54_y_0 = const()[name = tensor<string, []>("mul_54_y_0"), val = tensor<fp32, []>(-0x1p+0)];
            tensor<fp32, [1, 128, 57, 237]> mul_54 = mul(x = gather_7, y = mul_54_y_0)[name = tensor<string, []>("mul_54")];
            tensor<int32, []> concat_122_axis_0 = const()[name = tensor<string, []>("concat_122_axis_0"), val = tensor<int32, []>(3)];
            tensor<bool, []> concat_122_interleave_0 = const()[name = tensor<string, []>("concat_122_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 128, 57, 476]> concat_122 = concat(axis = concat_122_axis_0, interleave = concat_122_interleave_0, values = (var_1617, gather_6))[name = tensor<string, []>("concat_122")];
            tensor<int32, []> concat_123_axis_0 = const()[name = tensor<string, []>("concat_123_axis_0"), val = tensor<int32, []>(3)];
            tensor<bool, []> concat_123_interleave_0 = const()[name = tensor<string, []>("concat_123_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 128, 57, 476]> concat_123 = concat(axis = concat_123_axis_0, interleave = concat_123_interleave_0, values = (var_1621, mul_54))[name = tensor<string, []>("concat_123")];
            tensor<int32, [4]> transpose_12_perm_0 = const()[name = tensor<string, []>("transpose_12_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [4]> transpose_13_perm_0 = const()[name = tensor<string, []>("transpose_13_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [2]> reshape_210_shape_0 = const()[name = tensor<string, []>("reshape_210_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_12 = transpose(perm = transpose_12_perm_0, x = concat_122)[name = tensor<string, []>("transpose_65")];
            tensor<fp32, [476, 7296]> reshape_210 = reshape(shape = reshape_210_shape_0, x = transpose_12)[name = tensor<string, []>("reshape_210")];
            tensor<int32, [2]> reshape_211_shape_0 = const()[name = tensor<string, []>("reshape_211_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_13 = transpose(perm = transpose_13_perm_0, x = concat_123)[name = tensor<string, []>("transpose_64")];
            tensor<fp32, [476, 7296]> reshape_211 = reshape(shape = reshape_211_shape_0, x = transpose_13)[name = tensor<string, []>("reshape_211")];
            tensor<bool, []> matmul_16_transpose_x_0 = const()[name = tensor<string, []>("matmul_16_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_16_transpose_y_0 = const()[name = tensor<string, []>("matmul_16_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_16 = matmul(transpose_x = matmul_16_transpose_x_0, transpose_y = matmul_16_transpose_y_0, x = cos_0, y = reshape_210)[name = tensor<string, []>("matmul_16")];
            tensor<bool, []> matmul_17_transpose_x_0 = const()[name = tensor<string, []>("matmul_17_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_17_transpose_y_0 = const()[name = tensor<string, []>("matmul_17_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_17 = matmul(transpose_x = matmul_17_transpose_x_0, transpose_y = matmul_17_transpose_y_0, x = sin_0, y = reshape_211)[name = tensor<string, []>("matmul_17")];
            tensor<fp32, [476, 7296]> sub_99 = sub(x = matmul_16, y = matmul_17)[name = tensor<string, []>("sub_99")];
            tensor<int32, [4]> reshape_214_shape_0 = const()[name = tensor<string, []>("reshape_214_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_214 = reshape(shape = reshape_214_shape_0, x = sub_99)[name = tensor<string, []>("reshape_214")];
            tensor<int32, [4]> transpose_14_perm_0 = const()[name = tensor<string, []>("transpose_14_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<fp32, []> _inversed_real_div_58_y_0 = const()[name = tensor<string, []>("_inversed_real_div_58_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [1, 128, 57, 476]> transpose_14 = transpose(perm = transpose_14_perm_0, x = reshape_214)[name = tensor<string, []>("transpose_63")];
            tensor<fp32, [1, 128, 57, 476]> _inversed_real_div_58 = mul(x = transpose_14, y = _inversed_real_div_58_y_0)[name = tensor<string, []>("_inversed_real_div_58")];
            tensor<int32, [5]> reshape_176_shape_0 = const()[name = tensor<string, []>("reshape_176_shape_0"), val = tensor<int32, [5]>([1, 1, 128, 57, 476])];
            tensor<fp32, [1, 1, 128, 57, 476]> reshape_176 = reshape(shape = reshape_176_shape_0, x = _inversed_real_div_58)[name = tensor<string, []>("reshape_176")];
            tensor<int32, [3]> reduce_mean_132_axes_0 = const()[name = tensor<string, []>("reduce_mean_132_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_132_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_132_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_132 = reduce_mean(axes = reduce_mean_132_axes_0, keep_dims = reduce_mean_132_keep_dims_0, x = reshape_176)[name = tensor<string, []>("reduce_mean_132")];
            tensor<fp32, [1, 1, 128, 57, 476]> sub_88 = sub(x = reshape_176, y = reduce_mean_132)[name = tensor<string, []>("sub_88")];
            tensor<fp32, [1, 1, 128, 57, 476]> square_44 = square(x = sub_88)[name = tensor<string, []>("square_44")];
            tensor<int32, [3]> reduce_mean_134_axes_0 = const()[name = tensor<string, []>("reduce_mean_134_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_134_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_134_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_134 = reduce_mean(axes = reduce_mean_134_axes_0, keep_dims = reduce_mean_134_keep_dims_0, x = square_44)[name = tensor<string, []>("reduce_mean_134")];
            tensor<fp32, []> add_104_y_0 = const()[name = tensor<string, []>("add_104_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_104 = add(x = reduce_mean_134, y = add_104_y_0)[name = tensor<string, []>("add_104")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_44 = sqrt(x = add_104)[name = tensor<string, []>("sqrt_44")];
            tensor<fp32, [1, 1, 128, 57, 476]> real_div_44 = real_div(x = sub_88, y = sqrt_44)[name = tensor<string, []>("real_div_44")];
            tensor<int32, [4]> reshape_177_shape_0 = const()[name = tensor<string, []>("reshape_177_shape_0"), val = tensor<int32, [4]>([1, 128, 57, 476])];
            tensor<fp32, [1, 128, 57, 476]> reshape_177 = reshape(shape = reshape_177_shape_0, x = real_div_44)[name = tensor<string, []>("reshape_177")];
            tensor<fp32, [128]> add_105_gamma_0 = const()[name = tensor<string, []>("add_105_gamma_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34711936)))];
            tensor<fp32, [128]> add_105_beta_0 = const()[name = tensor<string, []>("add_105_beta_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34712512)))];
            tensor<fp32, []> add_105_epsilon_0 = const()[name = tensor<string, []>("add_105_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 128, 57, 476]> add_105 = batch_norm(beta = add_105_beta_0, epsilon = add_105_epsilon_0, gamma = add_105_gamma_0, mean = add_73_mean_0, variance = add_73_variance_0, x = reshape_177)[name = tensor<string, []>("add_105")];
            tensor<int32, [4]> var_1648_perm_0 = const()[name = tensor<string, []>("op_1648_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<int32, [3]> var_1652 = const()[name = tensor<string, []>("op_1652"), val = tensor<int32, [3]>([476, 57, 128])];
            tensor<fp32, [1, 476, 57, 128]> var_1648 = transpose(perm = var_1648_perm_0, x = add_105)[name = tensor<string, []>("transpose_62")];
            tensor<fp32, [476, 57, 128]> input_281 = reshape(shape = var_1652, x = var_1648)[name = tensor<string, []>("input_281")];
            tensor<int32, [3]> input_281_batch_first_transpose_perm_0 = const()[name = tensor<string, []>("input_281_batch_first_transpose_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [512]> add_106 = const()[name = tensor<string, []>("add_106"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34713088)))];
            tensor<fp32, [512]> add_107 = const()[name = tensor<string, []>("add_107"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34715200)))];
            tensor<fp32, [512, 128]> concat_84 = const()[name = tensor<string, []>("concat_84"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34717312)))];
            tensor<fp32, [512, 128]> concat_85 = const()[name = tensor<string, []>("concat_85"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(34979520)))];
            tensor<fp32, [512, 128]> concat_86 = const()[name = tensor<string, []>("concat_86"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35241728)))];
            tensor<fp32, [512, 128]> concat_87 = const()[name = tensor<string, []>("concat_87"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35503936)))];
            tensor<string, []> input_283_batch_first_direction_0 = const()[name = tensor<string, []>("input_283_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_283_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_283_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_283_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_283_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_283_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_283_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_283_batch_first_activation_0 = const()[name = tensor<string, []>("input_283_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [57, 476, 128]> input_281_batch_first_transpose = transpose(perm = input_281_batch_first_transpose_perm_0, x = input_281)[name = tensor<string, []>("transpose_61")];
            tensor<fp32, [57, 476, 256]> input_283_batch_first_0, tensor<fp32, [476, 256]> input_283_batch_first_1, tensor<fp32, [476, 256]> input_283_batch_first_2 = lstm(activation = input_283_batch_first_activation_0, bias = add_106, bias_back = add_107, cell_activation = input_283_batch_first_cell_activation_0, direction = input_283_batch_first_direction_0, initial_c = input_243_batch_first_lstm_h0_reshaped, initial_h = input_243_batch_first_lstm_h0_reshaped, output_sequence = input_283_batch_first_output_sequence_0, recurrent_activation = input_283_batch_first_recurrent_activation_0, weight_hh = concat_85, weight_hh_back = concat_87, weight_ih = concat_84, weight_ih_back = concat_86, x = input_281_batch_first_transpose)[name = tensor<string, []>("input_283_batch_first")];
            tensor<int32, [3]> input_283_perm_0 = const()[name = tensor<string, []>("input_283_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [476, 57, 256]> input_283 = transpose(perm = input_283_perm_0, x = input_283_batch_first_0)[name = tensor<string, []>("transpose_60")];
            tensor<fp32, [476, 57, 128]> x_127 = linear(bias = separation_net_dp_modules_4_linear_layers_0_bias, weight = separation_net_dp_modules_4_linear_layers_0_weight, x = input_283)[name = tensor<string, []>("linear_8")];
            tensor<int32, [4]> var_1675 = const()[name = tensor<string, []>("op_1675"), val = tensor<int32, [4]>([1, 476, 57, 128])];
            tensor<fp32, [1, 476, 57, 128]> var_1676 = reshape(shape = var_1675, x = x_127)[name = tensor<string, []>("op_1676")];
            tensor<int32, [4]> x_129_perm_0 = const()[name = tensor<string, []>("x_129_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<fp32, [1, 128, 57, 476]> x_129 = transpose(perm = x_129_perm_0, x = var_1676)[name = tensor<string, []>("transpose_59")];
            tensor<fp32, [1, 128, 57, 476]> input_285 = add(x = x_129, y = _inversed_real_div_58)[name = tensor<string, []>("input_285")];
            tensor<int32, [5]> reshape_180_shape_0 = const()[name = tensor<string, []>("reshape_180_shape_0"), val = tensor<int32, [5]>([1, 1, 128, 57, 476])];
            tensor<fp32, [1, 1, 128, 57, 476]> reshape_180 = reshape(shape = reshape_180_shape_0, x = input_285)[name = tensor<string, []>("reshape_180")];
            tensor<int32, [3]> reduce_mean_135_axes_0 = const()[name = tensor<string, []>("reduce_mean_135_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_135_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_135_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_135 = reduce_mean(axes = reduce_mean_135_axes_0, keep_dims = reduce_mean_135_keep_dims_0, x = reshape_180)[name = tensor<string, []>("reduce_mean_135")];
            tensor<fp32, [1, 1, 128, 57, 476]> sub_90 = sub(x = reshape_180, y = reduce_mean_135)[name = tensor<string, []>("sub_90")];
            tensor<fp32, [1, 1, 128, 57, 476]> square_45 = square(x = sub_90)[name = tensor<string, []>("square_45")];
            tensor<int32, [3]> reduce_mean_137_axes_0 = const()[name = tensor<string, []>("reduce_mean_137_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_137_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_137_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_137 = reduce_mean(axes = reduce_mean_137_axes_0, keep_dims = reduce_mean_137_keep_dims_0, x = square_45)[name = tensor<string, []>("reduce_mean_137")];
            tensor<fp32, []> add_108_y_0 = const()[name = tensor<string, []>("add_108_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_108 = add(x = reduce_mean_137, y = add_108_y_0)[name = tensor<string, []>("add_108")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_45 = sqrt(x = add_108)[name = tensor<string, []>("sqrt_45")];
            tensor<fp32, [1, 1, 128, 57, 476]> real_div_45 = real_div(x = sub_90, y = sqrt_45)[name = tensor<string, []>("real_div_45")];
            tensor<int32, [4]> reshape_181_shape_0 = const()[name = tensor<string, []>("reshape_181_shape_0"), val = tensor<int32, [4]>([1, 128, 57, 476])];
            tensor<fp32, [1, 128, 57, 476]> reshape_181 = reshape(shape = reshape_181_shape_0, x = real_div_45)[name = tensor<string, []>("reshape_181")];
            tensor<fp32, [128]> add_109_gamma_0 = const()[name = tensor<string, []>("add_109_gamma_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35766144)))];
            tensor<fp32, [128]> add_109_beta_0 = const()[name = tensor<string, []>("add_109_beta_0"), val = tensor<fp32, [128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35766720)))];
            tensor<fp32, []> add_109_epsilon_0 = const()[name = tensor<string, []>("add_109_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 128, 57, 476]> add_109 = batch_norm(beta = add_109_beta_0, epsilon = add_109_epsilon_0, gamma = add_109_gamma_0, mean = add_73_mean_0, variance = add_73_variance_0, x = reshape_181)[name = tensor<string, []>("add_109")];
            tensor<int32, [4]> var_1682_perm_0 = const()[name = tensor<string, []>("op_1682_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_1686 = const()[name = tensor<string, []>("op_1686"), val = tensor<int32, [3]>([57, 128, 476])];
            tensor<fp32, [1, 57, 128, 476]> var_1682 = transpose(perm = var_1682_perm_0, x = add_109)[name = tensor<string, []>("transpose_58")];
            tensor<fp32, [57, 128, 476]> var_1687 = reshape(shape = var_1686, x = var_1682)[name = tensor<string, []>("op_1687")];
            tensor<int32, [3]> transpose_35_perm_0 = const()[name = tensor<string, []>("transpose_35_perm_0"), val = tensor<int32, [3]>([2, 0, 1])];
            tensor<fp32, [512]> add_110 = const()[name = tensor<string, []>("add_110"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35767296)))];
            tensor<fp32, [512]> add_111 = const()[name = tensor<string, []>("add_111"), val = tensor<fp32, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35769408)))];
            tensor<fp32, [512, 128]> concat_94 = const()[name = tensor<string, []>("concat_94"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(35771520)))];
            tensor<fp32, [512, 128]> concat_95 = const()[name = tensor<string, []>("concat_95"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36033728)))];
            tensor<fp32, [512, 128]> concat_96 = const()[name = tensor<string, []>("concat_96"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36295936)))];
            tensor<fp32, [512, 128]> concat_97 = const()[name = tensor<string, []>("concat_97"), val = tensor<fp32, [512, 128]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36558144)))];
            tensor<string, []> input_289_batch_first_direction_0 = const()[name = tensor<string, []>("input_289_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_289_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_289_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_289_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_289_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_289_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_289_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_289_batch_first_activation_0 = const()[name = tensor<string, []>("input_289_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [476, 57, 128]> transpose_35 = transpose(perm = transpose_35_perm_0, x = var_1687)[name = tensor<string, []>("transpose_57")];
            tensor<fp32, [476, 57, 256]> input_289_batch_first_0, tensor<fp32, [57, 256]> input_289_batch_first_1, tensor<fp32, [57, 256]> input_289_batch_first_2 = lstm(activation = input_289_batch_first_activation_0, bias = add_110, bias_back = add_111, cell_activation = input_289_batch_first_cell_activation_0, direction = input_289_batch_first_direction_0, initial_c = input_249_batch_first_lstm_h0_reshaped, initial_h = input_249_batch_first_lstm_h0_reshaped, output_sequence = input_289_batch_first_output_sequence_0, recurrent_activation = input_289_batch_first_recurrent_activation_0, weight_hh = concat_95, weight_hh_back = concat_97, weight_ih = concat_94, weight_ih_back = concat_96, x = transpose_35)[name = tensor<string, []>("input_289_batch_first")];
            tensor<int32, [3]> input_289_perm_0 = const()[name = tensor<string, []>("input_289_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [57, 476, 256]> input_289 = transpose(perm = input_289_perm_0, x = input_289_batch_first_0)[name = tensor<string, []>("transpose_56")];
            tensor<fp32, [57, 476, 128]> x_133 = linear(bias = separation_net_dp_modules_4_linear_layers_1_bias, weight = separation_net_dp_modules_4_linear_layers_1_weight, x = input_289)[name = tensor<string, []>("linear_9")];
            tensor<int32, [3]> var_1710_perm_0 = const()[name = tensor<string, []>("op_1710_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
            tensor<int32, [4]> var_1712 = const()[name = tensor<string, []>("op_1712"), val = tensor<int32, [4]>([1, 57, 128, 476])];
            tensor<fp32, [57, 128, 476]> var_1710 = transpose(perm = var_1710_perm_0, x = x_133)[name = tensor<string, []>("transpose_55")];
            tensor<fp32, [1, 57, 128, 476]> var_1713 = reshape(shape = var_1712, x = var_1710)[name = tensor<string, []>("op_1713")];
            tensor<int32, [4]> x_135_perm_0 = const()[name = tensor<string, []>("x_135_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 128, 57, 476]> x_135 = transpose(perm = x_135_perm_0, x = var_1713)[name = tensor<string, []>("transpose_54")];
            tensor<fp32, [1, 128, 57, 476]> x_137 = add(x = x_135, y = input_285)[name = tensor<string, []>("x_137")];
            tensor<int32, [4]> transpose_16_perm_0 = const()[name = tensor<string, []>("transpose_16_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [2]> reshape_216_shape_0 = const()[name = tensor<string, []>("reshape_216_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_16 = transpose(perm = transpose_16_perm_0, x = x_137)[name = tensor<string, []>("transpose_53")];
            tensor<fp32, [476, 7296]> reshape_216 = reshape(shape = reshape_216_shape_0, x = transpose_16)[name = tensor<string, []>("reshape_216")];
            tensor<bool, []> matmul_21_transpose_x_0 = const()[name = tensor<string, []>("matmul_21_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_21_transpose_y_0 = const()[name = tensor<string, []>("matmul_21_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_21 = matmul(transpose_x = matmul_21_transpose_x_0, transpose_y = matmul_21_transpose_y_0, x = cos_0, y = reshape_216)[name = tensor<string, []>("matmul_21")];
            tensor<bool, []> matmul_23_transpose_x_0 = const()[name = tensor<string, []>("matmul_23_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_23_transpose_y_0 = const()[name = tensor<string, []>("matmul_23_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_23 = matmul(transpose_x = matmul_23_transpose_x_0, transpose_y = matmul_23_transpose_y_0, x = sin_0, y = reshape_216)[name = tensor<string, []>("matmul_23")];
            tensor<fp32, []> mul_57_y_0 = const()[name = tensor<string, []>("mul_57_y_0"), val = tensor<fp32, []>(-0x1p+0)];
            tensor<fp32, [476, 7296]> mul_57 = mul(x = matmul_23, y = mul_57_y_0)[name = tensor<string, []>("mul_57")];
            tensor<int32, [4]> reshape_220_shape_0 = const()[name = tensor<string, []>("reshape_220_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_220 = reshape(shape = reshape_220_shape_0, x = matmul_21)[name = tensor<string, []>("reshape_220")];
            tensor<int32, [4]> reshape_221_shape_0 = const()[name = tensor<string, []>("reshape_221_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_221 = reshape(shape = reshape_221_shape_0, x = mul_57)[name = tensor<string, []>("reshape_221")];
            tensor<int32, [4]> transpose_18_perm_0 = const()[name = tensor<string, []>("transpose_18_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [4]> transpose_19_perm_0 = const()[name = tensor<string, []>("transpose_19_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<fp32, []> _inversed_real_div_61_y_0 = const()[name = tensor<string, []>("_inversed_real_div_61_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [1, 128, 57, 476]> transpose_18 = transpose(perm = transpose_18_perm_0, x = reshape_220)[name = tensor<string, []>("transpose_52")];
            tensor<fp32, [1, 128, 57, 476]> _inversed_real_div_61 = mul(x = transpose_18, y = _inversed_real_div_61_y_0)[name = tensor<string, []>("_inversed_real_div_61")];
            tensor<fp32, []> _inversed_real_div_62_y_0 = const()[name = tensor<string, []>("_inversed_real_div_62_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [1, 128, 57, 476]> transpose_19 = transpose(perm = transpose_19_perm_0, x = reshape_221)[name = tensor<string, []>("transpose_51")];
            tensor<fp32, [1, 128, 57, 476]> _inversed_real_div_62 = mul(x = transpose_19, y = _inversed_real_div_62_y_0)[name = tensor<string, []>("_inversed_real_div_62")];
            tensor<int32, []> gather_8_batch_dims_0 = const()[name = tensor<string, []>("gather_8_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_8_validate_indices_0 = const()[name = tensor<string, []>("gather_8_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_8_axis_1 = const()[name = tensor<string, []>("gather_8_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 239]> gather_8 = gather(axis = gather_8_axis_1, batch_dims = gather_8_batch_dims_0, indices = select_0, validate_indices = gather_8_validate_indices_0, x = _inversed_real_div_61)[name = tensor<string, []>("gather_8")];
            tensor<int32, []> gather_9_batch_dims_0 = const()[name = tensor<string, []>("gather_9_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_9_validate_indices_0 = const()[name = tensor<string, []>("gather_9_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_9_axis_1 = const()[name = tensor<string, []>("gather_9_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 239]> gather_9 = gather(axis = gather_9_axis_1, batch_dims = gather_9_batch_dims_0, indices = select_0, validate_indices = gather_9_validate_indices_0, x = _inversed_real_div_62)[name = tensor<string, []>("gather_9")];
            tensor<bool, []> x_143_interleave_0 = const()[name = tensor<string, []>("x_143_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 256, 57, 239]> x_143 = concat(axis = var_1209, interleave = x_143_interleave_0, values = (gather_8, gather_9))[name = tensor<string, []>("x_143")];
            tensor<int32, [5]> reshape_184_shape_0 = const()[name = tensor<string, []>("reshape_184_shape_0"), val = tensor<int32, [5]>([1, 1, 256, 57, 239])];
            tensor<fp32, [1, 1, 256, 57, 239]> reshape_184 = reshape(shape = reshape_184_shape_0, x = x_143)[name = tensor<string, []>("reshape_184")];
            tensor<int32, [3]> reduce_mean_138_axes_0 = const()[name = tensor<string, []>("reduce_mean_138_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_138_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_138_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_138 = reduce_mean(axes = reduce_mean_138_axes_0, keep_dims = reduce_mean_138_keep_dims_0, x = reshape_184)[name = tensor<string, []>("reduce_mean_138")];
            tensor<fp32, [1, 1, 256, 57, 239]> sub_92 = sub(x = reshape_184, y = reduce_mean_138)[name = tensor<string, []>("sub_92")];
            tensor<fp32, [1, 1, 256, 57, 239]> square_46 = square(x = sub_92)[name = tensor<string, []>("square_46")];
            tensor<int32, [3]> reduce_mean_140_axes_0 = const()[name = tensor<string, []>("reduce_mean_140_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_140_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_140_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_140 = reduce_mean(axes = reduce_mean_140_axes_0, keep_dims = reduce_mean_140_keep_dims_0, x = square_46)[name = tensor<string, []>("reduce_mean_140")];
            tensor<fp32, []> add_112_y_0 = const()[name = tensor<string, []>("add_112_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_112 = add(x = reduce_mean_140, y = add_112_y_0)[name = tensor<string, []>("add_112")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_46 = sqrt(x = add_112)[name = tensor<string, []>("sqrt_46")];
            tensor<fp32, [1, 1, 256, 57, 239]> real_div_46 = real_div(x = sub_92, y = sqrt_46)[name = tensor<string, []>("real_div_46")];
            tensor<int32, [4]> reshape_185_shape_0 = const()[name = tensor<string, []>("reshape_185_shape_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<fp32, [1, 256, 57, 239]> reshape_185 = reshape(shape = reshape_185_shape_0, x = real_div_46)[name = tensor<string, []>("reshape_185")];
            tensor<fp32, [256]> add_113_gamma_0 = const()[name = tensor<string, []>("add_113_gamma_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36820352)))];
            tensor<fp32, [256]> add_113_beta_0 = const()[name = tensor<string, []>("add_113_beta_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36821440)))];
            tensor<fp32, []> add_113_epsilon_0 = const()[name = tensor<string, []>("add_113_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 256, 57, 239]> add_113 = batch_norm(beta = add_113_beta_0, epsilon = add_113_epsilon_0, gamma = add_113_gamma_0, mean = add_81_mean_0, variance = add_81_variance_0, x = reshape_185)[name = tensor<string, []>("add_113")];
            tensor<int32, [4]> var_1744_perm_0 = const()[name = tensor<string, []>("op_1744_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<int32, [3]> var_1748 = const()[name = tensor<string, []>("op_1748"), val = tensor<int32, [3]>([239, 57, 256])];
            tensor<fp32, [1, 239, 57, 256]> var_1744 = transpose(perm = var_1744_perm_0, x = add_113)[name = tensor<string, []>("transpose_50")];
            tensor<fp32, [239, 57, 256]> input_291 = reshape(shape = var_1748, x = var_1744)[name = tensor<string, []>("input_291")];
            tensor<int32, [3]> input_291_batch_first_transpose_perm_0 = const()[name = tensor<string, []>("input_291_batch_first_transpose_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [1024]> add_114 = const()[name = tensor<string, []>("add_114"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36822528)))];
            tensor<fp32, [1024]> add_115 = const()[name = tensor<string, []>("add_115"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36826688)))];
            tensor<fp32, [1024, 256]> concat_104 = const()[name = tensor<string, []>("concat_104"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(36830848)))];
            tensor<fp32, [1024, 256]> concat_105 = const()[name = tensor<string, []>("concat_105"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(37879488)))];
            tensor<fp32, [1024, 256]> concat_106 = const()[name = tensor<string, []>("concat_106"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(38928128)))];
            tensor<fp32, [1024, 256]> concat_107 = const()[name = tensor<string, []>("concat_107"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(39976768)))];
            tensor<string, []> input_293_batch_first_direction_0 = const()[name = tensor<string, []>("input_293_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_293_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_293_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_293_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_293_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_293_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_293_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_293_batch_first_activation_0 = const()[name = tensor<string, []>("input_293_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [57, 239, 256]> input_291_batch_first_transpose = transpose(perm = input_291_batch_first_transpose_perm_0, x = input_291)[name = tensor<string, []>("transpose_49")];
            tensor<fp32, [57, 239, 512]> input_293_batch_first_0, tensor<fp32, [239, 512]> input_293_batch_first_1, tensor<fp32, [239, 512]> input_293_batch_first_2 = lstm(activation = input_293_batch_first_activation_0, bias = add_114, bias_back = add_115, cell_activation = input_293_batch_first_cell_activation_0, direction = input_293_batch_first_direction_0, initial_c = input_253_batch_first_lstm_h0_reshaped, initial_h = input_253_batch_first_lstm_h0_reshaped, output_sequence = input_293_batch_first_output_sequence_0, recurrent_activation = input_293_batch_first_recurrent_activation_0, weight_hh = concat_105, weight_hh_back = concat_107, weight_ih = concat_104, weight_ih_back = concat_106, x = input_291_batch_first_transpose)[name = tensor<string, []>("input_293_batch_first")];
            tensor<int32, [3]> input_293_perm_0 = const()[name = tensor<string, []>("input_293_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [239, 57, 512]> input_293 = transpose(perm = input_293_perm_0, x = input_293_batch_first_0)[name = tensor<string, []>("transpose_48")];
            tensor<fp32, [239, 57, 256]> x_147 = linear(bias = separation_net_dp_modules_5_linear_layers_0_bias, weight = separation_net_dp_modules_5_linear_layers_0_weight, x = input_293)[name = tensor<string, []>("linear_10")];
            tensor<int32, [4]> var_1771 = const()[name = tensor<string, []>("op_1771"), val = tensor<int32, [4]>([1, 239, 57, 256])];
            tensor<fp32, [1, 239, 57, 256]> var_1772 = reshape(shape = var_1771, x = x_147)[name = tensor<string, []>("op_1772")];
            tensor<int32, [4]> x_149_perm_0 = const()[name = tensor<string, []>("x_149_perm_0"), val = tensor<int32, [4]>([0, 3, 2, 1])];
            tensor<fp32, [1, 256, 57, 239]> x_149 = transpose(perm = x_149_perm_0, x = var_1772)[name = tensor<string, []>("transpose_47")];
            tensor<fp32, [1, 256, 57, 239]> input_295 = add(x = x_149, y = x_143)[name = tensor<string, []>("input_295")];
            tensor<int32, [5]> reshape_188_shape_0 = const()[name = tensor<string, []>("reshape_188_shape_0"), val = tensor<int32, [5]>([1, 1, 256, 57, 239])];
            tensor<fp32, [1, 1, 256, 57, 239]> reshape_188 = reshape(shape = reshape_188_shape_0, x = input_295)[name = tensor<string, []>("reshape_188")];
            tensor<int32, [3]> reduce_mean_141_axes_0 = const()[name = tensor<string, []>("reduce_mean_141_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_141_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_141_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_141 = reduce_mean(axes = reduce_mean_141_axes_0, keep_dims = reduce_mean_141_keep_dims_0, x = reshape_188)[name = tensor<string, []>("reduce_mean_141")];
            tensor<fp32, [1, 1, 256, 57, 239]> sub_94 = sub(x = reshape_188, y = reduce_mean_141)[name = tensor<string, []>("sub_94")];
            tensor<fp32, [1, 1, 256, 57, 239]> square_47 = square(x = sub_94)[name = tensor<string, []>("square_47")];
            tensor<int32, [3]> reduce_mean_143_axes_0 = const()[name = tensor<string, []>("reduce_mean_143_axes_0"), val = tensor<int32, [3]>([2, 3, 4])];
            tensor<bool, []> reduce_mean_143_keep_dims_0 = const()[name = tensor<string, []>("reduce_mean_143_keep_dims_0"), val = tensor<bool, []>(true)];
            tensor<fp32, [1, 1, 1, 1, 1]> reduce_mean_143 = reduce_mean(axes = reduce_mean_143_axes_0, keep_dims = reduce_mean_143_keep_dims_0, x = square_47)[name = tensor<string, []>("reduce_mean_143")];
            tensor<fp32, []> add_116_y_0 = const()[name = tensor<string, []>("add_116_y_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 1, 1, 1, 1]> add_116 = add(x = reduce_mean_143, y = add_116_y_0)[name = tensor<string, []>("add_116")];
            tensor<fp32, [1, 1, 1, 1, 1]> sqrt_47 = sqrt(x = add_116)[name = tensor<string, []>("sqrt_47")];
            tensor<fp32, [1, 1, 256, 57, 239]> real_div_47 = real_div(x = sub_94, y = sqrt_47)[name = tensor<string, []>("real_div_47")];
            tensor<int32, [4]> reshape_189_shape_0 = const()[name = tensor<string, []>("reshape_189_shape_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<fp32, [1, 256, 57, 239]> reshape_189 = reshape(shape = reshape_189_shape_0, x = real_div_47)[name = tensor<string, []>("reshape_189")];
            tensor<fp32, [256]> add_117_gamma_0 = const()[name = tensor<string, []>("add_117_gamma_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(41025408)))];
            tensor<fp32, [256]> add_117_beta_0 = const()[name = tensor<string, []>("add_117_beta_0"), val = tensor<fp32, [256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(41026496)))];
            tensor<fp32, []> add_117_epsilon_0 = const()[name = tensor<string, []>("add_117_epsilon_0"), val = tensor<fp32, []>(0x1.4f8b58p-17)];
            tensor<fp32, [1, 256, 57, 239]> add_117 = batch_norm(beta = add_117_beta_0, epsilon = add_117_epsilon_0, gamma = add_117_gamma_0, mean = add_81_mean_0, variance = add_81_variance_0, x = reshape_189)[name = tensor<string, []>("add_117")];
            tensor<int32, [4]> var_1778_perm_0 = const()[name = tensor<string, []>("op_1778_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<int32, [3]> var_1782 = const()[name = tensor<string, []>("op_1782"), val = tensor<int32, [3]>([57, 256, 239])];
            tensor<fp32, [1, 57, 256, 239]> var_1778 = transpose(perm = var_1778_perm_0, x = add_117)[name = tensor<string, []>("transpose_46")];
            tensor<fp32, [57, 256, 239]> var_1783 = reshape(shape = var_1782, x = var_1778)[name = tensor<string, []>("op_1783")];
            tensor<int32, [3]> transpose_36_perm_0 = const()[name = tensor<string, []>("transpose_36_perm_0"), val = tensor<int32, [3]>([2, 0, 1])];
            tensor<fp32, [1024]> add_118 = const()[name = tensor<string, []>("add_118"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(41027584)))];
            tensor<fp32, [1024]> add_119 = const()[name = tensor<string, []>("add_119"), val = tensor<fp32, [1024]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(41031744)))];
            tensor<fp32, [1024, 256]> concat_114 = const()[name = tensor<string, []>("concat_114"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(41035904)))];
            tensor<fp32, [1024, 256]> concat_115 = const()[name = tensor<string, []>("concat_115"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(42084544)))];
            tensor<fp32, [1024, 256]> concat_116 = const()[name = tensor<string, []>("concat_116"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(43133184)))];
            tensor<fp32, [1024, 256]> concat_117 = const()[name = tensor<string, []>("concat_117"), val = tensor<fp32, [1024, 256]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(44181824)))];
            tensor<string, []> input_299_batch_first_direction_0 = const()[name = tensor<string, []>("input_299_batch_first_direction_0"), val = tensor<string, []>("bidirectional")];
            tensor<bool, []> input_299_batch_first_output_sequence_0 = const()[name = tensor<string, []>("input_299_batch_first_output_sequence_0"), val = tensor<bool, []>(true)];
            tensor<string, []> input_299_batch_first_recurrent_activation_0 = const()[name = tensor<string, []>("input_299_batch_first_recurrent_activation_0"), val = tensor<string, []>("sigmoid")];
            tensor<string, []> input_299_batch_first_cell_activation_0 = const()[name = tensor<string, []>("input_299_batch_first_cell_activation_0"), val = tensor<string, []>("tanh")];
            tensor<string, []> input_299_batch_first_activation_0 = const()[name = tensor<string, []>("input_299_batch_first_activation_0"), val = tensor<string, []>("tanh")];
            tensor<fp32, [239, 57, 256]> transpose_36 = transpose(perm = transpose_36_perm_0, x = var_1783)[name = tensor<string, []>("transpose_45")];
            tensor<fp32, [239, 57, 512]> input_299_batch_first_0, tensor<fp32, [57, 512]> input_299_batch_first_1, tensor<fp32, [57, 512]> input_299_batch_first_2 = lstm(activation = input_299_batch_first_activation_0, bias = add_118, bias_back = add_119, cell_activation = input_299_batch_first_cell_activation_0, direction = input_299_batch_first_direction_0, initial_c = input_259_batch_first_lstm_h0_reshaped, initial_h = input_259_batch_first_lstm_h0_reshaped, output_sequence = input_299_batch_first_output_sequence_0, recurrent_activation = input_299_batch_first_recurrent_activation_0, weight_hh = concat_115, weight_hh_back = concat_117, weight_ih = concat_114, weight_ih_back = concat_116, x = transpose_36)[name = tensor<string, []>("input_299_batch_first")];
            tensor<int32, [3]> input_299_perm_0 = const()[name = tensor<string, []>("input_299_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
            tensor<fp32, [57, 239, 512]> input_299 = transpose(perm = input_299_perm_0, x = input_299_batch_first_0)[name = tensor<string, []>("transpose_44")];
            tensor<fp32, [57, 239, 256]> x_153 = linear(bias = separation_net_dp_modules_5_linear_layers_1_bias, weight = separation_net_dp_modules_5_linear_layers_1_weight, x = input_299)[name = tensor<string, []>("linear_11")];
            tensor<int32, [3]> var_1806_perm_0 = const()[name = tensor<string, []>("op_1806_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
            tensor<int32, [4]> var_1808 = const()[name = tensor<string, []>("op_1808"), val = tensor<int32, [4]>([1, 57, 256, 239])];
            tensor<fp32, [57, 256, 239]> var_1806 = transpose(perm = var_1806_perm_0, x = x_153)[name = tensor<string, []>("transpose_43")];
            tensor<fp32, [1, 57, 256, 239]> var_1809 = reshape(shape = var_1808, x = var_1806)[name = tensor<string, []>("op_1809")];
            tensor<int32, [4]> x_155_perm_0 = const()[name = tensor<string, []>("x_155_perm_0"), val = tensor<int32, [4]>([0, 2, 1, 3])];
            tensor<fp32, [1, 256, 57, 239]> x_155 = transpose(perm = x_155_perm_0, x = var_1809)[name = tensor<string, []>("transpose_42")];
            tensor<fp32, [1, 256, 57, 239]> x_157 = add(x = x_155, y = input_295)[name = tensor<string, []>("x_157")];
            tensor<int32, [4]> var_1814_begin_0 = const()[name = tensor<string, []>("op_1814_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_1814_end_0 = const()[name = tensor<string, []>("op_1814_end_0"), val = tensor<int32, [4]>([1, 128, 57, 239])];
            tensor<bool, [4]> var_1814_end_mask_0 = const()[name = tensor<string, []>("op_1814_end_mask_0"), val = tensor<bool, [4]>([true, false, true, true])];
            tensor<fp32, [1, 128, 57, 239]> var_1814 = slice_by_index(begin = var_1814_begin_0, end = var_1814_end_0, end_mask = var_1814_end_mask_0, x = x_157)[name = tensor<string, []>("op_1814")];
            tensor<int32, [4]> var_1818_begin_0 = const()[name = tensor<string, []>("op_1818_begin_0"), val = tensor<int32, [4]>([0, 128, 0, 0])];
            tensor<int32, [4]> var_1818_end_0 = const()[name = tensor<string, []>("op_1818_end_0"), val = tensor<int32, [4]>([1, 256, 57, 239])];
            tensor<bool, [4]> var_1818_end_mask_0 = const()[name = tensor<string, []>("op_1818_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 128, 57, 239]> var_1818 = slice_by_index(begin = var_1818_begin_0, end = var_1818_end_0, end_mask = var_1818_end_mask_0, x = x_157)[name = tensor<string, []>("op_1818")];
            tensor<int32, []> gather_10_batch_dims_0 = const()[name = tensor<string, []>("gather_10_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_10_validate_indices_0 = const()[name = tensor<string, []>("gather_10_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_10_axis_1 = const()[name = tensor<string, []>("gather_10_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 237]> gather_10 = gather(axis = gather_10_axis_1, batch_dims = gather_10_batch_dims_0, indices = select_2, validate_indices = gather_10_validate_indices_0, x = var_1814)[name = tensor<string, []>("gather_10")];
            tensor<int32, []> gather_11_batch_dims_0 = const()[name = tensor<string, []>("gather_11_batch_dims_0"), val = tensor<int32, []>(0)];
            tensor<bool, []> gather_11_validate_indices_0 = const()[name = tensor<string, []>("gather_11_validate_indices_0"), val = tensor<bool, []>(false)];
            tensor<int32, []> gather_11_axis_1 = const()[name = tensor<string, []>("gather_11_axis_1"), val = tensor<int32, []>(3)];
            tensor<fp32, [1, 128, 57, 237]> gather_11 = gather(axis = gather_11_axis_1, batch_dims = gather_11_batch_dims_0, indices = select_2, validate_indices = gather_11_validate_indices_0, x = var_1818)[name = tensor<string, []>("gather_11")];
            tensor<fp32, []> mul_58_y_0 = const()[name = tensor<string, []>("mul_58_y_0"), val = tensor<fp32, []>(-0x1p+0)];
            tensor<fp32, [1, 128, 57, 237]> mul_58 = mul(x = gather_11, y = mul_58_y_0)[name = tensor<string, []>("mul_58")];
            tensor<int32, []> concat_124_axis_0 = const()[name = tensor<string, []>("concat_124_axis_0"), val = tensor<int32, []>(3)];
            tensor<bool, []> concat_124_interleave_0 = const()[name = tensor<string, []>("concat_124_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 128, 57, 476]> concat_124 = concat(axis = concat_124_axis_0, interleave = concat_124_interleave_0, values = (var_1814, gather_10))[name = tensor<string, []>("concat_124")];
            tensor<int32, []> concat_125_axis_0 = const()[name = tensor<string, []>("concat_125_axis_0"), val = tensor<int32, []>(3)];
            tensor<bool, []> concat_125_interleave_0 = const()[name = tensor<string, []>("concat_125_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 128, 57, 476]> concat_125 = concat(axis = concat_125_axis_0, interleave = concat_125_interleave_0, values = (var_1818, mul_58))[name = tensor<string, []>("concat_125")];
            tensor<int32, [4]> transpose_20_perm_0 = const()[name = tensor<string, []>("transpose_20_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [4]> transpose_21_perm_0 = const()[name = tensor<string, []>("transpose_21_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<int32, [2]> reshape_222_shape_0 = const()[name = tensor<string, []>("reshape_222_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_20 = transpose(perm = transpose_20_perm_0, x = concat_124)[name = tensor<string, []>("transpose_41")];
            tensor<fp32, [476, 7296]> reshape_222 = reshape(shape = reshape_222_shape_0, x = transpose_20)[name = tensor<string, []>("reshape_222")];
            tensor<int32, [2]> reshape_223_shape_0 = const()[name = tensor<string, []>("reshape_223_shape_0"), val = tensor<int32, [2]>([476, -1])];
            tensor<fp32, [476, 128, 57, 1]> transpose_21 = transpose(perm = transpose_21_perm_0, x = concat_125)[name = tensor<string, []>("transpose_40")];
            tensor<fp32, [476, 7296]> reshape_223 = reshape(shape = reshape_223_shape_0, x = transpose_21)[name = tensor<string, []>("reshape_223")];
            tensor<bool, []> matmul_26_transpose_x_0 = const()[name = tensor<string, []>("matmul_26_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_26_transpose_y_0 = const()[name = tensor<string, []>("matmul_26_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_26 = matmul(transpose_x = matmul_26_transpose_x_0, transpose_y = matmul_26_transpose_y_0, x = cos_0, y = reshape_222)[name = tensor<string, []>("matmul_26")];
            tensor<bool, []> matmul_27_transpose_x_0 = const()[name = tensor<string, []>("matmul_27_transpose_x_0"), val = tensor<bool, []>(false)];
            tensor<bool, []> matmul_27_transpose_y_0 = const()[name = tensor<string, []>("matmul_27_transpose_y_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [476, 7296]> matmul_27 = matmul(transpose_x = matmul_27_transpose_x_0, transpose_y = matmul_27_transpose_y_0, x = sin_0, y = reshape_223)[name = tensor<string, []>("matmul_27")];
            tensor<fp32, [476, 7296]> sub_101 = sub(x = matmul_26, y = matmul_27)[name = tensor<string, []>("sub_101")];
            tensor<int32, [4]> reshape_226_shape_0 = const()[name = tensor<string, []>("reshape_226_shape_0"), val = tensor<int32, [4]>([476, 128, 57, 1])];
            tensor<fp32, [476, 128, 57, 1]> reshape_226 = reshape(shape = reshape_226_shape_0, x = sub_101)[name = tensor<string, []>("reshape_226")];
            tensor<fp32, []> _inversed_real_div_64_y_0 = const()[name = tensor<string, []>("_inversed_real_div_64_y_0"), val = tensor<fp32, []>(0x1.777acep-5)];
            tensor<fp32, [476, 128, 57, 1]> _inversed_real_div_64 = mul(x = reshape_226, y = _inversed_real_div_64_y_0)[name = tensor<string, []>("_inversed_real_div_64")];
            tensor<int32, [4]> transpose_37_perm_0 = const()[name = tensor<string, []>("transpose_37_perm_0"), val = tensor<int32, [4]>([3, 1, 2, 0])];
            tensor<fp32, [1, 128, 57, 476]> transpose_37 = transpose(perm = transpose_37_perm_0, x = _inversed_real_div_64)[name = tensor<string, []>("transpose_39")];
            tensor<fp32, [1, 128, 57, 476]> x_165 = add(x = transpose_37, y = transpose_28)[name = tensor<string, []>("x_165")];
            tensor<int32, [4]> var_1830 = const()[name = tensor<string, []>("op_1830"), val = tensor<int32, [4]>([1, 2, 1, 1])];
            tensor<fp32, [1, 256, 57, 476]> input_301 = tile(reps = var_1830, x = x_165)[name = tensor<string, []>("input_301")];
            tensor<string, []> input_303_pad_type_0 = const()[name = tensor<string, []>("input_303_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [4]> input_303_pad_0 = const()[name = tensor<string, []>("input_303_pad_0"), val = tensor<int32, [4]>([1, 1, 1, 1])];
            tensor<int32, [2]> input_303_strides_0 = const()[name = tensor<string, []>("input_303_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> input_303_dilations_0 = const()[name = tensor<string, []>("input_303_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_303_groups_0 = const()[name = tensor<string, []>("input_303_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 256, 57, 476]> input_303 = conv(bias = decoder_0_0_conv_bias, dilations = input_303_dilations_0, groups = input_303_groups_0, pad = input_303_pad_0, pad_type = input_303_pad_type_0, strides = input_303_strides_0, weight = decoder_0_0_conv_weight, x = input_301)[name = tensor<string, []>("input_303")];
            tensor<int32, []> x_167_split_num_splits_0 = const()[name = tensor<string, []>("x_167_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> x_167_split_axis_0 = const()[name = tensor<string, []>("x_167_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 128, 57, 476]> x_167_split_0, tensor<fp32, [1, 128, 57, 476]> x_167_split_1 = split(axis = x_167_split_axis_0, num_splits = x_167_split_num_splits_0, x = input_303)[name = tensor<string, []>("x_167_split")];
            tensor<fp32, [1, 128, 57, 476]> x_167_split_1_sigmoid = sigmoid(x = x_167_split_1)[name = tensor<string, []>("x_167_split_1_sigmoid")];
            tensor<fp32, [1, 128, 57, 476]> x_167 = mul(x = x_167_split_0, y = x_167_split_1_sigmoid)[name = tensor<string, []>("x_167")];
            tensor<int32, []> var_1848 = const()[name = tensor<string, []>("op_1848"), val = tensor<int32, []>(2)];
            tensor<int32, [4]> var_1864_begin_0 = const()[name = tensor<string, []>("op_1864_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_1864_end_0 = const()[name = tensor<string, []>("op_1864_end_0"), val = tensor<int32, [4]>([1, 128, 33, 476])];
            tensor<bool, [4]> var_1864_end_mask_0 = const()[name = tensor<string, []>("op_1864_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 128, 33, 476]> var_1864 = slice_by_index(begin = var_1864_begin_0, end = var_1864_end_0, end_mask = var_1864_end_mask_0, x = x_167)[name = tensor<string, []>("op_1864")];
            tensor<string, []> out_1_pad_type_0 = const()[name = tensor<string, []>("out_1_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [4]> out_1_pad_0 = const()[name = tensor<string, []>("out_1_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_1_strides_0 = const()[name = tensor<string, []>("out_1_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> out_1_dilations_0 = const()[name = tensor<string, []>("out_1_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_1_groups_0 = const()[name = tensor<string, []>("out_1_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_1_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_1_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 64, 35, 476])];
            tensor<fp32, [1, 64, 35, 476]> out_1_has_output_shape = conv_transpose(bias = decoder_0_1_convtrs_0_bias, dilations = out_1_dilations_0, groups = out_1_groups_0, output_shape = out_1_has_output_shape_output_shape_0, pad = out_1_pad_0, pad_type = out_1_pad_type_0, strides = out_1_strides_0, weight = decoder_0_1_convtrs_0_weight, x = var_1864)[name = tensor<string, []>("out_1_has_output_shape")];
            tensor<int32, [4]> var_1884_begin_0 = const()[name = tensor<string, []>("op_1884_begin_0"), val = tensor<int32, [4]>([0, 0, 1, 0])];
            tensor<int32, [4]> var_1884_end_0 = const()[name = tensor<string, []>("op_1884_end_0"), val = tensor<int32, [4]>([1, 64, 34, 476])];
            tensor<bool, [4]> var_1884_end_mask_0 = const()[name = tensor<string, []>("op_1884_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 64, 33, 476]> var_1884 = slice_by_index(begin = var_1884_begin_0, end = var_1884_end_0, end_mask = var_1884_end_mask_0, x = out_1_has_output_shape)[name = tensor<string, []>("op_1884")];
            tensor<int32, [4]> var_1888_begin_0 = const()[name = tensor<string, []>("op_1888_begin_0"), val = tensor<int32, [4]>([0, 0, 33, 0])];
            tensor<int32, [4]> var_1888_end_0 = const()[name = tensor<string, []>("op_1888_end_0"), val = tensor<int32, [4]>([1, 128, 52, 476])];
            tensor<bool, [4]> var_1888_end_mask_0 = const()[name = tensor<string, []>("op_1888_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 128, 19, 476]> var_1888 = slice_by_index(begin = var_1888_begin_0, end = var_1888_end_0, end_mask = var_1888_end_mask_0, x = x_167)[name = tensor<string, []>("op_1888")];
            tensor<string, []> out_3_pad_type_0 = const()[name = tensor<string, []>("out_3_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> out_3_strides_0 = const()[name = tensor<string, []>("out_3_strides_0"), val = tensor<int32, [2]>([4, 1])];
            tensor<int32, [4]> out_3_pad_0 = const()[name = tensor<string, []>("out_3_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_3_dilations_0 = const()[name = tensor<string, []>("out_3_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_3_groups_0 = const()[name = tensor<string, []>("out_3_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_3_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_3_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 64, 76, 476])];
            tensor<fp32, [1, 64, 76, 476]> out_3_has_output_shape = conv_transpose(bias = decoder_0_1_convtrs_1_bias, dilations = out_3_dilations_0, groups = out_3_groups_0, output_shape = out_3_has_output_shape_output_shape_0, pad = out_3_pad_0, pad_type = out_3_pad_type_0, strides = out_3_strides_0, weight = decoder_0_1_convtrs_1_weight, x = var_1888)[name = tensor<string, []>("out_3_has_output_shape")];
            tensor<int32, [4]> var_1908_begin_0 = const()[name = tensor<string, []>("op_1908_begin_0"), val = tensor<int32, [4]>([0, 0, 1, 0])];
            tensor<int32, [4]> var_1908_end_0 = const()[name = tensor<string, []>("op_1908_end_0"), val = tensor<int32, [4]>([1, 64, 74, 476])];
            tensor<bool, [4]> var_1908_end_mask_0 = const()[name = tensor<string, []>("op_1908_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 64, 73, 476]> var_1908 = slice_by_index(begin = var_1908_begin_0, end = var_1908_end_0, end_mask = var_1908_end_mask_0, x = out_3_has_output_shape)[name = tensor<string, []>("op_1908")];
            tensor<int32, [4]> var_1912_begin_0 = const()[name = tensor<string, []>("op_1912_begin_0"), val = tensor<int32, [4]>([0, 0, 52, 0])];
            tensor<int32, [4]> var_1912_end_0 = const()[name = tensor<string, []>("op_1912_end_0"), val = tensor<int32, [4]>([1, 128, 57, 476])];
            tensor<bool, [4]> var_1912_end_mask_0 = const()[name = tensor<string, []>("op_1912_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 128, 5, 476]> var_1912 = slice_by_index(begin = var_1912_begin_0, end = var_1912_end_0, end_mask = var_1912_end_mask_0, x = x_167)[name = tensor<string, []>("op_1912")];
            tensor<string, []> out_5_pad_type_0 = const()[name = tensor<string, []>("out_5_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> out_5_strides_0 = const()[name = tensor<string, []>("out_5_strides_0"), val = tensor<int32, [2]>([16, 1])];
            tensor<int32, [4]> out_5_pad_0 = const()[name = tensor<string, []>("out_5_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_5_dilations_0 = const()[name = tensor<string, []>("out_5_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_5_groups_0 = const()[name = tensor<string, []>("out_5_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_5_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_5_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 64, 80, 476])];
            tensor<fp32, [1, 64, 80, 476]> out_5_has_output_shape = conv_transpose(bias = decoder_0_1_convtrs_2_bias, dilations = out_5_dilations_0, groups = out_5_groups_0, output_shape = out_5_has_output_shape_output_shape_0, pad = out_5_pad_0, pad_type = out_5_pad_type_0, strides = out_5_strides_0, weight = decoder_0_1_convtrs_2_weight, x = var_1912)[name = tensor<string, []>("out_5_has_output_shape")];
            tensor<bool, []> x_169_interleave_0 = const()[name = tensor<string, []>("x_169_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 64, 186, 476]> x_169 = concat(axis = var_1848, interleave = x_169_interleave_0, values = (var_1884, var_1908, out_5_has_output_shape))[name = tensor<string, []>("x_169")];
            tensor<fp32, [1, 64, 186, 476]> x_171 = add(x = x_169, y = transpose_26)[name = tensor<string, []>("x_171")];
            tensor<int32, [4]> var_1942 = const()[name = tensor<string, []>("op_1942"), val = tensor<int32, [4]>([1, 2, 1, 1])];
            tensor<fp32, [1, 128, 186, 476]> input_311 = tile(reps = var_1942, x = x_171)[name = tensor<string, []>("input_311")];
            tensor<string, []> input_313_pad_type_0 = const()[name = tensor<string, []>("input_313_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [4]> input_313_pad_0 = const()[name = tensor<string, []>("input_313_pad_0"), val = tensor<int32, [4]>([1, 1, 1, 1])];
            tensor<int32, [2]> input_313_strides_0 = const()[name = tensor<string, []>("input_313_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> input_313_dilations_0 = const()[name = tensor<string, []>("input_313_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_313_groups_0 = const()[name = tensor<string, []>("input_313_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 128, 186, 476]> input_313 = conv(bias = decoder_1_0_conv_bias, dilations = input_313_dilations_0, groups = input_313_groups_0, pad = input_313_pad_0, pad_type = input_313_pad_type_0, strides = input_313_strides_0, weight = decoder_1_0_conv_weight, x = input_311)[name = tensor<string, []>("input_313")];
            tensor<int32, []> x_173_split_num_splits_0 = const()[name = tensor<string, []>("x_173_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> x_173_split_axis_0 = const()[name = tensor<string, []>("x_173_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 64, 186, 476]> x_173_split_0, tensor<fp32, [1, 64, 186, 476]> x_173_split_1 = split(axis = x_173_split_axis_0, num_splits = x_173_split_num_splits_0, x = input_313)[name = tensor<string, []>("x_173_split")];
            tensor<fp32, [1, 64, 186, 476]> x_173_split_1_sigmoid = sigmoid(x = x_173_split_1)[name = tensor<string, []>("x_173_split_1_sigmoid")];
            tensor<fp32, [1, 64, 186, 476]> x_173 = mul(x = x_173_split_0, y = x_173_split_1_sigmoid)[name = tensor<string, []>("x_173")];
            tensor<int32, []> var_1960 = const()[name = tensor<string, []>("op_1960"), val = tensor<int32, []>(2)];
            tensor<int32, [4]> var_1976_begin_0 = const()[name = tensor<string, []>("op_1976_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_1976_end_0 = const()[name = tensor<string, []>("op_1976_end_0"), val = tensor<int32, [4]>([1, 64, 108, 476])];
            tensor<bool, [4]> var_1976_end_mask_0 = const()[name = tensor<string, []>("op_1976_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 64, 108, 476]> var_1976 = slice_by_index(begin = var_1976_begin_0, end = var_1976_end_0, end_mask = var_1976_end_mask_0, x = x_173)[name = tensor<string, []>("op_1976")];
            tensor<string, []> out_7_pad_type_0 = const()[name = tensor<string, []>("out_7_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [4]> out_7_pad_0 = const()[name = tensor<string, []>("out_7_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_7_strides_0 = const()[name = tensor<string, []>("out_7_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> out_7_dilations_0 = const()[name = tensor<string, []>("out_7_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_7_groups_0 = const()[name = tensor<string, []>("out_7_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_7_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_7_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 32, 110, 476])];
            tensor<fp32, [1, 32, 110, 476]> out_7_has_output_shape = conv_transpose(bias = decoder_1_1_convtrs_0_bias, dilations = out_7_dilations_0, groups = out_7_groups_0, output_shape = out_7_has_output_shape_output_shape_0, pad = out_7_pad_0, pad_type = out_7_pad_type_0, strides = out_7_strides_0, weight = decoder_1_1_convtrs_0_weight, x = var_1976)[name = tensor<string, []>("out_7_has_output_shape")];
            tensor<int32, [4]> var_1996_begin_0 = const()[name = tensor<string, []>("op_1996_begin_0"), val = tensor<int32, [4]>([0, 0, 1, 0])];
            tensor<int32, [4]> var_1996_end_0 = const()[name = tensor<string, []>("op_1996_end_0"), val = tensor<int32, [4]>([1, 32, 109, 476])];
            tensor<bool, [4]> var_1996_end_mask_0 = const()[name = tensor<string, []>("op_1996_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 32, 108, 476]> var_1996 = slice_by_index(begin = var_1996_begin_0, end = var_1996_end_0, end_mask = var_1996_end_mask_0, x = out_7_has_output_shape)[name = tensor<string, []>("op_1996")];
            tensor<int32, [4]> var_2000_begin_0 = const()[name = tensor<string, []>("op_2000_begin_0"), val = tensor<int32, [4]>([0, 0, 108, 0])];
            tensor<int32, [4]> var_2000_end_0 = const()[name = tensor<string, []>("op_2000_end_0"), val = tensor<int32, [4]>([1, 64, 169, 476])];
            tensor<bool, [4]> var_2000_end_mask_0 = const()[name = tensor<string, []>("op_2000_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 64, 61, 476]> var_2000 = slice_by_index(begin = var_2000_begin_0, end = var_2000_end_0, end_mask = var_2000_end_mask_0, x = x_173)[name = tensor<string, []>("op_2000")];
            tensor<string, []> out_9_pad_type_0 = const()[name = tensor<string, []>("out_9_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> out_9_strides_0 = const()[name = tensor<string, []>("out_9_strides_0"), val = tensor<int32, [2]>([4, 1])];
            tensor<int32, [4]> out_9_pad_0 = const()[name = tensor<string, []>("out_9_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_9_dilations_0 = const()[name = tensor<string, []>("out_9_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_9_groups_0 = const()[name = tensor<string, []>("out_9_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_9_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_9_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 32, 244, 476])];
            tensor<fp32, [1, 32, 244, 476]> out_9_has_output_shape = conv_transpose(bias = decoder_1_1_convtrs_1_bias, dilations = out_9_dilations_0, groups = out_9_groups_0, output_shape = out_9_has_output_shape_output_shape_0, pad = out_9_pad_0, pad_type = out_9_pad_type_0, strides = out_9_strides_0, weight = decoder_1_1_convtrs_1_weight, x = var_2000)[name = tensor<string, []>("out_9_has_output_shape")];
            tensor<int32, [4]> var_2020_begin_0 = const()[name = tensor<string, []>("op_2020_begin_0"), val = tensor<int32, [4]>([0, 0, 1, 0])];
            tensor<int32, [4]> var_2020_end_0 = const()[name = tensor<string, []>("op_2020_end_0"), val = tensor<int32, [4]>([1, 32, 243, 476])];
            tensor<bool, [4]> var_2020_end_mask_0 = const()[name = tensor<string, []>("op_2020_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 32, 242, 476]> var_2020 = slice_by_index(begin = var_2020_begin_0, end = var_2020_end_0, end_mask = var_2020_end_mask_0, x = out_9_has_output_shape)[name = tensor<string, []>("op_2020")];
            tensor<int32, [4]> var_2024_begin_0 = const()[name = tensor<string, []>("op_2024_begin_0"), val = tensor<int32, [4]>([0, 0, 169, 0])];
            tensor<int32, [4]> var_2024_end_0 = const()[name = tensor<string, []>("op_2024_end_0"), val = tensor<int32, [4]>([1, 64, 186, 476])];
            tensor<bool, [4]> var_2024_end_mask_0 = const()[name = tensor<string, []>("op_2024_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 64, 17, 476]> var_2024 = slice_by_index(begin = var_2024_begin_0, end = var_2024_end_0, end_mask = var_2024_end_mask_0, x = x_173)[name = tensor<string, []>("op_2024")];
            tensor<string, []> out_11_pad_type_0 = const()[name = tensor<string, []>("out_11_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> out_11_strides_0 = const()[name = tensor<string, []>("out_11_strides_0"), val = tensor<int32, [2]>([16, 1])];
            tensor<int32, [4]> out_11_pad_0 = const()[name = tensor<string, []>("out_11_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_11_dilations_0 = const()[name = tensor<string, []>("out_11_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_11_groups_0 = const()[name = tensor<string, []>("out_11_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_11_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_11_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 32, 272, 476])];
            tensor<fp32, [1, 32, 272, 476]> out_11_has_output_shape = conv_transpose(bias = decoder_1_1_convtrs_2_bias, dilations = out_11_dilations_0, groups = out_11_groups_0, output_shape = out_11_has_output_shape_output_shape_0, pad = out_11_pad_0, pad_type = out_11_pad_type_0, strides = out_11_strides_0, weight = decoder_1_1_convtrs_2_weight, x = var_2024)[name = tensor<string, []>("out_11_has_output_shape")];
            tensor<int32, [4]> var_2043_begin_0 = const()[name = tensor<string, []>("op_2043_begin_0"), val = tensor<int32, [4]>([0, 0, 3, 0])];
            tensor<int32, [4]> var_2043_end_0 = const()[name = tensor<string, []>("op_2043_end_0"), val = tensor<int32, [4]>([1, 32, 269, 476])];
            tensor<bool, [4]> var_2043_end_mask_0 = const()[name = tensor<string, []>("op_2043_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 32, 266, 476]> var_2043 = slice_by_index(begin = var_2043_begin_0, end = var_2043_end_0, end_mask = var_2043_end_mask_0, x = out_11_has_output_shape)[name = tensor<string, []>("op_2043")];
            tensor<bool, []> x_175_interleave_0 = const()[name = tensor<string, []>("x_175_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 32, 616, 476]> x_175 = concat(axis = var_1960, interleave = x_175_interleave_0, values = (var_1996, var_2020, var_2043))[name = tensor<string, []>("x_175")];
            tensor<fp32, [1, 32, 616, 476]> x_177 = add(x = x_175, y = transpose_24)[name = tensor<string, []>("x_177")];
            tensor<int32, [4]> var_2054 = const()[name = tensor<string, []>("op_2054"), val = tensor<int32, [4]>([1, 2, 1, 1])];
            tensor<fp32, [1, 64, 616, 476]> input_321 = tile(reps = var_2054, x = x_177)[name = tensor<string, []>("input_321")];
            tensor<string, []> input_323_pad_type_0 = const()[name = tensor<string, []>("input_323_pad_type_0"), val = tensor<string, []>("custom")];
            tensor<int32, [4]> input_323_pad_0 = const()[name = tensor<string, []>("input_323_pad_0"), val = tensor<int32, [4]>([1, 1, 1, 1])];
            tensor<int32, [2]> input_323_strides_0 = const()[name = tensor<string, []>("input_323_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> input_323_dilations_0 = const()[name = tensor<string, []>("input_323_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> input_323_groups_0 = const()[name = tensor<string, []>("input_323_groups_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 64, 616, 476]> input_323 = conv(bias = decoder_2_0_conv_bias, dilations = input_323_dilations_0, groups = input_323_groups_0, pad = input_323_pad_0, pad_type = input_323_pad_type_0, strides = input_323_strides_0, weight = decoder_2_0_conv_weight, x = input_321)[name = tensor<string, []>("input_323")];
            tensor<int32, []> x_split_num_splits_0 = const()[name = tensor<string, []>("x_split_num_splits_0"), val = tensor<int32, []>(2)];
            tensor<int32, []> x_split_axis_0 = const()[name = tensor<string, []>("x_split_axis_0"), val = tensor<int32, []>(1)];
            tensor<fp32, [1, 32, 616, 476]> x_split_0, tensor<fp32, [1, 32, 616, 476]> x_split_1 = split(axis = x_split_axis_0, num_splits = x_split_num_splits_0, x = input_323)[name = tensor<string, []>("x_split")];
            tensor<fp32, [1, 32, 616, 476]> x_split_1_sigmoid = sigmoid(x = x_split_1)[name = tensor<string, []>("x_split_1_sigmoid")];
            tensor<fp32, [1, 32, 616, 476]> x = mul(x = x_split_0, y = x_split_1_sigmoid)[name = tensor<string, []>("x")];
            tensor<int32, []> var_2072 = const()[name = tensor<string, []>("op_2072"), val = tensor<int32, []>(2)];
            tensor<int32, [4]> var_2088_begin_0 = const()[name = tensor<string, []>("op_2088_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_2088_end_0 = const()[name = tensor<string, []>("op_2088_end_0"), val = tensor<int32, [4]>([1, 32, 359, 476])];
            tensor<bool, [4]> var_2088_end_mask_0 = const()[name = tensor<string, []>("op_2088_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 32, 359, 476]> var_2088 = slice_by_index(begin = var_2088_begin_0, end = var_2088_end_0, end_mask = var_2088_end_mask_0, x = x)[name = tensor<string, []>("op_2088")];
            tensor<string, []> out_13_pad_type_0 = const()[name = tensor<string, []>("out_13_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [4]> out_13_pad_0 = const()[name = tensor<string, []>("out_13_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_13_strides_0 = const()[name = tensor<string, []>("out_13_strides_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, [2]> out_13_dilations_0 = const()[name = tensor<string, []>("out_13_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_13_groups_0 = const()[name = tensor<string, []>("out_13_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_13_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_13_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 16, 361, 476])];
            tensor<fp32, [1, 16, 361, 476]> out_13_has_output_shape = conv_transpose(bias = decoder_2_1_convtrs_0_bias, dilations = out_13_dilations_0, groups = out_13_groups_0, output_shape = out_13_has_output_shape_output_shape_0, pad = out_13_pad_0, pad_type = out_13_pad_type_0, strides = out_13_strides_0, weight = decoder_2_1_convtrs_0_weight, x = var_2088)[name = tensor<string, []>("out_13_has_output_shape")];
            tensor<int32, [4]> var_2108_begin_0 = const()[name = tensor<string, []>("op_2108_begin_0"), val = tensor<int32, [4]>([0, 0, 1, 0])];
            tensor<int32, [4]> var_2108_end_0 = const()[name = tensor<string, []>("op_2108_end_0"), val = tensor<int32, [4]>([1, 16, 360, 476])];
            tensor<bool, [4]> var_2108_end_mask_0 = const()[name = tensor<string, []>("op_2108_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 16, 359, 476]> var_2108 = slice_by_index(begin = var_2108_begin_0, end = var_2108_end_0, end_mask = var_2108_end_mask_0, x = out_13_has_output_shape)[name = tensor<string, []>("op_2108")];
            tensor<int32, [4]> var_2112_begin_0 = const()[name = tensor<string, []>("op_2112_begin_0"), val = tensor<int32, [4]>([0, 0, 359, 0])];
            tensor<int32, [4]> var_2112_end_0 = const()[name = tensor<string, []>("op_2112_end_0"), val = tensor<int32, [4]>([1, 32, 560, 476])];
            tensor<bool, [4]> var_2112_end_mask_0 = const()[name = tensor<string, []>("op_2112_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 32, 201, 476]> var_2112 = slice_by_index(begin = var_2112_begin_0, end = var_2112_end_0, end_mask = var_2112_end_mask_0, x = x)[name = tensor<string, []>("op_2112")];
            tensor<string, []> out_15_pad_type_0 = const()[name = tensor<string, []>("out_15_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> out_15_strides_0 = const()[name = tensor<string, []>("out_15_strides_0"), val = tensor<int32, [2]>([4, 1])];
            tensor<int32, [4]> out_15_pad_0 = const()[name = tensor<string, []>("out_15_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_15_dilations_0 = const()[name = tensor<string, []>("out_15_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_15_groups_0 = const()[name = tensor<string, []>("out_15_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_15_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_15_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 16, 804, 476])];
            tensor<fp32, [1, 16, 804, 476]> out_15_has_output_shape = conv_transpose(bias = decoder_2_1_convtrs_1_bias, dilations = out_15_dilations_0, groups = out_15_groups_0, output_shape = out_15_has_output_shape_output_shape_0, pad = out_15_pad_0, pad_type = out_15_pad_type_0, strides = out_15_strides_0, weight = decoder_2_1_convtrs_1_weight, x = var_2112)[name = tensor<string, []>("out_15_has_output_shape")];
            tensor<int32, [4]> var_2132_begin_0 = const()[name = tensor<string, []>("op_2132_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [4]> var_2132_end_0 = const()[name = tensor<string, []>("op_2132_end_0"), val = tensor<int32, [4]>([1, 16, 803, 476])];
            tensor<bool, [4]> var_2132_end_mask_0 = const()[name = tensor<string, []>("op_2132_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 16, 803, 476]> var_2132 = slice_by_index(begin = var_2132_begin_0, end = var_2132_end_0, end_mask = var_2132_end_mask_0, x = out_15_has_output_shape)[name = tensor<string, []>("op_2132")];
            tensor<int32, [4]> var_2136_begin_0 = const()[name = tensor<string, []>("op_2136_begin_0"), val = tensor<int32, [4]>([0, 0, 560, 0])];
            tensor<int32, [4]> var_2136_end_0 = const()[name = tensor<string, []>("op_2136_end_0"), val = tensor<int32, [4]>([1, 32, 616, 476])];
            tensor<bool, [4]> var_2136_end_mask_0 = const()[name = tensor<string, []>("op_2136_end_mask_0"), val = tensor<bool, [4]>([true, true, true, true])];
            tensor<fp32, [1, 32, 56, 476]> var_2136 = slice_by_index(begin = var_2136_begin_0, end = var_2136_end_0, end_mask = var_2136_end_mask_0, x = x)[name = tensor<string, []>("op_2136")];
            tensor<string, []> out_pad_type_0 = const()[name = tensor<string, []>("out_pad_type_0"), val = tensor<string, []>("valid")];
            tensor<int32, [2]> out_strides_0 = const()[name = tensor<string, []>("out_strides_0"), val = tensor<int32, [2]>([16, 1])];
            tensor<int32, [4]> out_pad_0 = const()[name = tensor<string, []>("out_pad_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
            tensor<int32, [2]> out_dilations_0 = const()[name = tensor<string, []>("out_dilations_0"), val = tensor<int32, [2]>([1, 1])];
            tensor<int32, []> out_groups_0 = const()[name = tensor<string, []>("out_groups_0"), val = tensor<int32, []>(1)];
            tensor<int32, [4]> out_has_output_shape_output_shape_0 = const()[name = tensor<string, []>("out_has_output_shape_output_shape_0"), val = tensor<int32, [4]>([1, 16, 896, 476])];
            tensor<fp32, [1, 16, 896, 476]> out_has_output_shape = conv_transpose(bias = decoder_2_1_convtrs_2_bias, dilations = out_dilations_0, groups = out_groups_0, output_shape = out_has_output_shape_output_shape_0, pad = out_pad_0, pad_type = out_pad_type_0, strides = out_strides_0, weight = decoder_2_1_convtrs_2_weight, x = var_2136)[name = tensor<string, []>("out_has_output_shape")];
            tensor<int32, [4]> var_2155_begin_0 = const()[name = tensor<string, []>("op_2155_begin_0"), val = tensor<int32, [4]>([0, 0, 4, 0])];
            tensor<int32, [4]> var_2155_end_0 = const()[name = tensor<string, []>("op_2155_end_0"), val = tensor<int32, [4]>([1, 16, 891, 476])];
            tensor<bool, [4]> var_2155_end_mask_0 = const()[name = tensor<string, []>("op_2155_end_mask_0"), val = tensor<bool, [4]>([true, true, false, true])];
            tensor<fp32, [1, 16, 887, 476]> var_2155 = slice_by_index(begin = var_2155_begin_0, end = var_2155_end_0, end_mask = var_2155_end_mask_0, x = out_has_output_shape)[name = tensor<string, []>("op_2155")];
            tensor<bool, []> var_2158_interleave_0 = const()[name = tensor<string, []>("op_2158_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp32, [1, 16, 2049, 476]> separated = concat(axis = var_2072, interleave = var_2158_interleave_0, values = (var_2108, var_2132, var_2155))[name = tensor<string, []>("op_2158")];
        } -> (separated);
}