program(1.0)
[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3500.12.1"}, {"coremlc-version", "3500.27.1"}, {"coremltools-component-torch", "2.5.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "8.2"}, {"mldb_token", "mldb-8mbymokg7x"}})]
{
    func main<ios15>(tensor<fp32, [?, 27]> input_feature_vector) [FlexibleShapeInformation = tuple<tuple<tensor<string, []>, dict<tensor<string, []>, tensor<int32, [?]>>>, tuple<tensor<string, []>, dict<tensor<string, []>, list<tensor<int32, [2]>, ?>>>>((("DefaultShapes", {{"input_feature_vector", [1, 27]}}), ("RangeDims", {{"input_feature_vector", [[1, 512], [27, 27]]}})))] {
            tensor<int32, [2]> count_confirms_begin_0 = const()[name = tensor<string, []>("count_confirms_begin_0"), val = tensor<int32, [2]>([0, 23])];
            tensor<int32, [2]> count_confirms_end_0 = const()[name = tensor<string, []>("count_confirms_end_0"), val = tensor<int32, [2]>([0, 24])];
            tensor<bool, [2]> count_confirms_end_mask_0 = const()[name = tensor<string, []>("count_confirms_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<string, []> input_feature_vector_to_fp16_dtype_0 = const()[name = tensor<string, []>("input_feature_vector_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
            tensor<fp16, [?, 27]> input_feature_vector_to_fp16 = cast(dtype = input_feature_vector_to_fp16_dtype_0, x = input_feature_vector)[name = tensor<string, []>("cast_14")];
            tensor<fp16, [?, 1]> count_confirms_cast_fp16 = slice_by_index(begin = count_confirms_begin_0, end = count_confirms_end_0, end_mask = count_confirms_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("count_confirms_cast_fp16")];
            tensor<fp16, []> var_12_promoted_to_fp16 = const()[name = tensor<string, []>("op_12_promoted_to_fp16"), val = tensor<fp16, []>(0x1.8p+1)];
            tensor<bool, [?, 1]> var_13_cast_fp16 = greater_equal(x = count_confirms_cast_fp16, y = var_12_promoted_to_fp16)[name = tensor<string, []>("op_13_cast_fp16")];
            tensor<int32, [2]> shape_0_cast_fp16 = shape(x = count_confirms_cast_fp16)[name = tensor<string, []>("shape_0_cast_fp16")];
            tensor<int32, []> var_20_value_0 = const()[name = tensor<string, []>("op_20_value_0"), val = tensor<int32, []>(5)];
            tensor<int32, [?, 1]> var_20 = fill(shape = shape_0_cast_fp16, value = var_20_value_0)[name = tensor<string, []>("op_20")];
            tensor<fp16, []> var_27_value_0_to_fp16 = const()[name = tensor<string, []>("op_27_value_0_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
            tensor<fp16, [?, 1]> var_27_cast_fp16 = fill(shape = shape_0_cast_fp16, value = var_27_value_0_to_fp16)[name = tensor<string, []>("op_27_cast_fp16")];
            tensor<string, []> var_20_promoted_to_fp16_dtype_0 = const()[name = tensor<string, []>("op_20_promoted_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
            tensor<fp16, [?, 1]> var_20_to_fp16 = cast(dtype = var_20_promoted_to_fp16_dtype_0, x = var_20)[name = tensor<string, []>("cast_13")];
            tensor<fp16, [?, 1]> sub_score_confirms_cast_fp16 = select(a = var_20_to_fp16, b = var_27_cast_fp16, cond = var_13_cast_fp16)[name = tensor<string, []>("sub_score_confirms_cast_fp16")];
            tensor<string, []> sub_score_confirms_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("sub_score_confirms_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<int32, [2]> count_rejects_begin_0 = const()[name = tensor<string, []>("count_rejects_begin_0"), val = tensor<int32, [2]>([0, 25])];
            tensor<int32, [2]> count_rejects_end_0 = const()[name = tensor<string, []>("count_rejects_end_0"), val = tensor<int32, [2]>([0, 26])];
            tensor<bool, [2]> count_rejects_end_mask_0 = const()[name = tensor<string, []>("count_rejects_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> count_rejects_cast_fp16 = slice_by_index(begin = count_rejects_begin_0, end = count_rejects_end_0, end_mask = count_rejects_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("count_rejects_cast_fp16")];
            tensor<fp16, []> var_39_promoted_to_fp16 = const()[name = tensor<string, []>("op_39_promoted_to_fp16"), val = tensor<fp16, []>(0x1.8p+1)];
            tensor<bool, [?, 1]> var_40_cast_fp16 = greater_equal(x = count_rejects_cast_fp16, y = var_39_promoted_to_fp16)[name = tensor<string, []>("op_40_cast_fp16")];
            tensor<fp16, []> var_41_promoted_to_fp16 = const()[name = tensor<string, []>("op_41_promoted_to_fp16"), val = tensor<fp16, []>(0x1p+0)];
            tensor<bool, [?, 1]> var_42_cast_fp16 = less(x = count_confirms_cast_fp16, y = var_41_promoted_to_fp16)[name = tensor<string, []>("op_42_cast_fp16")];
            tensor<bool, [?, 1]> var_43 = logical_and(x = var_40_cast_fp16, y = var_42_cast_fp16)[name = tensor<string, []>("op_43")];
            tensor<int32, [2]> shape_2_cast_fp16 = shape(x = count_rejects_cast_fp16)[name = tensor<string, []>("shape_2_cast_fp16")];
            tensor<int32, []> var_50_value_0 = const()[name = tensor<string, []>("op_50_value_0"), val = tensor<int32, []>(-10)];
            tensor<int32, [?, 1]> var_50 = fill(shape = shape_2_cast_fp16, value = var_50_value_0)[name = tensor<string, []>("op_50")];
            tensor<fp16, []> var_57_value_0_to_fp16 = const()[name = tensor<string, []>("op_57_value_0_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
            tensor<fp16, [?, 1]> var_57_cast_fp16 = fill(shape = shape_2_cast_fp16, value = var_57_value_0_to_fp16)[name = tensor<string, []>("op_57_cast_fp16")];
            tensor<string, []> var_50_promoted_to_fp16_dtype_0 = const()[name = tensor<string, []>("op_50_promoted_to_fp16_dtype_0"), val = tensor<string, []>("fp16")];
            tensor<fp16, [?, 1]> var_50_to_fp16 = cast(dtype = var_50_promoted_to_fp16_dtype_0, x = var_50)[name = tensor<string, []>("cast_11")];
            tensor<fp16, [?, 1]> sub_score_rejects_cast_fp16 = select(a = var_50_to_fp16, b = var_57_cast_fp16, cond = var_43)[name = tensor<string, []>("sub_score_rejects_cast_fp16")];
            tensor<string, []> sub_score_rejects_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("sub_score_rejects_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<int32, [2]> last_open_age_begin_0 = const()[name = tensor<string, []>("last_open_age_begin_0"), val = tensor<int32, [2]>([0, 15])];
            tensor<int32, [2]> last_open_age_end_0 = const()[name = tensor<string, []>("last_open_age_end_0"), val = tensor<int32, [2]>([0, 16])];
            tensor<bool, [2]> last_open_age_end_mask_0 = const()[name = tensor<string, []>("last_open_age_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> last_open_age_cast_fp16 = slice_by_index(begin = last_open_age_begin_0, end = last_open_age_end_0, end_mask = last_open_age_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("last_open_age_cast_fp16")];
            tensor<fp16, []> var_69_promoted_to_fp16 = const()[name = tensor<string, []>("op_69_promoted_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
            tensor<bool, [?, 1]> var_70_cast_fp16 = greater(x = last_open_age_cast_fp16, y = var_69_promoted_to_fp16)[name = tensor<string, []>("op_70_cast_fp16")];
            tensor<fp16, []> var_71_promoted_to_fp16 = const()[name = tensor<string, []>("op_71_promoted_to_fp16"), val = tensor<fp16, []>(-0x1p+0)];
            tensor<fp16, [?, 1]> var_72_cast_fp16 = mul(x = last_open_age_cast_fp16, y = var_71_promoted_to_fp16)[name = tensor<string, []>("op_72_cast_fp16")];
            tensor<fp16, []> _inversed_74_y_0_to_fp16 = const()[name = tensor<string, []>("_inversed_74_y_0_to_fp16"), val = tensor<fp16, []>(0x1.04p-18)];
            tensor<fp16, [?, 1]> _inversed_74_cast_fp16 = mul(x = var_72_cast_fp16, y = _inversed_74_y_0_to_fp16)[name = tensor<string, []>("_inversed_74_cast_fp16")];
            tensor<fp16, []> var_75_promoted_to_fp16 = const()[name = tensor<string, []>("op_75_promoted_to_fp16"), val = tensor<fp16, []>(0x1p+1)];
            tensor<fp16, [?, 1]> var_76_cast_fp16 = pow(x = var_75_promoted_to_fp16, y = _inversed_74_cast_fp16)[name = tensor<string, []>("op_76_cast_fp16")];
            tensor<fp16, []> var_77_promoted_to_fp16 = const()[name = tensor<string, []>("op_77_promoted_to_fp16"), val = tensor<fp16, []>(0x1.4p+3)];
            tensor<fp16, [?, 1]> var_78_cast_fp16 = mul(x = var_76_cast_fp16, y = var_77_promoted_to_fp16)[name = tensor<string, []>("op_78_cast_fp16")];
            tensor<fp16, []> const_0_promoted_to_fp16 = const()[name = tensor<string, []>("const_0_promoted_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
            tensor<fp16, [?, 1]> sub_score_open_age_cast_fp16 = select(a = var_78_cast_fp16, b = const_0_promoted_to_fp16, cond = var_70_cast_fp16)[name = tensor<string, []>("sub_score_open_age_cast_fp16")];
            tensor<string, []> sub_score_open_age_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("sub_score_open_age_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<int32, [2]> ssid_popularity_begin_0 = const()[name = tensor<string, []>("ssid_popularity_begin_0"), val = tensor<int32, [2]>([0, 0])];
            tensor<int32, [2]> ssid_popularity_end_0 = const()[name = tensor<string, []>("ssid_popularity_end_0"), val = tensor<int32, [2]>([0, 1])];
            tensor<bool, [2]> ssid_popularity_end_mask_0 = const()[name = tensor<string, []>("ssid_popularity_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> ssid_popularity_cast_fp16 = slice_by_index(begin = ssid_popularity_begin_0, end = ssid_popularity_end_0, end_mask = ssid_popularity_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("ssid_popularity_cast_fp16")];
            tensor<fp16, []> var_98_to_fp16 = const()[name = tensor<string, []>("op_98_to_fp16"), val = tensor<fp16, []>(0x1.2cp+7)];
            tensor<fp16, [?, 1]> sub_score_ssid_popularity_cast_fp16 = mul(x = ssid_popularity_cast_fp16, y = var_98_to_fp16)[name = tensor<string, []>("sub_score_ssid_popularity_cast_fp16")];
            tensor<string, []> sub_score_ssid_popularity_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("sub_score_ssid_popularity_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<int32, [2]> popularity_pow_begin_0 = const()[name = tensor<string, []>("popularity_pow_begin_0"), val = tensor<int32, [2]>([0, 6])];
            tensor<int32, [2]> popularity_pow_end_0 = const()[name = tensor<string, []>("popularity_pow_end_0"), val = tensor<int32, [2]>([0, 7])];
            tensor<bool, [2]> popularity_pow_end_mask_0 = const()[name = tensor<string, []>("popularity_pow_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> popularity_pow_cast_fp16 = slice_by_index(begin = popularity_pow_begin_0, end = popularity_pow_end_0, end_mask = popularity_pow_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("popularity_pow_cast_fp16")];
            tensor<int32, [2]> popularity_dow_begin_0 = const()[name = tensor<string, []>("popularity_dow_begin_0"), val = tensor<int32, [2]>([0, 3])];
            tensor<int32, [2]> popularity_dow_end_0 = const()[name = tensor<string, []>("popularity_dow_end_0"), val = tensor<int32, [2]>([0, 4])];
            tensor<bool, [2]> popularity_dow_end_mask_0 = const()[name = tensor<string, []>("popularity_dow_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> popularity_dow_cast_fp16 = slice_by_index(begin = popularity_dow_begin_0, end = popularity_dow_end_0, end_mask = popularity_dow_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("popularity_dow_cast_fp16")];
            tensor<int32, [2]> popularity_tod_begin_0 = const()[name = tensor<string, []>("popularity_tod_begin_0"), val = tensor<int32, [2]>([0, 4])];
            tensor<int32, [2]> popularity_tod_end_0 = const()[name = tensor<string, []>("popularity_tod_end_0"), val = tensor<int32, [2]>([0, 5])];
            tensor<bool, [2]> popularity_tod_end_mask_0 = const()[name = tensor<string, []>("popularity_tod_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> popularity_tod_cast_fp16 = slice_by_index(begin = popularity_tod_begin_0, end = popularity_tod_end_0, end_mask = popularity_tod_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("popularity_tod_cast_fp16")];
            tensor<fp16, []> var_130_to_fp16 = const()[name = tensor<string, []>("op_130_to_fp16"), val = tensor<fp16, []>(0x1.998p-2)];
            tensor<fp16, [?, 1]> var_131_cast_fp16 = mul(x = popularity_pow_cast_fp16, y = var_130_to_fp16)[name = tensor<string, []>("op_131_cast_fp16")];
            tensor<fp16, []> var_132_to_fp16 = const()[name = tensor<string, []>("op_132_to_fp16"), val = tensor<fp16, []>(0x1.668p-2)];
            tensor<fp16, [?, 1]> var_133_cast_fp16 = mul(x = popularity_dow_cast_fp16, y = var_132_to_fp16)[name = tensor<string, []>("op_133_cast_fp16")];
            tensor<fp16, [?, 1]> var_135_cast_fp16 = add(x = var_131_cast_fp16, y = var_133_cast_fp16)[name = tensor<string, []>("op_135_cast_fp16")];
            tensor<fp16, []> var_136_to_fp16 = const()[name = tensor<string, []>("op_136_to_fp16"), val = tensor<fp16, []>(0x1p-2)];
            tensor<fp16, [?, 1]> var_137_cast_fp16 = mul(x = popularity_tod_cast_fp16, y = var_136_to_fp16)[name = tensor<string, []>("op_137_cast_fp16")];
            tensor<fp16, [?, 1]> var_139_cast_fp16 = add(x = var_135_cast_fp16, y = var_137_cast_fp16)[name = tensor<string, []>("op_139_cast_fp16")];
            tensor<fp16, []> var_140_promoted_to_fp16 = const()[name = tensor<string, []>("op_140_promoted_to_fp16"), val = tensor<fp16, []>(0x1.9p+6)];
            tensor<fp16, [?, 1]> sub_score_document_popularity_cast_fp16 = mul(x = var_139_cast_fp16, y = var_140_promoted_to_fp16)[name = tensor<string, []>("sub_score_document_popularity_cast_fp16")];
            tensor<string, []> sub_score_document_popularity_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("sub_score_document_popularity_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<int32, [2]> app_score_1_begin_0 = const()[name = tensor<string, []>("app_score_1_begin_0"), val = tensor<int32, [2]>([0, 22])];
            tensor<int32, [2]> app_score_1_end_0 = const()[name = tensor<string, []>("app_score_1_end_0"), val = tensor<int32, [2]>([0, 23])];
            tensor<bool, [2]> app_score_1_end_mask_0 = const()[name = tensor<string, []>("app_score_1_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> app_score_1_cast_fp16 = slice_by_index(begin = app_score_1_begin_0, end = app_score_1_end_0, end_mask = app_score_1_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("app_score_1_cast_fp16")];
            tensor<fp16, []> var_152_promoted_to_fp16 = const()[name = tensor<string, []>("op_152_promoted_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
            tensor<bool, [?, 1]> var_153_cast_fp16 = greater_equal(x = app_score_1_cast_fp16, y = var_152_promoted_to_fp16)[name = tensor<string, []>("op_153_cast_fp16")];
            tensor<fp16, []> var_154_promoted_to_fp16 = const()[name = tensor<string, []>("op_154_promoted_to_fp16"), val = tensor<fp16, []>(0x1p+0)];
            tensor<bool, [?, 1]> var_155_cast_fp16 = less_equal(x = app_score_1_cast_fp16, y = var_154_promoted_to_fp16)[name = tensor<string, []>("op_155_cast_fp16")];
            tensor<bool, [?, 1]> AppScorePresumedProbabilistic__pre__output__fp32__cast = logical_and(x = var_153_cast_fp16, y = var_155_cast_fp16)[name = tensor<string, []>("app_score_presumed_probabilistic")];
            tensor<fp16, []> var_157_promoted_to_fp16 = const()[name = tensor<string, []>("op_157_promoted_to_fp16"), val = tensor<fp16, []>(0x1.9p+7)];
            tensor<fp16, [?, 1]> var_158_cast_fp16 = mul(x = app_score_1_cast_fp16, y = var_157_promoted_to_fp16)[name = tensor<string, []>("op_158_cast_fp16")];
            tensor<fp16, [?, 1]> app_score_3_cast_fp16 = select(a = var_158_cast_fp16, b = app_score_1_cast_fp16, cond = AppScorePresumedProbabilistic__pre__output__fp32__cast)[name = tensor<string, []>("app_score_3_cast_fp16")];
            tensor<fp16, []> var_160_promoted_to_fp16 = const()[name = tensor<string, []>("op_160_promoted_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
            tensor<fp16, []> var_161_promoted_to_fp16 = const()[name = tensor<string, []>("op_161_promoted_to_fp16"), val = tensor<fp16, []>(0x1.9p+7)];
            tensor<fp16, [?, 1]> clip_0_cast_fp16 = clip(alpha = var_160_promoted_to_fp16, beta = var_161_promoted_to_fp16, x = app_score_3_cast_fp16)[name = tensor<string, []>("clip_0_cast_fp16")];
            tensor<fp16, []> var_164_promoted_to_fp16 = const()[name = tensor<string, []>("op_164_promoted_to_fp16"), val = tensor<fp16, []>(0x1p+0)];
            tensor<fp16, [?, 1]> var_165_cast_fp16 = add(x = clip_0_cast_fp16, y = var_164_promoted_to_fp16)[name = tensor<string, []>("op_165_cast_fp16")];
            tensor<fp16, []> var_166_epsilon_0_to_fp16 = const()[name = tensor<string, []>("op_166_epsilon_0_to_fp16"), val = tensor<fp16, []>(0x0p+0)];
            tensor<fp16, [?, 1]> var_166_cast_fp16 = log(epsilon = var_166_epsilon_0_to_fp16, x = var_165_cast_fp16)[name = tensor<string, []>("op_166_cast_fp16")];
            tensor<fp16, []> var_167_to_fp16 = const()[name = tensor<string, []>("op_167_to_fp16"), val = tensor<fp16, []>(0x1p-1)];
            tensor<fp16, [?, 1]> sub_score_associated_app_cast_fp16 = mul(x = var_166_cast_fp16, y = var_167_to_fp16)[name = tensor<string, []>("sub_score_associated_app_cast_fp16")];
            tensor<string, []> sub_score_associated_app_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("sub_score_associated_app_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<int32, [2]> cat_popularity_pow_begin_0 = const()[name = tensor<string, []>("cat_popularity_pow_begin_0"), val = tensor<int32, [2]>([0, 20])];
            tensor<int32, [2]> cat_popularity_pow_end_0 = const()[name = tensor<string, []>("cat_popularity_pow_end_0"), val = tensor<int32, [2]>([0, 21])];
            tensor<bool, [2]> cat_popularity_pow_end_mask_0 = const()[name = tensor<string, []>("cat_popularity_pow_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> cat_popularity_pow_cast_fp16 = slice_by_index(begin = cat_popularity_pow_begin_0, end = cat_popularity_pow_end_0, end_mask = cat_popularity_pow_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("cat_popularity_pow_cast_fp16")];
            tensor<int32, [2]> cat_popularity_dow_begin_0 = const()[name = tensor<string, []>("cat_popularity_dow_begin_0"), val = tensor<int32, [2]>([0, 18])];
            tensor<int32, [2]> cat_popularity_dow_end_0 = const()[name = tensor<string, []>("cat_popularity_dow_end_0"), val = tensor<int32, [2]>([0, 19])];
            tensor<bool, [2]> cat_popularity_dow_end_mask_0 = const()[name = tensor<string, []>("cat_popularity_dow_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> cat_popularity_dow_cast_fp16 = slice_by_index(begin = cat_popularity_dow_begin_0, end = cat_popularity_dow_end_0, end_mask = cat_popularity_dow_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("cat_popularity_dow_cast_fp16")];
            tensor<int32, [2]> cat_popularity_begin_0 = const()[name = tensor<string, []>("cat_popularity_begin_0"), val = tensor<int32, [2]>([0, 16])];
            tensor<int32, [2]> cat_popularity_end_0 = const()[name = tensor<string, []>("cat_popularity_end_0"), val = tensor<int32, [2]>([0, 17])];
            tensor<bool, [2]> cat_popularity_end_mask_0 = const()[name = tensor<string, []>("cat_popularity_end_mask_0"), val = tensor<bool, [2]>([true, false])];
            tensor<fp16, [?, 1]> cat_popularity_cast_fp16 = slice_by_index(begin = cat_popularity_begin_0, end = cat_popularity_end_0, end_mask = cat_popularity_end_mask_0, x = input_feature_vector_to_fp16)[name = tensor<string, []>("cat_popularity_cast_fp16")];
            tensor<fp16, []> var_199_to_fp16 = const()[name = tensor<string, []>("op_199_to_fp16"), val = tensor<fp16, []>(0x1p-1)];
            tensor<fp16, [?, 1]> var_200_cast_fp16 = mul(x = cat_popularity_pow_cast_fp16, y = var_199_to_fp16)[name = tensor<string, []>("op_200_cast_fp16")];
            tensor<fp16, []> var_201_to_fp16 = const()[name = tensor<string, []>("op_201_to_fp16"), val = tensor<fp16, []>(0x1.998p-2)];
            tensor<fp16, [?, 1]> var_202_cast_fp16 = mul(x = cat_popularity_dow_cast_fp16, y = var_201_to_fp16)[name = tensor<string, []>("op_202_cast_fp16")];
            tensor<fp16, [?, 1]> var_204_cast_fp16 = add(x = var_200_cast_fp16, y = var_202_cast_fp16)[name = tensor<string, []>("op_204_cast_fp16")];
            tensor<fp16, []> var_205_to_fp16 = const()[name = tensor<string, []>("op_205_to_fp16"), val = tensor<fp16, []>(0x1.998p-4)];
            tensor<fp16, [?, 1]> var_206_cast_fp16 = mul(x = cat_popularity_cast_fp16, y = var_205_to_fp16)[name = tensor<string, []>("op_206_cast_fp16")];
            tensor<fp16, [?, 1]> var_208_cast_fp16 = add(x = var_204_cast_fp16, y = var_206_cast_fp16)[name = tensor<string, []>("op_208_cast_fp16")];
            tensor<fp16, []> var_209_to_fp16 = const()[name = tensor<string, []>("op_209_to_fp16"), val = tensor<fp16, []>(0x1.9p+5)];
            tensor<fp16, [?, 1]> sub_score_category_popularity_cast_fp16 = mul(x = var_208_cast_fp16, y = var_209_to_fp16)[name = tensor<string, []>("sub_score_category_popularity_cast_fp16")];
            tensor<string, []> sub_score_category_popularity_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("sub_score_category_popularity_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<int32, []> var_212 = const()[name = tensor<string, []>("op_212"), val = tensor<int32, []>(1)];
            tensor<bool, []> all_sub_scores_interleave_0 = const()[name = tensor<string, []>("all_sub_scores_interleave_0"), val = tensor<bool, []>(false)];
            tensor<fp16, [?, 7]> all_sub_scores_cast_fp16 = concat(axis = var_212, interleave = all_sub_scores_interleave_0, values = (sub_score_confirms_cast_fp16, sub_score_rejects_cast_fp16, sub_score_open_age_cast_fp16, sub_score_ssid_popularity_cast_fp16, sub_score_document_popularity_cast_fp16, sub_score_associated_app_cast_fp16, sub_score_category_popularity_cast_fp16))[name = tensor<string, []>("all_sub_scores_cast_fp16")];
            tensor<int32, [1]> var_218_axes_0 = const()[name = tensor<string, []>("op_218_axes_0"), val = tensor<int32, [1]>([1])];
            tensor<bool, []> var_218_keep_dims_0 = const()[name = tensor<string, []>("op_218_keep_dims_0"), val = tensor<bool, []>(false)];
            tensor<fp16, [?]> var_218_cast_fp16 = reduce_sum(axes = var_218_axes_0, keep_dims = var_218_keep_dims_0, x = all_sub_scores_cast_fp16)[name = tensor<string, []>("op_218_cast_fp16")];
            tensor<string, []> var_218_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("op_218_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<fp32, [?]> score = cast(dtype = var_218_cast_fp16_to_fp32_dtype_0, x = var_218_cast_fp16)[name = tensor<string, []>("cast_4")];
            tensor<fp32, [?, 1]> SubScoreCategoryPopularity = cast(dtype = sub_score_category_popularity_cast_fp16_to_fp32_dtype_0, x = sub_score_category_popularity_cast_fp16)[name = tensor<string, []>("cast_5")];
            tensor<fp32, [?, 1]> SubScoreAssociatedApp = cast(dtype = sub_score_associated_app_cast_fp16_to_fp32_dtype_0, x = sub_score_associated_app_cast_fp16)[name = tensor<string, []>("cast_6")];
            tensor<fp32, [?, 1]> SubScoreDocumentPopularity = cast(dtype = sub_score_document_popularity_cast_fp16_to_fp32_dtype_0, x = sub_score_document_popularity_cast_fp16)[name = tensor<string, []>("cast_7")];
            tensor<fp32, [?, 1]> SubScoreSSIDPopularity = cast(dtype = sub_score_ssid_popularity_cast_fp16_to_fp32_dtype_0, x = sub_score_ssid_popularity_cast_fp16)[name = tensor<string, []>("cast_8")];
            tensor<fp32, [?, 1]> SubScoreOpenAge = cast(dtype = sub_score_open_age_cast_fp16_to_fp32_dtype_0, x = sub_score_open_age_cast_fp16)[name = tensor<string, []>("cast_9")];
            tensor<fp32, [?, 1]> SubScoreRejects = cast(dtype = sub_score_rejects_cast_fp16_to_fp32_dtype_0, x = sub_score_rejects_cast_fp16)[name = tensor<string, []>("cast_10")];
            tensor<fp32, [?, 1]> SubScoreConfirms = cast(dtype = sub_score_confirms_cast_fp16_to_fp32_dtype_0, x = sub_score_confirms_cast_fp16)[name = tensor<string, []>("cast_12")];
            tensor<string, []> cast_15_dtype_0 = const()[name = tensor<string, []>("cast_15_dtype_0"), val = tensor<string, []>("fp32")];
            tensor<fp32, [?, 1]> AppScorePresumedProbabilistic = cast(dtype = cast_15_dtype_0, x = AppScorePresumedProbabilistic__pre__output__fp32__cast)[name = tensor<string, []>("cast_15")];
        } -> (score, SubScoreConfirms, SubScoreRejects, SubScoreOpenAge, SubScoreDocumentPopularity, SubScoreAssociatedApp, SubScoreCategoryPopularity, SubScoreSSIDPopularity, AppScorePresumedProbabilistic);
}